From 0ed4e08e45bb73927f1496fcbbbb57d5b99d2767 Mon Sep 17 00:00:00 2001
From: Harshada Balasubramanian <harshada.balasubramanian@mpimet.mpg.de>
Date: Mon, 24 Feb 2025 11:53:11 +0100
Subject: [PATCH 01/36] structure
---
src/interpolation/CMakeLists.txt | 3 +
.../mo_lib_interpolation_vector.cpp | 167 +++++++-----------
.../mo_lib_interpolation_vector.hpp | 33 ++--
test/c/test_interpolation_vector.cpp | 5 +-
4 files changed, 84 insertions(+), 124 deletions(-)
diff --git a/src/interpolation/CMakeLists.txt b/src/interpolation/CMakeLists.txt
index 37c3ad0..01a20eb 100644
--- a/src/interpolation/CMakeLists.txt
+++ b/src/interpolation/CMakeLists.txt
@@ -13,10 +13,13 @@ add_library(
iconmath-interpolation
mo_lib_interpolation_scalar.F90
mo_lib_interpolation_vector.F90
+ mo_lib_interpolation_vector.hpp
mo_lib_interpolation_vector.cpp
mo_lib_intp_rbf.F90
mo_lib_intp_rbf-rbf_vec_interpol_vertex_lib.cpp
mo_lib_intp_rbf-rbf_vec_interpol_vertex_lib_bindings.cpp
+ interpolation_bindings.h
+ interpolation_bindings.cpp
)
add_library(${PROJECT_NAME}::interpolation ALIAS iconmath-interpolation)
diff --git a/src/interpolation/mo_lib_interpolation_vector.cpp b/src/interpolation/mo_lib_interpolation_vector.cpp
index 00a914a..40ef003 100644
--- a/src/interpolation/mo_lib_interpolation_vector.cpp
+++ b/src/interpolation/mo_lib_interpolation_vector.cpp
@@ -1,24 +1,7 @@
-// ICON
-//
-// ---------------------------------------------------------------
-// Copyright (C) 2004-2024, DWD, MPI-M, DKRZ, KIT, ETH, MeteoSwiss
-// Contact information: icon-model.org
-//
-// See AUTHORS.TXT for a list of authors
-// See LICENSES/ for license information
-// SPDX-License-Identifier: BSD-3-Clause
-// ---------------------------------------------------------------
-
-#include "mo_lib_loopindices.hpp"
#include "mo_lib_interpolation_vector.hpp"
-// The templated C++ function using Kokkos.
-// Raw pointer arguments are wrapped into unmanaged Kokkos::Views.
-// Note: The dimensions below must match the Fortran arrays.
-// - p_vn_in and p_vt_in: dimensions [nproma, nlev, nblks_e]
-// - cell_edge_idx and cell_edge_blk: dimensions [nproma, nblks_c, 3]
-// - e_bln_c_u and e_bln_c_v: dimensions [nproma, 6, nblks_c]
-// - p_u_out and p_v_out: dimensions [nproma, nlev, nblks_c]
+
+
template <typename T>
void edges2cells_vector_lib(
const T* p_vn_in, const T* p_vt_in,
@@ -33,109 +16,95 @@ void edges2cells_vector_lib(
// Dimensions for the arrays.
int nlev, int nblks_e, int nblks_c)
{
- // Wrap raw pointers in unmanaged Kokkos Views.
- typedef Kokkos::View<const T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstT3D;
- typedef Kokkos::View<T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedT3D;
- typedef Kokkos::View<const int***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstInt3D;
- UnmanagedConstT3D p_vn_in_view(p_vn_in, nproma, nlev, nblks_e);
- UnmanagedConstT3D p_vt_in_view(p_vt_in, nproma, nlev, nblks_e);
+ // Wrap raw pointers in unmanaged Kokkos Views.
+ typedef Kokkos::View<const T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstT3D;
+ typedef Kokkos::View<T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedT3D;
+ typedef Kokkos::View<const int***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstInt3D;
+
+
+ UnmanagedConstT3D p_vn_in_view(p_vn_in, nproma, nlev, nblks_e);
+ UnmanagedConstT3D p_vt_in_view(p_vt_in, nproma, nlev, nblks_e);
- UnmanagedConstInt3D cell_edge_idx_view(cell_edge_idx, nproma, nblks_c, 3);
- UnmanagedConstInt3D cell_edge_blk_view(cell_edge_blk, nproma, nblks_c, 3);
+ UnmanagedConstInt3D cell_edge_idx_view(cell_edge_idx, nproma, nblks_c, 3);
+ UnmanagedConstInt3D cell_edge_blk_view(cell_edge_blk, nproma, nblks_c, 3);
- UnmanagedConstT3D e_bln_c_u_view(e_bln_c_u, nproma, 6, nblks_c);
- UnmanagedConstT3D e_bln_c_v_view(e_bln_c_v, nproma, 6, nblks_c);
+ UnmanagedConstT3D e_bln_c_u_view(e_bln_c_u, nproma, 6, nblks_c);
+ UnmanagedConstT3D e_bln_c_v_view(e_bln_c_v, nproma, 6, nblks_c);
- UnmanagedT3D p_u_out_view(p_u_out, nproma, nlev, nblks_c);
- UnmanagedT3D p_v_out_view(p_v_out, nproma, nlev, nblks_c);
+ UnmanagedT3D p_u_out_view(p_u_out, nproma, nlev, nblks_c);
+ UnmanagedT3D p_v_out_view(p_v_out, nproma, nlev, nblks_c);
- // Loop over cell blocks as in the original Fortran code.
- for (int jb = i_startblk; jb <= i_endblk; ++jb) {
- // Call get_indices_c_lib to get inner loop indices for block jb.
- int i_startidx, i_endidx;
- get_indices_c_lib(i_startidx_in, i_endidx_in, nproma,
- jb, i_startblk, i_endblk,
- i_startidx, i_endidx);
+ // Loop over cell blocks as in the original Fortran code.
+ for (int jb = i_startblk; jb <= i_endblk; ++jb) {
+ // Call get_indices_c_lib to get inner loop indices for block jb.
+ int i_startidx, i_endidx;
+ get_indices_c_lib(i_startidx_in, i_endidx_in, nproma,
+ jb, i_startblk, i_endblk,
+ i_startidx, i_endidx);
- Kokkos::MDRangePolicy<Kokkos::Rank<2>> innerPolicy(
- {slev, i_startidx}, {elev + 1, i_endidx + 1});
- Kokkos::parallel_for("edges2cells_inner", innerPolicy,
- KOKKOS_LAMBDA(const int jk, const int jc) {
- // Compute the bilinear interpolation for cell (jc, jk, jb).
- p_u_out_view(jc, jk, jb) =
- e_bln_c_u_view(jc, 0, jb) *
- p_vn_in_view(cell_edge_idx_view(jc, jb, 0) - 1, jk, cell_edge_blk_view(jc, jb, 0) - 1) +
- e_bln_c_u_view(jc, 1, jb) *
- p_vt_in_view(cell_edge_idx_view(jc, jb, 0) - 1, jk, cell_edge_blk_view(jc, jb, 0) - 1) +
- e_bln_c_u_view(jc, 2, jb) *
- p_vn_in_view(cell_edge_idx_view(jc, jb, 1) - 1, jk, cell_edge_blk_view(jc, jb, 1) - 1) +
- e_bln_c_u_view(jc, 3, jb) *
- p_vt_in_view(cell_edge_idx_view(jc, jb, 1) - 1, jk, cell_edge_blk_view(jc, jb, 1) - 1) +
- e_bln_c_u_view(jc, 4, jb) *
- p_vn_in_view(cell_edge_idx_view(jc, jb, 2) - 1, jk, cell_edge_blk_view(jc, jb, 2) - 1) +
- e_bln_c_u_view(jc, 5, jb) *
- p_vt_in_view(cell_edge_idx_view(jc, jb, 2) - 1, jk, cell_edge_blk_view(jc, jb, 2) - 1);
+ Kokkos::MDRangePolicy<Kokkos::Rank<2>> innerPolicy(
+ {slev, i_startidx}, {elev + 1, i_endidx + 1});
+ Kokkos::parallel_for("edges2cells_inner", innerPolicy,
+ KOKKOS_LAMBDA(const int jk, const int jc) {
+ // Compute the bilinear interpolation for cell (jc, jk, jb).
+ p_u_out_view(jc, jk, jb) =
+ e_bln_c_u_view(jc, 0, jb) *
+ p_vn_in_view(cell_edge_idx_view(jc, jb, 0) - 1, jk, cell_edge_blk_view(jc, jb, 0) - 1) +
+ e_bln_c_u_view(jc, 1, jb) *
+ p_vt_in_view(cell_edge_idx_view(jc, jb, 0) - 1, jk, cell_edge_blk_view(jc, jb, 0) - 1) +
+ e_bln_c_u_view(jc, 2, jb) *
+ p_vn_in_view(cell_edge_idx_view(jc, jb, 1) - 1, jk, cell_edge_blk_view(jc, jb, 1) - 1) +
+ e_bln_c_u_view(jc, 3, jb) *
+ p_vt_in_view(cell_edge_idx_view(jc, jb, 1) - 1, jk, cell_edge_blk_view(jc, jb, 1) - 1) +
+ e_bln_c_u_view(jc, 4, jb) *
+ p_vn_in_view(cell_edge_idx_view(jc, jb, 2) - 1, jk, cell_edge_blk_view(jc, jb, 2) - 1) +
+ e_bln_c_u_view(jc, 5, jb) *
+ p_vt_in_view(cell_edge_idx_view(jc, jb, 2) - 1, jk, cell_edge_blk_view(jc, jb, 2) - 1);
- p_v_out_view(jc, jk, jb) =
- e_bln_c_v_view(jc, 0, jb) *
- p_vn_in_view(cell_edge_idx_view(jc, jb, 0) - 1, jk, cell_edge_blk_view(jc, jb, 0) - 1) +
- e_bln_c_v_view(jc, 1, jb) *
- p_vt_in_view(cell_edge_idx_view(jc, jb, 0) - 1, jk, cell_edge_blk_view(jc, jb, 0) - 1) +
- e_bln_c_v_view(jc, 2, jb) *
- p_vn_in_view(cell_edge_idx_view(jc, jb, 1) - 1, jk, cell_edge_blk_view(jc, jb, 1) - 1) +
- e_bln_c_v_view(jc, 3, jb) *
- p_vt_in_view(cell_edge_idx_view(jc, jb, 1) - 1, jk, cell_edge_blk_view(jc, jb, 1) - 1) +
- e_bln_c_v_view(jc, 4, jb) *
- p_vn_in_view(cell_edge_idx_view(jc, jb, 2) - 1, jk, cell_edge_blk_view(jc, jb, 2) - 1) +
- e_bln_c_v_view(jc, 5, jb) *
- p_vt_in_view(cell_edge_idx_view(jc, jb, 2) - 1, jk, cell_edge_blk_view(jc, jb, 2) - 1);
- });
- // Optionally fence after each block if required.
- Kokkos::fence();
- }
+ p_v_out_view(jc, jk, jb) =
+ e_bln_c_v_view(jc, 0, jb) *
+ p_vn_in_view(cell_edge_idx_view(jc, jb, 0) - 1, jk, cell_edge_blk_view(jc, jb, 0) - 1) +
+ e_bln_c_v_view(jc, 1, jb) *
+ p_vt_in_view(cell_edge_idx_view(jc, jb, 0) - 1, jk, cell_edge_blk_view(jc, jb, 0) - 1) +
+ e_bln_c_v_view(jc, 2, jb) *
+ p_vn_in_view(cell_edge_idx_view(jc, jb, 1) - 1, jk, cell_edge_blk_view(jc, jb, 1) - 1) +
+ e_bln_c_v_view(jc, 3, jb) *
+ p_vt_in_view(cell_edge_idx_view(jc, jb, 1) - 1, jk, cell_edge_blk_view(jc, jb, 1) - 1) +
+ e_bln_c_v_view(jc, 4, jb) *
+ p_vn_in_view(cell_edge_idx_view(jc, jb, 2) - 1, jk, cell_edge_blk_view(jc, jb, 2) - 1) +
+ e_bln_c_v_view(jc, 5, jb) *
+ p_vt_in_view(cell_edge_idx_view(jc, jb, 2) - 1, jk, cell_edge_blk_view(jc, jb, 2) - 1);
+ });
+ // Optionally fence after each block if required.
+ Kokkos::fence();
+ }
}
-extern "C" void edges2cells_vector_lib_dp(
+template
+void edges2cells_vector_lib<double>(
const double* p_vn_in, const double* p_vt_in,
const int* cell_edge_idx, const int* cell_edge_blk,
const double* e_bln_c_u, const double* e_bln_c_v,
double* p_u_out, double* p_v_out,
+ // Additional integer parameters.
int i_startblk, int i_endblk,
int i_startidx_in, int i_endidx_in,
int slev, int elev,
int nproma,
- int nlev, int nblks_e, int nblks_c)
-{
- edges2cells_vector_lib<double>(p_vn_in, p_vt_in,
- cell_edge_idx, cell_edge_blk,
- e_bln_c_u, e_bln_c_v,
- p_u_out, p_v_out,
- i_startblk, i_endblk,
- i_startidx_in, i_endidx_in,
- slev, elev,
- nproma,
- nlev, nblks_e, nblks_c);
-}
+ // Dimensions for the arrays.
+ int nlev, int nblks_e, int nblks_c);
-extern "C" void edges2cells_vector_lib_sp(
+template
+void edges2cells_vector_lib<float>(
const float* p_vn_in, const float* p_vt_in,
const int* cell_edge_idx, const int* cell_edge_blk,
const float* e_bln_c_u, const float* e_bln_c_v,
float* p_u_out, float* p_v_out,
+ // Additional integer parameters.
int i_startblk, int i_endblk,
int i_startidx_in, int i_endidx_in,
int slev, int elev,
int nproma,
- int nlev, int nblks_e, int nblks_c)
-{
- edges2cells_vector_lib<float>(p_vn_in, p_vt_in,
- cell_edge_idx, cell_edge_blk,
- e_bln_c_u, e_bln_c_v,
- p_u_out, p_v_out,
- i_startblk, i_endblk,
- i_startidx_in, i_endidx_in,
- slev, elev,
- nproma,
- nlev, nblks_e, nblks_c);
-}
+ // Dimensions for the arrays.
+ int nlev, int nblks_e, int nblks_c);
\ No newline at end of file
diff --git a/src/interpolation/mo_lib_interpolation_vector.hpp b/src/interpolation/mo_lib_interpolation_vector.hpp
index 0d19b24..1ee6c5b 100644
--- a/src/interpolation/mo_lib_interpolation_vector.hpp
+++ b/src/interpolation/mo_lib_interpolation_vector.hpp
@@ -8,10 +8,19 @@
// See LICENSES/ for license information
// SPDX-License-Identifier: BSD-3-Clause
// ---------------------------------------------------------------
+#pragma once
+#include "mo_lib_loopindices.hpp"
#include <Kokkos_Core.hpp>
#include <vector>
+// The templated C++ function using Kokkos.
+// Raw pointer arguments are wrapped into unmanaged Kokkos::Views.
+// Note: The dimensions below must match the Fortran arrays.
+// - p_vn_in and p_vt_in: dimensions [nproma, nlev, nblks_e]
+// - cell_edge_idx and cell_edge_blk: dimensions [nproma, nblks_c, 3]
+// - e_bln_c_u and e_bln_c_v: dimensions [nproma, 6, nblks_c]
+// - p_u_out and p_v_out: dimensions [nproma, nlev, nblks_c]
template <typename T>
void edges2cells_vector_lib(
const T* p_vn_in, const T* p_vt_in,
@@ -24,26 +33,4 @@ void edges2cells_vector_lib(
int slev, int elev,
int nproma,
// Dimensions for the arrays.
- int nlev, int nblks_e, int nblks_c);
-
-extern "C" void edges2cells_vector_lib_dp(
- const double* p_vn_in, const double* p_vt_in,
- const int* cell_edge_idx, const int* cell_edge_blk,
- const double* e_bln_c_u, const double* e_bln_c_v,
- double* p_u_out, double* p_v_out,
- int i_startblk, int i_endblk,
- int i_startidx_in, int i_endidx_in,
- int slev, int elev,
- int nproma,
- int nlev, int nblks_e, int nblks_c);
-
-extern "C" void edges2cells_vector_lib_sp(
- const float* p_vn_in, const float* p_vt_in,
- const int* cell_edge_idx, const int* cell_edge_blk,
- const float* e_bln_c_u, const float* e_bln_c_v,
- float* p_u_out, float* p_v_out,
- int i_startblk, int i_endblk,
- int i_startidx_in, int i_endidx_in,
- int slev, int elev,
- int nproma,
- int nlev, int nblks_e, int nblks_c);
+ int nlev, int nblks_e, int nblks_c);
\ No newline at end of file
diff --git a/test/c/test_interpolation_vector.cpp b/test/c/test_interpolation_vector.cpp
index 0eb5a8d..974efcd 100644
--- a/test/c/test_interpolation_vector.cpp
+++ b/test/c/test_interpolation_vector.cpp
@@ -12,6 +12,7 @@
#include <gtest/gtest.h>
#include <Kokkos_Core.hpp>
#include <vector>
+
#include "mo_lib_interpolation_vector.hpp"
// Dimensions for the test (small, trivial test).
@@ -62,7 +63,7 @@ TEST(Edges2CellsTest, DPTest) {
std::vector<double> p_v_ref(num_elements<double>(nproma, nlev, nblks_c), 6.0);
// Call the dp (double precision) version.
- edges2cells_vector_lib_dp(
+ edges2cells_vector_lib<double>(
p_vn_in.data(), p_vt_in.data(),
cell_edge_idx.data(), cell_edge_blk.data(),
e_bln_c_u.data(), e_bln_c_v.data(),
@@ -103,7 +104,7 @@ TEST(Edges2CellsTest, SPTest) {
std::vector<float> p_v_ref(num_elements<float>(nproma, nlev, nblks_c), 6.0f);
// Call the sp (float precision) version.
- edges2cells_vector_lib_sp(
+ edges2cells_vector_lib<float>(
p_vn_in.data(), p_vt_in.data(),
cell_edge_idx.data(), cell_edge_blk.data(),
e_bln_c_u.data(), e_bln_c_v.data(),
--
GitLab
From 4b9ef5dbc75f803f4e67adf7d40cddc529c59837 Mon Sep 17 00:00:00 2001
From: Harshada Balasubramanian <harshada.balasubramanian@mpimet.mpg.de>
Date: Mon, 24 Feb 2025 12:45:08 +0100
Subject: [PATCH 02/36] fix cmake
---
src/interpolation/CMakeLists.txt | 1 -
src/interpolation/interpolation_bindings.cpp | 36 ++++++++++++++++++++
src/interpolation/interpolation_bindings.h | 23 +++++++++++++
3 files changed, 59 insertions(+), 1 deletion(-)
create mode 100644 src/interpolation/interpolation_bindings.cpp
create mode 100644 src/interpolation/interpolation_bindings.h
diff --git a/src/interpolation/CMakeLists.txt b/src/interpolation/CMakeLists.txt
index 01a20eb..eade738 100644
--- a/src/interpolation/CMakeLists.txt
+++ b/src/interpolation/CMakeLists.txt
@@ -13,7 +13,6 @@ add_library(
iconmath-interpolation
mo_lib_interpolation_scalar.F90
mo_lib_interpolation_vector.F90
- mo_lib_interpolation_vector.hpp
mo_lib_interpolation_vector.cpp
mo_lib_intp_rbf.F90
mo_lib_intp_rbf-rbf_vec_interpol_vertex_lib.cpp
diff --git a/src/interpolation/interpolation_bindings.cpp b/src/interpolation/interpolation_bindings.cpp
new file mode 100644
index 0000000..daa1c44
--- /dev/null
+++ b/src/interpolation/interpolation_bindings.cpp
@@ -0,0 +1,36 @@
+// ICON
+//
+// ---------------------------------------------------------------
+// Copyright (C) 2004-2024, DWD, MPI-M, DKRZ, KIT, ETH, MeteoSwiss
+// Contact information: icon-model.org
+//
+// See AUTHORS.TXT for a list of authors
+// See LICENSES/ for license information
+// SPDX-License-Identifier: BSD-3-Clause
+// ---------------------------------------------------------------
+
+#include "interpolation_bindings.h"
+#include "mo_lib_interpolation_vector.hpp"
+
+void edges2cells_vector_lib_c(
+ const wp* p_vn_in, const wp* p_vt_in,
+ const int* cell_edge_idx, const int* cell_edge_blk,
+ const wp* e_bln_c_u, const wp* e_bln_c_v,
+ wp* p_u_out, wp* p_v_out,
+ int i_startblk, int i_endblk,
+ int i_startidx_in, int i_endidx_in,
+ int slev, int elev,
+ int nproma,
+ int nlev, int nblks_e, int nblks_c)
+{
+ edges2cells_vector_lib<wp>(p_vn_in, p_vt_in,
+ cell_edge_idx, cell_edge_blk,
+ e_bln_c_u, e_bln_c_v,
+ p_u_out, p_v_out,
+ i_startblk, i_endblk,
+ i_startidx_in, i_endidx_in,
+ slev, elev,
+ nproma,
+ nlev, nblks_e, nblks_c);
+}
+
diff --git a/src/interpolation/interpolation_bindings.h b/src/interpolation/interpolation_bindings.h
new file mode 100644
index 0000000..a712f53
--- /dev/null
+++ b/src/interpolation/interpolation_bindings.h
@@ -0,0 +1,23 @@
+#pragma once
+
+#ifdef __SINGLE_PRECISION
+ using wp = single;
+#else
+ using wp = double;
+#endif
+
+
+extern "C"{
+
+ void edges2cells_vector_lib_c(
+ const wp* p_vn_in, const wp* p_vt_in,
+ const int* cell_edge_idx, const int* cell_edge_blk,
+ const wp* e_bln_c_u, const wp* e_bln_c_v,
+ wp* p_u_out, wp* p_v_out,
+ int i_startblk, int i_endblk,
+ int i_startidx_in, int i_endidx_in,
+ int slev, int elev,
+ int nproma,
+ int nlev, int nblks_e, int nblks_c);
+
+}
\ No newline at end of file
--
GitLab
From d9ba3f7897888f8495f87439a78d766a056ce01f Mon Sep 17 00:00:00 2001
From: Harshada Balasubramanian <harshada.balasubramanian@mpimet.mpg.de>
Date: Mon, 24 Feb 2025 12:46:55 +0100
Subject: [PATCH 03/36] add license header
---
src/interpolation/interpolation_bindings.h | 10 ++++++++++
src/interpolation/mo_lib_interpolation_vector.cpp | 13 +++++++++++--
2 files changed, 21 insertions(+), 2 deletions(-)
diff --git a/src/interpolation/interpolation_bindings.h b/src/interpolation/interpolation_bindings.h
index a712f53..5134365 100644
--- a/src/interpolation/interpolation_bindings.h
+++ b/src/interpolation/interpolation_bindings.h
@@ -1,3 +1,13 @@
+// ICON
+//
+// ---------------------------------------------------------------
+// Copyright (C) 2004-2024, DWD, MPI-M, DKRZ, KIT, ETH, MeteoSwiss
+// Contact information: icon-model.org
+//
+// See AUTHORS.TXT for a list of authors
+// See LICENSES/ for license information
+// SPDX-License-Identifier: BSD-3-Clause
+// ---------------------------------------------------------------
#pragma once
#ifdef __SINGLE_PRECISION
diff --git a/src/interpolation/mo_lib_interpolation_vector.cpp b/src/interpolation/mo_lib_interpolation_vector.cpp
index 40ef003..74c2be7 100644
--- a/src/interpolation/mo_lib_interpolation_vector.cpp
+++ b/src/interpolation/mo_lib_interpolation_vector.cpp
@@ -1,6 +1,15 @@
-#include "mo_lib_interpolation_vector.hpp"
-
+// ICON
+//
+// ---------------------------------------------------------------
+// Copyright (C) 2004-2024, DWD, MPI-M, DKRZ, KIT, ETH, MeteoSwiss
+// Contact information: icon-model.org
+//
+// See AUTHORS.TXT for a list of authors
+// See LICENSES/ for license information
+// SPDX-License-Identifier: BSD-3-Clause
+// ---------------------------------------------------------------
+#include "mo_lib_interpolation_vector.hpp"
template <typename T>
void edges2cells_vector_lib(
--
GitLab
From 14b309458936d393dc93eb55f10cf9d47c0ea772 Mon Sep 17 00:00:00 2001
From: Harshada Balasubramanian <harshada.balasubramanian@mpimet.mpg.de>
Date: Tue, 25 Feb 2025 10:53:17 +0100
Subject: [PATCH 04/36] verts2edges_scalar_lib
---
src/interpolation/interpolation_bindings.cpp | 79 +++++++++++++++++--
src/interpolation/interpolation_bindings.h | 55 +++++++++----
.../mo_lib_interpolation_scalar.cpp | 62 +++++++++++++++
.../mo_lib_interpolation_scalar.hpp | 10 +++
4 files changed, 182 insertions(+), 24 deletions(-)
create mode 100644 src/interpolation/mo_lib_interpolation_scalar.cpp
create mode 100644 src/interpolation/mo_lib_interpolation_scalar.hpp
diff --git a/src/interpolation/interpolation_bindings.cpp b/src/interpolation/interpolation_bindings.cpp
index daa1c44..c2f0c5e 100644
--- a/src/interpolation/interpolation_bindings.cpp
+++ b/src/interpolation/interpolation_bindings.cpp
@@ -11,19 +11,21 @@
#include "interpolation_bindings.h"
#include "mo_lib_interpolation_vector.hpp"
-
-void edges2cells_vector_lib_c(
- const wp* p_vn_in, const wp* p_vt_in,
+#include "mo_lib_interpolation_scalar.hpp"
+
+//mo_lib_interpolation_vector.F90
+void edges2cells_vector_lib_dp(
+ const double* p_vn_in, const double* p_vt_in,
const int* cell_edge_idx, const int* cell_edge_blk,
- const wp* e_bln_c_u, const wp* e_bln_c_v,
- wp* p_u_out, wp* p_v_out,
+ const double* e_bln_c_u, const double* e_bln_c_v,
+ double* p_u_out, double* p_v_out,
int i_startblk, int i_endblk,
int i_startidx_in, int i_endidx_in,
int slev, int elev,
int nproma,
- int nlev, int nblks_e, int nblks_c)
-{
- edges2cells_vector_lib<wp>(p_vn_in, p_vt_in,
+ int nlev, int nblks_e, int nblks_c){
+
+ edges2cells_vector_lib<double>(p_vn_in, p_vt_in,
cell_edge_idx, cell_edge_blk,
e_bln_c_u, e_bln_c_v,
p_u_out, p_v_out,
@@ -34,3 +36,64 @@ void edges2cells_vector_lib_c(
nlev, nblks_e, nblks_c);
}
+void edges2cells_vector_lib_sp(
+ const float* p_vn_in, const float* p_vt_in,
+ const int* cell_edge_idx, const int* cell_edge_blk,
+ const float* e_bln_c_u, const float* e_bln_c_v,
+ float* p_u_out, float* p_v_out,
+ int i_startblk, int i_endblk,
+ int i_startidx_in, int i_endidx_in,
+ int slev, int elev,
+ int nproma,
+ int nlev, int nblks_e, int nblks_c){
+
+ edges2cells_vector_lib<float>(p_vn_in, p_vt_in,
+ cell_edge_idx, cell_edge_blk,
+ e_bln_c_u, e_bln_c_v,
+ p_u_out, p_v_out,
+ i_startblk, i_endblk,
+ i_startidx_in, i_endidx_in,
+ slev, elev,
+ nproma,
+ nlev, nblks_e, nblks_c);
+}
+
+//mo_lib_interpolation_scalar.F90
+void verts2edges_scalar_lib_dp(
+ const double* p_vertex_in, const int* shape_p_vertex_in,
+ const int* edge_vertex_idx, const int* edge_vertex_blk, const int* shape_edge_vertex,
+ const double* coeff_int, const int* shape_coeff_int,
+ double* p_edge_out, const int* shape_p_edge_out,
+ const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
+ const int nproma, const bool lacc){
+
+ verts2edges_scalar_lib<double>(p_vertex_in, shape_p_vertex_in,
+ edge_vertex_idx, edge_vertex_blk, shape_edge_vertex,
+ coeff_int, shape_coeff_int,
+ p_edge_out, shape_p_edge_out,
+ i_startblk, i_endblk,
+ i_startidx_in, i_endidx_in, slev, elev,
+ nproma, lacc);
+
+}
+
+void verts2edges_scalar_lib_sp(
+ const float* p_vertex_in, const int* shape_p_vertex_in,
+ const int* edge_vertex_idx, const int* edge_vertex_blk, const int* shape_edge_vertex,
+ const float* coeff_int, const int* shape_coeff_int,
+ float* p_edge_out, const int* shape_p_edge_out,
+ const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
+ const int nproma, const bool lacc){
+
+ verts2edges_scalar_lib<float>(p_vertex_in, shape_p_vertex_in,
+ edge_vertex_idx, edge_vertex_blk, shape_edge_vertex,
+ coeff_int, shape_coeff_int,
+ p_edge_out, shape_p_edge_out,
+ i_startblk, i_endblk,
+ i_startidx_in, i_endidx_in, slev, elev,
+ nproma, lacc);
+
+}
+
diff --git a/src/interpolation/interpolation_bindings.h b/src/interpolation/interpolation_bindings.h
index 5134365..dceba97 100644
--- a/src/interpolation/interpolation_bindings.h
+++ b/src/interpolation/interpolation_bindings.h
@@ -10,24 +10,47 @@
// ---------------------------------------------------------------
#pragma once
-#ifdef __SINGLE_PRECISION
- using wp = single;
-#else
- using wp = double;
-#endif
+extern "C"{
+ //mo_lib_interpolation_vector.F90
+ void edges2cells_vector_lib_dp(
+ const double* p_vn_in, const double* p_vt_in,
+ const int* cell_edge_idx, const int* cell_edge_blk,
+ const double* e_bln_c_u, const double* e_bln_c_v,
+ double* p_u_out, double* p_v_out,
+ int i_startblk, int i_endblk,
+ int i_startidx_in, int i_endidx_in,
+ int slev, int elev,
+ int nproma,
+ int nlev, int nblks_e, int nblks_c);
-extern "C"{
+ void edges2cells_vector_lib_sp(
+ const float* p_vn_in, const float* p_vt_in,
+ const int* cell_edge_idx, const int* cell_edge_blk,
+ const float* e_bln_c_u, const float* e_bln_c_v,
+ float* p_u_out, float* p_v_out,
+ int i_startblk, int i_endblk,
+ int i_startidx_in, int i_endidx_in,
+ int slev, int elev,
+ int nproma,
+ int nlev, int nblks_e, int nblks_c);
- void edges2cells_vector_lib_c(
- const wp* p_vn_in, const wp* p_vt_in,
- const int* cell_edge_idx, const int* cell_edge_blk,
- const wp* e_bln_c_u, const wp* e_bln_c_v,
- wp* p_u_out, wp* p_v_out,
- int i_startblk, int i_endblk,
- int i_startidx_in, int i_endidx_in,
- int slev, int elev,
- int nproma,
- int nlev, int nblks_e, int nblks_c);
+ //mo_lib_interpolation_scalar.F90
+ void verts2edges_scalar_lib_dp(
+ const double* p_vertex_in, const int* shape_p_vertex_in,
+ const int* edge_vertex_idx, const int* edge_vertex_blk, const int* shape_edge_vertex,
+ const double* coeff_int, const int* shape_coeff_int,
+ double* p_edge_out, const int* shape_p_edge_out,
+ const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
+ const int nproma, const bool lacc);
+ void verts2edges_scalar_lib_sp(
+ const float* p_vertex_in, const int* shape_p_vertex_in,
+ const int* edge_vertex_idx, const int* edge_vertex_blk, const int* shape_edge_vertex,
+ const float* coeff_int, const int* shape_coeff_int,
+ float* p_edge_out, const int* shape_p_edge_out,
+ const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
+ const int nproma, const bool lacc);
}
\ No newline at end of file
diff --git a/src/interpolation/mo_lib_interpolation_scalar.cpp b/src/interpolation/mo_lib_interpolation_scalar.cpp
new file mode 100644
index 0000000..ac9e4f0
--- /dev/null
+++ b/src/interpolation/mo_lib_interpolation_scalar.cpp
@@ -0,0 +1,62 @@
+#include <Kokkos_Core.hpp>
+#include <vector>
+#include "mo_lib_loopindices.hpp"
+#include "mo_lib_interpolation_scalar.hpp"
+
+template <typename T>
+void verts2edges_scalar_lib(const T* p_vertex_in, const int* shape_p_vertex_in,
+ const int* edge_vertex_idx, const int* edge_vertex_blk, const int* shape_edge_vertex,
+ const T* coeff_int, const int* shape_coeff_int,
+ T* p_edge_out, const int* shape_p_edge_out,
+ const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
+ const int nproma, const bool lacc){
+
+ // Wrap raw pointers in unmanaged Kokkos Views.
+ typedef Kokkos::View<const T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstT3D;
+ typedef Kokkos::View<T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedT3D;
+ typedef Kokkos::View<const int***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstInt3D;
+
+ UnmanagedConstT3D p_vertex_in_view (p_vertex_in, shape_p_vertex_in[0], shape_p_vertex_in[1], shape_p_vertex_in[2]);
+ UnmanagedConstInt3D iidx_view (edge_vertex_idx, shape_edge_vertex[0], shape_edge_vertex[1], shape_edge_vertex[2]);
+ UnmanagedConstInt3D iblk_view (edge_vertex_blk, shape_edge_vertex[0], shape_edge_vertex[1], shape_edge_vertex[2]);
+ UnmanagedConstT3D coeff_int_view (coeff_int, shape_coeff_int[0], shape_coeff_int[1], shape_coeff_int[2]);
+ UnmanagedT3D p_edge_out_view (p_edge_out, shape_p_edge_out[0], shape_p_edge_out[1], shape_p_edge_out[2]);
+
+ for (int jb = i_startblk; jb < i_endblk; ++jb) {
+
+ int i_startidx, i_endidx;
+ get_indices_c_lib(i_startidx_in, i_endidx_in, nproma,
+ jb, i_startblk, i_endblk,
+ i_startidx, i_endidx);
+
+ Kokkos::parallel_for("verts2edges_scalar", Kokkos::MDRangePolicy<Kokkos::Rank<2>>
+ ({slev, i_startidx}, {elev, i_endidx}),
+ KOKKOS_LAMBDA(const int jk, const int je) {
+
+ p_edge_out_view(je, jk, jb) = coeff_int_view(je, 0, jb) *
+ p_vertex_in_view(iidx_view(je, jb, 0), jk, iblk_view(je, jb, 0)) +
+ coeff_int_view(je, 1, jb)*p_vertex_in_view(iidx_view(je, jb, 1), jk, iblk_view(je, jb, 1));
+
+ });
+ Kokkos::fence();
+ }
+}
+
+template
+void verts2edges_scalar_lib<double>(const double* p_vertex_in, const int* shape_p_vertex_in,
+ const int* edge_vertex_idx, const int* edge_vertex_blk, const int* shape_edge_vertex,
+ const double* coeff_int, const int* shape_coeff_int,
+ double* p_edge_out, const int* shape_p_edge_out,
+ const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
+ const int nproma, const bool lacc);
+
+template
+void verts2edges_scalar_lib<float>(const float* p_vertex_in, const int* shape_p_vertex_in,
+ const int* edge_vertex_idx, const int* edge_vertex_blk, const int* shape_edge_vertex,
+ const float* coeff_int, const int* shape_coeff_int,
+ float* p_edge_out, const int* shape_p_edge_out,
+ const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
+ const int nproma, const bool lacc);
\ No newline at end of file
diff --git a/src/interpolation/mo_lib_interpolation_scalar.hpp b/src/interpolation/mo_lib_interpolation_scalar.hpp
new file mode 100644
index 0000000..5178308
--- /dev/null
+++ b/src/interpolation/mo_lib_interpolation_scalar.hpp
@@ -0,0 +1,10 @@
+#pragma once
+
+template <typename T>
+void verts2edges_scalar_lib(const T* p_vertex_in, const int* shape_p_vertex_in,
+ const int* edge_vertex_idx, const int* edge_vertex_blk, const int* shape_edge_vertex,
+ const T* coeff_int, const int* shape_coeff_int,
+ T* p_edge_out, const int* shape_p_edge_out,
+ const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
+ const int nproma, const bool lacc);
\ No newline at end of file
--
GitLab
From a5760248c767ce77c8e39b427f801e4e7cb4f1c2 Mon Sep 17 00:00:00 2001
From: Harshada Balasubramanian <harshada.balasubramanian@mpimet.mpg.de>
Date: Tue, 25 Feb 2025 12:31:16 +0100
Subject: [PATCH 05/36] cells2edges_scalar_lib
---
src/interpolation/interpolation_bindings.cpp | 42 +++++++
src/interpolation/interpolation_bindings.h | 24 ++++
.../mo_lib_interpolation_scalar.cpp | 105 +++++++++++++++++-
.../mo_lib_interpolation_scalar.hpp | 10 +-
4 files changed, 179 insertions(+), 2 deletions(-)
diff --git a/src/interpolation/interpolation_bindings.cpp b/src/interpolation/interpolation_bindings.cpp
index c2f0c5e..61e9c89 100644
--- a/src/interpolation/interpolation_bindings.cpp
+++ b/src/interpolation/interpolation_bindings.cpp
@@ -97,3 +97,45 @@ void verts2edges_scalar_lib_sp(
}
+void cells2edges_scalar_lib_dp(const double* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
+ const double* coeff_int, double* p_edge_out, const int* i_startblk_in,
+ const int* i_endblk_in, const int* i_startidx_in, const int* i_endidx_in,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblk_c, const int nblks_e, const int patch_id,
+ const bool l_limited_area, const bool lfill_latbc, const bool lacc){
+
+ cells2edges_scalar_lib<double, double>(p_cell_in,edge_cell_idx,edge_cell_blk,coeff_int,p_edge_out,i_startblk_in,
+ i_endblk_in,i_startidx_in,i_endidx_in,slev,elev,nproma,nlev,
+ nblk_c,nblks_e,patch_id,l_limited_area,lfill_latbc,lacc);
+
+}
+
+
+void cells2edges_scalar_lib_sp(const float* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
+ const float* coeff_int, float* p_edge_out, const int* i_startblk_in,
+ const int* i_endblk_in, const int* i_startidx_in, const int* i_endidx_in,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblk_c, const int nblks_e, const int patch_id,
+ const bool l_limited_area, const bool lfill_latbc, const bool lacc){
+
+ cells2edges_scalar_lib<float, float>(p_cell_in,edge_cell_idx,edge_cell_blk,coeff_int,p_edge_out,i_startblk_in,
+ i_endblk_in,i_startidx_in,i_endidx_in,slev,elev,nproma,nlev,
+ nblk_c,nblks_e,patch_id,l_limited_area,lfill_latbc,lacc);
+
+}
+
+void cells2edges_scalar_lib_sp2dp(const float* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
+ const double* coeff_int, double* p_edge_out, const int* i_startblk_in,
+ const int* i_endblk_in, const int* i_startidx_in, const int* i_endidx_in,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblk_c, const int nblks_e, const int patch_id,
+ const bool l_limited_area, const bool lfill_latbc, const bool lacc){
+
+ cells2edges_scalar_lib<double, float>(p_cell_in,edge_cell_idx,edge_cell_blk,coeff_int,p_edge_out,i_startblk_in,
+ i_endblk_in,i_startidx_in,i_endidx_in,slev,elev,nproma,nlev,
+ nblk_c,nblks_e,patch_id,l_limited_area,lfill_latbc,lacc);
+
+}
+
+
+
diff --git a/src/interpolation/interpolation_bindings.h b/src/interpolation/interpolation_bindings.h
index dceba97..d2411af 100644
--- a/src/interpolation/interpolation_bindings.h
+++ b/src/interpolation/interpolation_bindings.h
@@ -53,4 +53,28 @@ extern "C"{
const int i_startblk, const int i_endblk,
const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
const int nproma, const bool lacc);
+
+
+ void cells2edges_scalar_lib_dp(const double* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
+ const double* coeff_int, double* p_edge_out, const int* i_startblk_in,
+ const int* i_endblk_in, const int* i_startidx_in, const int* i_endidx_in,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblk_c, const int nblks_e, const int patch_id,
+ const bool l_limited_area, const bool lfill_latbc, const bool lacc);
+
+
+ void cells2edges_scalar_lib_sp(const float* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
+ const float* coeff_int, float* p_edge_out, const int* i_startblk_in,
+ const int* i_endblk_in, const int* i_startidx_in, const int* i_endidx_in,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblk_c, const int nblks_e, const int patch_id,
+ const bool l_limited_area, const bool lfill_latbc, const bool lacc);
+
+ void cells2edges_scalar_lib_sp2dp(const float* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
+ const double* coeff_int, double* p_edge_out, const int* i_startblk_in,
+ const int* i_endblk_in, const int* i_startidx_in, const int* i_endidx_in,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblk_c, const int nblks_e, const int patch_id,
+ const bool l_limited_area, const bool lfill_latbc, const bool lacc);
+
}
\ No newline at end of file
diff --git a/src/interpolation/mo_lib_interpolation_scalar.cpp b/src/interpolation/mo_lib_interpolation_scalar.cpp
index ac9e4f0..03822b6 100644
--- a/src/interpolation/mo_lib_interpolation_scalar.cpp
+++ b/src/interpolation/mo_lib_interpolation_scalar.cpp
@@ -1,5 +1,6 @@
#include <Kokkos_Core.hpp>
#include <vector>
+#include <iostream>
#include "mo_lib_loopindices.hpp"
#include "mo_lib_interpolation_scalar.hpp"
@@ -43,6 +44,82 @@ void verts2edges_scalar_lib(const T* p_vertex_in, const int* shape_p_vertex_in,
}
}
+template <typename T, typename S>
+void cells2edges_scalar_lib(const S* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
+ const T* coeff_int, T* p_edge_out, const int* i_startblk_in,
+ const int* i_endblk_in, const int* i_startidx_in, const int* i_endidx_in,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblk_c, const int nblks_e, const int patch_id,
+ const bool l_limited_area, const bool lfill_latbc, const bool lacc){
+
+ // Wrap raw pointers in unmanaged Kokkos Views.
+ typedef Kokkos::View<const T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstT3D;
+ typedef Kokkos::View<const S***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstS3D;
+ typedef Kokkos::View<T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedT3D;
+ typedef Kokkos::View<const int***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstInt3D;
+
+ UnmanagedConstS3D p_cell_in_view (p_cell_in, nproma, nlev, nblk_c);
+ UnmanagedConstInt3D iidx_view (edge_cell_idx, nproma, nblks_e, 2);
+ UnmanagedConstInt3D iblk_view (edge_cell_blk, nproma, nblks_e, 2);
+ UnmanagedConstT3D coeff_int_view (coeff_int, nproma, 2, nblks_e);
+ UnmanagedT3D p_edge_out_view (p_edge_out, nproma, nlev, nblks_e);
+
+ //Fill outermost nest boundary
+ int i_startblk, i_endblk;
+ if ((l_limited_area || patch_id > 0) && (lfill_latbc)){
+ i_startblk = i_startblk_in[0];
+ i_endblk = i_endblk_in[0];
+
+ for (int jb = i_startblk; jb < i_endblk; ++jb) {
+
+ int i_startidx, i_endidx;
+ get_indices_e_lib(i_startidx_in[0], i_endidx_in[0], nproma, jb,
+ i_startblk, i_endblk, i_startidx, i_endidx);
+
+ Kokkos::parallel_for("cells2edges_scalar", Kokkos::MDRangePolicy<Kokkos::Rank<2>>
+ ({slev, i_startidx}, {elev, i_endidx}),
+ KOKKOS_LAMBDA(const int jk, const int je) {
+
+ if (iidx_view(je, jb, 0) >= 0 && iblk_view(je, jb, 0) >= 0){
+ p_edge_out_view(je, jk, jb) = p_cell_in_view(iidx_view(je, jb, 0), jk, iblk_view(je, jb, 0));
+ }
+ else if (iidx_view(je, jb, 1) >= 0 && iblk_view(je, jb, 1) >= 0){
+ p_edge_out_view(je, jk, jb) = p_cell_in_view(iidx_view(je, jb, 1), jk, iblk_view(je, jb, 1));
+ }
+ else{
+ std::cerr << "mo_interpolation:cells2edges_scalar_lib: error in lateral boundary filling" << std::endl;
+ std::exit(EXIT_FAILURE);
+ }
+ });
+ Kokkos::fence();
+ }
+ }
+ else{
+ //Process the remaining grid points for which a real interpolation is possible
+ i_startblk = i_startblk_in[1];
+ i_endblk = i_endblk_in[1];
+
+ for (int jb = i_startblk; jb < i_endblk; ++jb) {
+
+ int i_startidx, i_endidx;
+ get_indices_e_lib(i_startidx_in[1], i_endidx_in[1], nproma, jb,
+ i_startblk, i_endblk, i_startidx, i_endidx);
+
+ Kokkos::parallel_for("cells2edges_scalar", Kokkos::MDRangePolicy<Kokkos::Rank<2>>
+ ({slev, i_startidx}, {elev, i_endidx}),
+ KOKKOS_LAMBDA(const int jk, const int je) {
+
+ p_edge_out_view(je, jk, jb) = coeff_int_view(je, 0, jb) *
+ p_cell_in_view(iidx_view(je, jb, 0), jk, iblk_view(je, jb, 0)) +
+ coeff_int_view(je, 1, jb) * p_cell_in_view(iidx_view(je, jb, 1), jk, iblk_view(je, jb, 1));
+
+ });
+ Kokkos::fence();
+ }
+ }
+
+}
+
template
void verts2edges_scalar_lib<double>(const double* p_vertex_in, const int* shape_p_vertex_in,
const int* edge_vertex_idx, const int* edge_vertex_blk, const int* shape_edge_vertex,
@@ -59,4 +136,30 @@ void verts2edges_scalar_lib<float>(const float* p_vertex_in, const int* shape_p_
float* p_edge_out, const int* shape_p_edge_out,
const int i_startblk, const int i_endblk,
const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
- const int nproma, const bool lacc);
\ No newline at end of file
+ const int nproma, const bool lacc);
+
+template
+void cells2edges_scalar_lib<double, double>(const double* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
+ const double* coeff_int, double* p_edge_out, const int* i_startblk_in,
+ const int* i_endblk_in, const int* i_startidx_in, const int* i_endidx_in,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblk_c, const int nblks_e, const int patch_id,
+ const bool l_limited_area, const bool lfill_latbc, const bool lacc);
+
+template
+void cells2edges_scalar_lib<float, float>(const float* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
+ const float* coeff_int, float* p_edge_out, const int* i_startblk_in,
+ const int* i_endblk_in, const int* i_startidx_in, const int* i_endidx_in,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblk_c, const int nblks_e, const int patch_id,
+ const bool l_limited_area, const bool lfill_latbc, const bool lacc);
+
+template
+void cells2edges_scalar_lib<double, float>(const float* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
+ const double* coeff_int, double* p_edge_out, const int* i_startblk_in,
+ const int* i_endblk_in, const int* i_startidx_in, const int* i_endidx_in,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblk_c, const int nblks_e, const int patch_id,
+ const bool l_limited_area, const bool lfill_latbc, const bool lacc);
+
+
diff --git a/src/interpolation/mo_lib_interpolation_scalar.hpp b/src/interpolation/mo_lib_interpolation_scalar.hpp
index 5178308..b6a2509 100644
--- a/src/interpolation/mo_lib_interpolation_scalar.hpp
+++ b/src/interpolation/mo_lib_interpolation_scalar.hpp
@@ -7,4 +7,12 @@ void verts2edges_scalar_lib(const T* p_vertex_in, const int* shape_p_vertex_in,
T* p_edge_out, const int* shape_p_edge_out,
const int i_startblk, const int i_endblk,
const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
- const int nproma, const bool lacc);
\ No newline at end of file
+ const int nproma, const bool lacc);
+
+template <typename T, typename S>
+void cells2edges_scalar_lib(const S* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
+ const T* coeff_int, T* p_edge_out, const int* i_startblk_in,
+ const int* i_endblk_in, const int* i_startidx_in, const int* i_endidx_in,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblk_c, const int nblks_e, const int patch_id,
+ const bool l_limited_area, const bool lfill_latbc, const bool lacc);
--
GitLab
From cc01fe7c6c42c7102f23907b97a6f940719991ac Mon Sep 17 00:00:00 2001
From: Harshada Balasubramanian <harshada.balasubramanian@mpimet.mpg.de>
Date: Tue, 25 Feb 2025 15:26:59 +0100
Subject: [PATCH 06/36] edges2verts_scalar_lib
---
src/interpolation/interpolation_bindings.cpp | 22 ++++++++
src/interpolation/interpolation_bindings.h | 14 +++++
.../mo_lib_interpolation_scalar.cpp | 56 +++++++++++++++++++
.../mo_lib_interpolation_scalar.hpp | 7 +++
4 files changed, 99 insertions(+)
diff --git a/src/interpolation/interpolation_bindings.cpp b/src/interpolation/interpolation_bindings.cpp
index 61e9c89..509be83 100644
--- a/src/interpolation/interpolation_bindings.cpp
+++ b/src/interpolation/interpolation_bindings.cpp
@@ -137,5 +137,27 @@ void cells2edges_scalar_lib_sp2dp(const float* p_cell_in, const int* edge_cell_i
}
+void edges2verts_scalar_lib_dp (const double* p_edge_in, const int* vert_edge_idx, const int* vert_edge_blk,
+ const double* v_int, double* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in, const int slev,
+ const int elev, const int nproma, const int nlev, const int nblks_e,
+ const int nblks_v, const int cell_type, const bool lacc){
+
+ edges2verts_scalar_lib<double>(p_edge_in, vert_edge_idx, vert_edge_blk, v_int,p_vert_out,i_startblk,
+ i_endblk,i_startidx_in,i_endidx_in, slev, elev, nproma, nlev,nblks_e, nblks_v, cell_type,lacc);
+}
+
+void edges2verts_scalar_lib_sp(const float* p_edge_in, const int* vert_edge_idx, const int* vert_edge_blk,
+ const float* v_int, float* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in, const int slev,
+ const int elev, const int nproma, const int nlev, const int nblks_e,
+ const int nblks_v, const int cell_type, const bool lacc){
+
+ edges2verts_scalar_lib<float>(p_edge_in, vert_edge_idx, vert_edge_blk, v_int,p_vert_out,i_startblk,
+ i_endblk,i_startidx_in,i_endidx_in, slev, elev, nproma, nlev,nblks_e, nblks_v, cell_type,lacc);
+
+}
+
+
diff --git a/src/interpolation/interpolation_bindings.h b/src/interpolation/interpolation_bindings.h
index d2411af..c630ae8 100644
--- a/src/interpolation/interpolation_bindings.h
+++ b/src/interpolation/interpolation_bindings.h
@@ -77,4 +77,18 @@ extern "C"{
const int nblk_c, const int nblks_e, const int patch_id,
const bool l_limited_area, const bool lfill_latbc, const bool lacc);
+ void edges2verts_scalar_lib_dp (const double* p_edge_in, const int* vert_edge_idx, const int* vert_edge_blk,
+ const double* v_int, double* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in, const int slev,
+ const int elev, const int nproma, const int nlev, const int nblks_e,
+ const int nblks_v, const int cell_type, const bool lacc);
+
+
+ void edges2verts_scalar_lib_sp (const float* p_edge_in, const int* vert_edge_idx, const int* vert_edge_blk,
+ const float* v_int, float* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in, const int slev,
+ const int elev, const int nproma, const int nlev, const int nblks_e,
+ const int nblks_v, const int cell_type, const bool lacc);
+
+
}
\ No newline at end of file
diff --git a/src/interpolation/mo_lib_interpolation_scalar.cpp b/src/interpolation/mo_lib_interpolation_scalar.cpp
index 03822b6..34d3336 100644
--- a/src/interpolation/mo_lib_interpolation_scalar.cpp
+++ b/src/interpolation/mo_lib_interpolation_scalar.cpp
@@ -120,6 +120,47 @@ void cells2edges_scalar_lib(const S* p_cell_in, const int* edge_cell_idx, const
}
+template <typename T>
+void edges2verts_scalar_lib (const T* p_edge_in, const int* vert_edge_idx, const int* vert_edge_blk,
+ const T* v_int, T* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in, const int slev,
+ const int elev, const int nproma, const int nlev, const int nblks_e,
+ const int nblks_v, const int cell_type, const bool lacc){
+
+ // Wrap raw pointers in unmanaged Kokkos Views.
+ typedef Kokkos::View<const T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstT3D;
+ typedef Kokkos::View<T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedT3D;
+ typedef Kokkos::View<const int***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstInt3D;
+
+ UnmanagedConstT3D p_edge_in_view (p_edge_in, nproma, nlev, nblks_e);
+ UnmanagedConstInt3D iidx_view (vert_edge_idx, nproma, nblks_v, 5);
+ UnmanagedConstInt3D iblk_view (vert_edge_blk, nproma,nblks_v, 5);
+ UnmanagedConstT3D v_int_view (v_int, nproma,cell_type,nblks_v);
+ UnmanagedT3D p_vert_out_view (p_vert_out, nproma,nlev,nblks_v);
+
+ for (int jb = i_startblk; jb < i_endblk; ++jb) {
+
+ int i_startidx, i_endidx;
+ get_indices_v_lib(i_startidx_in, i_endidx_in, nproma,
+ jb, i_startblk, i_endblk,
+ i_startidx, i_endidx);
+
+ Kokkos::parallel_for("edges2verts_scalar", Kokkos::MDRangePolicy<Kokkos::Rank<2>>
+ ({slev, i_startidx}, {elev, i_endidx}),
+ KOKKOS_LAMBDA(const int jk, const int jv) {
+
+ p_vert_out_view(jv, jk, jb) = v_int_view(jv, 0, jb)*p_edge_in_view(iidx_view(jv, jb, 0), jk, iblk_view(jv, jb, 0))
+ + v_int_view(jv, 1, jb)*p_edge_in_view(iidx_view(jv, jb, 1), jk, iblk_view(jv, jb, 1))
+ + v_int_view(jv, 2, jb)*p_edge_in_view(iidx_view(jv, jb, 2), jk, iblk_view(jv, jb, 2))
+ + v_int_view(jv, 3, jb)*p_edge_in_view(iidx_view(jv, jb, 3), jk, iblk_view(jv, jb, 3))
+ + v_int_view(jv, 4, jb)*p_edge_in_view(iidx_view(jv, jb, 4), jk, iblk_view(jv, jb, 4))
+ + v_int_view(jv, 5, jb)*p_edge_in_view(iidx_view(jv, jb, 5), jk, iblk_view(jv, jb, 5));
+ });
+ Kokkos::fence();
+ }
+
+}
+
template
void verts2edges_scalar_lib<double>(const double* p_vertex_in, const int* shape_p_vertex_in,
const int* edge_vertex_idx, const int* edge_vertex_blk, const int* shape_edge_vertex,
@@ -163,3 +204,18 @@ void cells2edges_scalar_lib<double, float>(const float* p_cell_in, const int* ed
const bool l_limited_area, const bool lfill_latbc, const bool lacc);
+template
+void edges2verts_scalar_lib<double> (const double* p_edge_in, const int* vert_edge_idx, const int* vert_edge_blk,
+ const double* v_int, double* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in, const int slev,
+ const int elev, const int nproma, const int nlev, const int nblks_e,
+ const int nblks_v, const int cell_type, const bool lacc);
+
+template
+void edges2verts_scalar_lib <float> (const float* p_edge_in, const int* vert_edge_idx, const int* vert_edge_blk,
+ const float* v_int, float* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in, const int slev,
+ const int elev, const int nproma, const int nlev, const int nblks_e,
+ const int nblks_v, const int cell_type, const bool lacc);
+
+
diff --git a/src/interpolation/mo_lib_interpolation_scalar.hpp b/src/interpolation/mo_lib_interpolation_scalar.hpp
index b6a2509..6145089 100644
--- a/src/interpolation/mo_lib_interpolation_scalar.hpp
+++ b/src/interpolation/mo_lib_interpolation_scalar.hpp
@@ -16,3 +16,10 @@ void cells2edges_scalar_lib(const S* p_cell_in, const int* edge_cell_idx, const
const int slev, const int elev, const int nproma, const int nlev,
const int nblk_c, const int nblks_e, const int patch_id,
const bool l_limited_area, const bool lfill_latbc, const bool lacc);
+
+template <typename T>
+void edges2verts_scalar_lib (const T* p_edge_in, const int* vert_edge_idx, const int* vert_edge_blk,
+ const T* v_int, T* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in, const int slev,
+ const int elev, const int nproma, const int nlev, const int nblks_e,
+ const int nblks_v, const int cell_type, const bool lacc);
--
GitLab
From 350143de7d8cce4b14eaf7b259f688766af04005 Mon Sep 17 00:00:00 2001
From: Harshada Balasubramanian <harshada.balasubramanian@mpimet.mpg.de>
Date: Tue, 25 Feb 2025 15:37:34 +0100
Subject: [PATCH 07/36] remove shape
---
src/interpolation/interpolation_bindings.cpp | 40 +++++++++----------
src/interpolation/interpolation_bindings.h | 33 +++++++--------
.../mo_lib_interpolation_scalar.cpp | 40 +++++++++----------
.../mo_lib_interpolation_scalar.hpp | 15 +++----
4 files changed, 61 insertions(+), 67 deletions(-)
diff --git a/src/interpolation/interpolation_bindings.cpp b/src/interpolation/interpolation_bindings.cpp
index 509be83..2589edf 100644
--- a/src/interpolation/interpolation_bindings.cpp
+++ b/src/interpolation/interpolation_bindings.cpp
@@ -59,41 +59,37 @@ void edges2cells_vector_lib_sp(
}
//mo_lib_interpolation_scalar.F90
-void verts2edges_scalar_lib_dp(
- const double* p_vertex_in, const int* shape_p_vertex_in,
- const int* edge_vertex_idx, const int* edge_vertex_blk, const int* shape_edge_vertex,
- const double* coeff_int, const int* shape_coeff_int,
- double* p_edge_out, const int* shape_p_edge_out,
+void verts2edges_scalar_lib_dp(const double* p_vertex_in,
+ const int* edge_vertex_idx, const int* edge_vertex_blk,
+ const double* coeff_int,
+ double* p_edge_out,
const int i_startblk, const int i_endblk,
const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
- const int nproma, const bool lacc){
+ const int nproma, const int nlev, const int nblks_v, const int nblks_e, const bool lacc){
- verts2edges_scalar_lib<double>(p_vertex_in, shape_p_vertex_in,
- edge_vertex_idx, edge_vertex_blk, shape_edge_vertex,
- coeff_int, shape_coeff_int,
- p_edge_out, shape_p_edge_out,
+ verts2edges_scalar_lib<double>(p_vertex_in,
+ edge_vertex_idx, edge_vertex_blk, coeff_int,
+ p_edge_out,
i_startblk, i_endblk,
i_startidx_in, i_endidx_in, slev, elev,
- nproma, lacc);
+ nproma, nlev, nblks_v, nblks_e, lacc);
}
-void verts2edges_scalar_lib_sp(
- const float* p_vertex_in, const int* shape_p_vertex_in,
- const int* edge_vertex_idx, const int* edge_vertex_blk, const int* shape_edge_vertex,
- const float* coeff_int, const int* shape_coeff_int,
- float* p_edge_out, const int* shape_p_edge_out,
+void verts2edges_scalar_lib_sp(const float* p_vertex_in,
+ const int* edge_vertex_idx, const int* edge_vertex_blk,
+ const float* coeff_int,
+ float* p_edge_out,
const int i_startblk, const int i_endblk,
const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
- const int nproma, const bool lacc){
+ const int nproma, const int nlev, const int nblks_v, const int nblks_e, const bool lacc){
- verts2edges_scalar_lib<float>(p_vertex_in, shape_p_vertex_in,
- edge_vertex_idx, edge_vertex_blk, shape_edge_vertex,
- coeff_int, shape_coeff_int,
- p_edge_out, shape_p_edge_out,
+ verts2edges_scalar_lib<float>(p_vertex_in,
+ edge_vertex_idx, edge_vertex_blk, coeff_int,
+ p_edge_out,
i_startblk, i_endblk,
i_startidx_in, i_endidx_in, slev, elev,
- nproma, lacc);
+ nproma, nlev, nblks_v, nblks_e, lacc);
}
diff --git a/src/interpolation/interpolation_bindings.h b/src/interpolation/interpolation_bindings.h
index c630ae8..8567ef9 100644
--- a/src/interpolation/interpolation_bindings.h
+++ b/src/interpolation/interpolation_bindings.h
@@ -36,25 +36,22 @@ extern "C"{
int nlev, int nblks_e, int nblks_c);
//mo_lib_interpolation_scalar.F90
- void verts2edges_scalar_lib_dp(
- const double* p_vertex_in, const int* shape_p_vertex_in,
- const int* edge_vertex_idx, const int* edge_vertex_blk, const int* shape_edge_vertex,
- const double* coeff_int, const int* shape_coeff_int,
- double* p_edge_out, const int* shape_p_edge_out,
- const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
- const int nproma, const bool lacc);
+ void verts2edges_scalar_lib_dp(const double* p_vertex_in,
+ const int* edge_vertex_idx, const int* edge_vertex_blk,
+ const double* coeff_int,
+ double* p_edge_out,
+ const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
+ const int nproma, const int nlev, const int nblks_v, const int nblks_e, const bool lacc);
- void verts2edges_scalar_lib_sp(
- const float* p_vertex_in, const int* shape_p_vertex_in,
- const int* edge_vertex_idx, const int* edge_vertex_blk, const int* shape_edge_vertex,
- const float* coeff_int, const int* shape_coeff_int,
- float* p_edge_out, const int* shape_p_edge_out,
- const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
- const int nproma, const bool lacc);
-
-
+ void verts2edges_scalar_lib_sp(const float* p_vertex_in,
+ const int* edge_vertex_idx, const int* edge_vertex_blk,
+ const float* coeff_int,
+ float* p_edge_out,
+ const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
+ const int nproma, const int nlev, const int nblks_v, const int nblks_e, const bool lacc);
+
void cells2edges_scalar_lib_dp(const double* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
const double* coeff_int, double* p_edge_out, const int* i_startblk_in,
const int* i_endblk_in, const int* i_startidx_in, const int* i_endidx_in,
diff --git a/src/interpolation/mo_lib_interpolation_scalar.cpp b/src/interpolation/mo_lib_interpolation_scalar.cpp
index 34d3336..0994ac1 100644
--- a/src/interpolation/mo_lib_interpolation_scalar.cpp
+++ b/src/interpolation/mo_lib_interpolation_scalar.cpp
@@ -5,24 +5,24 @@
#include "mo_lib_interpolation_scalar.hpp"
template <typename T>
-void verts2edges_scalar_lib(const T* p_vertex_in, const int* shape_p_vertex_in,
- const int* edge_vertex_idx, const int* edge_vertex_blk, const int* shape_edge_vertex,
- const T* coeff_int, const int* shape_coeff_int,
- T* p_edge_out, const int* shape_p_edge_out,
+void verts2edges_scalar_lib(const T* p_vertex_in,
+ const int* edge_vertex_idx, const int* edge_vertex_blk,
+ const T* coeff_int,
+ T* p_edge_out,
const int i_startblk, const int i_endblk,
const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
- const int nproma, const bool lacc){
+ const int nproma, const int nlev, const int nblks_v, const int nblks_e, const bool lacc){
// Wrap raw pointers in unmanaged Kokkos Views.
typedef Kokkos::View<const T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstT3D;
typedef Kokkos::View<T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedT3D;
typedef Kokkos::View<const int***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstInt3D;
- UnmanagedConstT3D p_vertex_in_view (p_vertex_in, shape_p_vertex_in[0], shape_p_vertex_in[1], shape_p_vertex_in[2]);
- UnmanagedConstInt3D iidx_view (edge_vertex_idx, shape_edge_vertex[0], shape_edge_vertex[1], shape_edge_vertex[2]);
- UnmanagedConstInt3D iblk_view (edge_vertex_blk, shape_edge_vertex[0], shape_edge_vertex[1], shape_edge_vertex[2]);
- UnmanagedConstT3D coeff_int_view (coeff_int, shape_coeff_int[0], shape_coeff_int[1], shape_coeff_int[2]);
- UnmanagedT3D p_edge_out_view (p_edge_out, shape_p_edge_out[0], shape_p_edge_out[1], shape_p_edge_out[2]);
+ UnmanagedConstT3D p_vertex_in_view (p_vertex_in, nproma,nlev,nblks_v);
+ UnmanagedConstInt3D iidx_view (edge_vertex_idx,nproma,nblks_e, 4);
+ UnmanagedConstInt3D iblk_view (edge_vertex_blk, nproma,nblks_e, 4);
+ UnmanagedConstT3D coeff_int_view (coeff_int, nproma,2,nblks_e);
+ UnmanagedT3D p_edge_out_view (p_edge_out, nproma,nlev,nblks_e);
for (int jb = i_startblk; jb < i_endblk; ++jb) {
@@ -162,22 +162,22 @@ void edges2verts_scalar_lib (const T* p_edge_in, const int* vert_edge_idx, const
}
template
-void verts2edges_scalar_lib<double>(const double* p_vertex_in, const int* shape_p_vertex_in,
- const int* edge_vertex_idx, const int* edge_vertex_blk, const int* shape_edge_vertex,
- const double* coeff_int, const int* shape_coeff_int,
- double* p_edge_out, const int* shape_p_edge_out,
+void verts2edges_scalar_lib<double>(const double* p_vertex_in,
+ const int* edge_vertex_idx, const int* edge_vertex_blk,
+ const double* coeff_int,
+ double* p_edge_out,
const int i_startblk, const int i_endblk,
const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
- const int nproma, const bool lacc);
+ const int nproma, const int nlev, const int nblks_v, const int nblks_e, const bool lacc);
template
-void verts2edges_scalar_lib<float>(const float* p_vertex_in, const int* shape_p_vertex_in,
- const int* edge_vertex_idx, const int* edge_vertex_blk, const int* shape_edge_vertex,
- const float* coeff_int, const int* shape_coeff_int,
- float* p_edge_out, const int* shape_p_edge_out,
+void verts2edges_scalar_lib<float>(const float* p_vertex_in,
+ const int* edge_vertex_idx, const int* edge_vertex_blk,
+ const float* coeff_int,
+ float* p_edge_out,
const int i_startblk, const int i_endblk,
const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
- const int nproma, const bool lacc);
+ const int nproma, const int nlev, const int nblks_v, const int nblks_e, const bool lacc);
template
void cells2edges_scalar_lib<double, double>(const double* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
diff --git a/src/interpolation/mo_lib_interpolation_scalar.hpp b/src/interpolation/mo_lib_interpolation_scalar.hpp
index 6145089..1c578b0 100644
--- a/src/interpolation/mo_lib_interpolation_scalar.hpp
+++ b/src/interpolation/mo_lib_interpolation_scalar.hpp
@@ -1,13 +1,14 @@
#pragma once
template <typename T>
-void verts2edges_scalar_lib(const T* p_vertex_in, const int* shape_p_vertex_in,
- const int* edge_vertex_idx, const int* edge_vertex_blk, const int* shape_edge_vertex,
- const T* coeff_int, const int* shape_coeff_int,
- T* p_edge_out, const int* shape_p_edge_out,
- const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
- const int nproma, const bool lacc);
+void verts2edges_scalar_lib(const T* p_vertex_in,
+ const int* edge_vertex_idx, const int* edge_vertex_blk,
+ const T* coeff_int,
+ T* p_edge_out,
+ const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
+ const int nproma, const int nlev, const int nblks_v, const int nblks_e, const bool lacc);
+;
template <typename T, typename S>
void cells2edges_scalar_lib(const S* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
--
GitLab
From a51b9a2b201d5932ee3dc5376324b30735cb0e51 Mon Sep 17 00:00:00 2001
From: Dylan Kierans <kierans@dkrz.de>
Date: Tue, 25 Feb 2025 15:51:51 +0100
Subject: [PATCH 08/36] Partial mo_lib_interpolation_scalar from Dylan
---
src/interpolation/CMakeLists.txt | 2 +
.../mo_lib_interpolation_scalar_Dylan.cpp | 408 ++++++++++++++++++
.../mo_lib_interpolation_scalar_Dylan.hpp | 54 +++
...ib_interpolation_scalar_bindings_Dylan.cpp | 219 ++++++++++
..._lib_interpolation_scalar_bindings_Dylan.h | 109 +++++
5 files changed, 792 insertions(+)
create mode 100644 src/interpolation/mo_lib_interpolation_scalar_Dylan.cpp
create mode 100644 src/interpolation/mo_lib_interpolation_scalar_Dylan.hpp
create mode 100644 src/interpolation/mo_lib_interpolation_scalar_bindings_Dylan.cpp
create mode 100644 src/interpolation/mo_lib_interpolation_scalar_bindings_Dylan.h
diff --git a/src/interpolation/CMakeLists.txt b/src/interpolation/CMakeLists.txt
index eade738..3e8ef1a 100644
--- a/src/interpolation/CMakeLists.txt
+++ b/src/interpolation/CMakeLists.txt
@@ -12,6 +12,8 @@
add_library(
iconmath-interpolation
mo_lib_interpolation_scalar.F90
+ mo_lib_interpolation_scalar_Dylan.cpp
+ mo_lib_interpolation_scalar_bindings_Dylan.cpp
mo_lib_interpolation_vector.F90
mo_lib_interpolation_vector.cpp
mo_lib_intp_rbf.F90
diff --git a/src/interpolation/mo_lib_interpolation_scalar_Dylan.cpp b/src/interpolation/mo_lib_interpolation_scalar_Dylan.cpp
new file mode 100644
index 0000000..8b133c5
--- /dev/null
+++ b/src/interpolation/mo_lib_interpolation_scalar_Dylan.cpp
@@ -0,0 +1,408 @@
+// ICON
+//
+// ---------------------------------------------------------------
+// Copyright (C) 2004-2024, DWD, MPI-M, DKRZ, KIT, ETH, MeteoSwiss
+// Contact information: icon-model.org
+//
+// See AUTHORS.TXT for a list of authors
+// See LICENSES/ for license information
+// SPDX-License-Identifier: BSD-3-Clause
+// ---------------------------------------------------------------
+
+#include <Kokkos_Core.hpp>
+#include "mo_lib_loopindices.hpp"
+#include "mo_lib_interpolation_scalar_Dylan.hpp"
+
+//------------------------------------------------------------------------
+//
+//>
+/// Computes interpolation from edges to cells
+///
+/// Computes interpolation of scalar fields from velocity points to
+/// cell centers via given interpolation weights
+///
+template <typename T>
+void edges2cells_scalar_lib(
+ const T* p_edge_in, const int* edge_idx, const int* edge_blk,
+ const T* coeff_int, T* p_cell_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_e, const int nblks_c,
+ const bool lacc)
+{
+ // Wrap raw pointers in unmanaged Kokkos Views.
+ typedef Kokkos::View<const T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstT3D;
+ typedef Kokkos::View<T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedT3D;
+ typedef Kokkos::View<const int***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstInt3D;
+
+ // edge based scalar input field, dim: (nproma,nlev,nblks_e)
+ UnmanagedConstT3D p_edge_in_view(p_edge_in, nproma, nlev, nblks_e);
+
+ // line indices of edges of triangles, dim: (nproma,nblks_c, 3)
+ UnmanagedConstInt3D iidx_view(edge_idx, nproma, nblks_c, 3); // edge_idx_view
+
+ // block indices of edges of triangles, dim: (nproma,nblks_c, 3)
+ UnmanagedConstInt3D iblk_view(edge_blk, nproma, nblks_c, 3); // edge_blk_view
+
+ // coefficients for (area weighted) interpolation, dim: (nproma,3-cell_type,nblks_c)
+ UnmanagedConstT3D coeff_int_view(coeff_int, nproma, 3, nblks_c);
+
+ // cell based scalar output field, dim: (nproma,nlev,nblks_c)
+ UnmanagedT3D p_cell_out_view(p_cell_out, nproma, nlev, nblks_c);
+
+ int i_startidx, i_endidx;
+
+ for (int jb = i_startblk; jb < i_endblk; ++jb){
+ get_indices_c_lib(i_startidx_in, i_endidx_in, nproma, jb, i_startblk, i_endblk,
+ i_startidx, i_endidx);
+
+ Kokkos::MDRangePolicy<Kokkos::Rank<2>> innerPolicy(
+ {slev, i_startidx}, {elev + 1, i_endidx + 1});
+
+ Kokkos::parallel_for("edges2cells_scalar_lib_inner", innerPolicy,
+ KOKKOS_LAMBDA(const int jk, const int jc) {
+ p_cell_out_view(jc, jk, jb) =
+ coeff_int_view(jc, 1, jb)*p_edge_in_view(iidx_view(jc, jb, 1), jk, iblk_view(jc, jb, 1)) +
+ coeff_int_view(jc, 2, jb)*p_edge_in_view(iidx_view(jc, jb, 2), jk, iblk_view(jc, jb, 2)) +
+ coeff_int_view(jc, 3, jb)*p_edge_in_view(iidx_view(jc, jb, 3), jk, iblk_view(jc, jb, 3));
+ });
+ }
+
+}
+
+template
+void edges2cells_scalar_lib<double>(
+ const double* p_edge_in, const int* edge_idx, const int* edge_blk,
+ const double* coeff_int, double* p_cell_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_e, const int nblks_c,
+ const bool lacc);
+
+template
+void edges2cells_scalar_lib<float>(
+ const float* p_edge_in, const int* edge_idx, const int* edge_blk,
+ const float* coeff_int, float* p_cell_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_e, const int nblks_c,
+ const bool lacc);
+
+//------------------------------------------------------------------------
+//>
+//// Computes average of scalar fields from centers of cells to vertices.
+////
+template <typename T, typename S>
+void cells2verts_scalar_lib(
+ const T* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const S* coeff_int, S* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async)
+{
+ // Wrap raw pointers in unmanaged Kokkos Views.
+ typedef Kokkos::View<const T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstT3D;
+ typedef Kokkos::View<const S***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstS3D;
+ typedef Kokkos::View<S***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedS3D;
+ typedef Kokkos::View<const int***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstInt3D;
+
+ // cell based scalar input field, dim: (nproma,nlev,nblks_c)
+ UnmanagedConstT3D p_cell_in_view(p_cell_in, nproma, nlev, nblks_c);
+
+ // line indices of cells around each vertex, dim: (nproma,nblks_v, 6)
+ UnmanagedConstInt3D iidx_view(vert_cell_idx, nproma, nblks_v, 6); // vert_cell_idx_view
+
+ // block indices of cells around each vertex, dim: (nproma,nblks_v, 6)
+ UnmanagedConstInt3D iblk_view(vert_cell_blk, nproma, nblks_v, 6); // vert_cell_blk_view
+
+ // coefficients for interpolation, dim: (nproma,9-cell_type,nblks_v)
+ UnmanagedConstS3D coeff_int_view(coeff_int, nproma, 9, nblks_v);
+
+ // vertex based scalar output field, dim: (nproma,nlev,nblks_c)
+ UnmanagedS3D p_vert_out_view(p_vert_out, nproma, nlev, nblks_c);
+
+ int i_startidx, i_endidx;
+
+ for (int jb = i_startblk; jb < i_endblk; ++jb){
+
+ get_indices_v_lib(i_startidx_in, i_endidx_in, nproma, jb, i_startblk, i_endblk,
+ i_startidx, i_endidx);
+
+ Kokkos::MDRangePolicy<Kokkos::Rank<2>> innerPolicy(
+ {slev, i_startidx}, {elev + 1, i_endidx + 1});
+
+ Kokkos::parallel_for("cells2verts_scalar_lib", innerPolicy,
+ KOKKOS_LAMBDA(const int jk, const int jv) {
+
+ p_vert_out_view(jv, jk, jb) =
+ coeff_int_view(jv, 0, jb)*p_cell_in_view(iidx_view(jv, jb, 0), jk, iblk_view(jv, jb, 0)) +
+ coeff_int_view(jv, 1, jb)*p_cell_in_view(iidx_view(jv, jb, 1), jk, iblk_view(jv, jb, 1)) +
+ coeff_int_view(jv, 2, jb)*p_cell_in_view(iidx_view(jv, jb, 2), jk, iblk_view(jv, jb, 2)) +
+ coeff_int_view(jv, 3, jb)*p_cell_in_view(iidx_view(jv, jb, 3), jk, iblk_view(jv, jb, 3)) +
+ coeff_int_view(jv, 4, jb)*p_cell_in_view(iidx_view(jv, jb, 4), jk, iblk_view(jv, jb, 4)) +
+ coeff_int_view(jv, 5, jb)*p_cell_in_view(iidx_view(jv, jb, 5), jk, iblk_view(jv, jb, 5));
+
+ });
+ }
+}
+
+template
+void cells2verts_scalar_lib<double,double>(
+ const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const double* coeff_int, double* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async);
+
+template
+void cells2verts_scalar_lib<double,float>(
+ const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async);
+
+template
+void cells2verts_scalar_lib<float,float>(
+ const float* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async);
+
+//------------------------------------------------------------------------
+
+
+//>
+/// Same as above, but provides output optionally in single precision and
+/// assumes reversed index order of the output field in loop exchange mode
+///
+template <typename T, typename S>
+void cells2verts_scalar_ri_lib(
+ const T* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const T* coeff_int, S* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async)
+{
+ // Wrap raw pointers in unmanaged Kokkos Views.
+ typedef Kokkos::View<const T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstT3D;
+ typedef Kokkos::View<S***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedS3D;
+ typedef Kokkos::View<const int***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstInt3D;
+
+ // cell based scalar input field, dim: (nproma,nlev,nblks_c)
+ UnmanagedConstT3D p_cell_in_view(p_cell_in, nproma, nlev, nblks_c);
+
+ // line indices of cells around each vertex, dim: (nproma,nblks_v, 6)
+ UnmanagedConstInt3D iidx_view(vert_cell_idx, nproma, nblks_v, 6); // vert_cell_idx_view
+
+ // block indices of cells around each vertex, dim: (nproma,nblks_v, 6)
+ UnmanagedConstInt3D iblk_view(vert_cell_blk, nproma, nblks_v, 6); // vert_cell_blk_view
+
+ // coefficients for interpolation, dim: (nproma,9-cell_type,nblks_v)
+ UnmanagedConstT3D coeff_int_view(coeff_int, nproma, 9, nblks_v);
+
+ // vertex based scalar output field, dim: (nproma,nlev,nblks_c)
+ UnmanagedS3D p_vert_out_view(p_vert_out, nproma, nlev, nblks_c);
+
+ int i_startidx, i_endidx;
+
+ for (int jb = i_startblk; jb < i_endblk; ++jb){
+
+ get_indices_v_lib(i_startidx_in, i_endidx_in, nproma, jb, i_startblk, i_endblk,
+ i_startidx, i_endidx);
+
+ Kokkos::MDRangePolicy<Kokkos::Rank<2>> innerPolicy(
+ {slev, i_startidx}, {elev + 1, i_endidx + 1});
+
+ Kokkos::parallel_for("cells2verts_scalar_ri_lib", innerPolicy,
+ KOKKOS_LAMBDA(const int jk, const int jv) {
+
+ p_vert_out_view(jv, jk, jb) =
+ coeff_int_view(jv, 0, jb)*p_cell_in_view(iidx_view(jv, jb, 0), jk, iblk_view(jv, jb, 0)) +
+ coeff_int_view(jv, 1, jb)*p_cell_in_view(iidx_view(jv, jb, 1), jk, iblk_view(jv, jb, 1)) +
+ coeff_int_view(jv, 2, jb)*p_cell_in_view(iidx_view(jv, jb, 2), jk, iblk_view(jv, jb, 2)) +
+ coeff_int_view(jv, 3, jb)*p_cell_in_view(iidx_view(jv, jb, 3), jk, iblk_view(jv, jb, 3)) +
+ coeff_int_view(jv, 4, jb)*p_cell_in_view(iidx_view(jv, jb, 4), jk, iblk_view(jv, jb, 4)) +
+ coeff_int_view(jv, 5, jb)*p_cell_in_view(iidx_view(jv, jb, 5), jk, iblk_view(jv, jb, 5));
+
+ });
+ }
+
+}
+
+template
+void cells2verts_scalar_ri_lib<double,double>(
+ const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const double* coeff_int, double* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async);
+
+template
+void cells2verts_scalar_ri_lib<double,float>(
+ const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const double* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async);
+
+template
+void cells2verts_scalar_ri_lib<float,float>(
+ const float* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async);
+
+
+//>
+/// Computes average of scalar fields from vertices to centers of cells.
+///
+template <typename T>
+void verts2cells_scalar_lib(
+ const T* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
+ const T* coeff_int, T* p_cell_out, const int nblks_c, const int npromz_c,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblks_v, const bool lacc)
+{
+ // Wrap raw pointers in unmanaged Kokkos Views.
+ typedef Kokkos::View<const T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstT3D;
+ typedef Kokkos::View<T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedT3D;
+ typedef Kokkos::View<const int***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstInt3D;
+
+ // cell based scalar input field, dim: (nproma,nlev,nblks_v)
+ UnmanagedConstT3D p_vert_in_view(p_vert_in, nproma, nlev, nblks_v);
+
+ // line indices of vertices of triangles, dim: (nproma,nblks_c, 3)
+ UnmanagedConstInt3D iidx_view(cell_index_idx, nproma, nblks_c, 3); // cell_vertex_idx
+
+ // block indices of vertices of triangles, dim: (nproma,nblks_c, 3)
+ UnmanagedConstInt3D iblk_view(cell_vertex_blk, nproma, nblks_c, 3); // cell_vertex_blk
+
+ // coefficients for interpolation, dim: (nproma,3-cell_type,nblks_c)
+ UnmanagedConstT3D coeff_int_view(coeff_int, nproma, 3, nblks_c);
+
+ // vertex based scalar output field, dim: (nproma,nlev,nblks_c)
+ UnmanagedT3D p_cell_out_view(p_cell_out, nproma, nlev, nblks_c);
+
+ for (int jb = 1; jb<nblks_c; ++jb){
+
+ int nlen = (jb != nblks_c ? nproma: npromz_c);
+
+ Kokkos::MDRangePolicy<Kokkos::Rank<2>> innerPolicy(
+ {slev, 0}, {elev + 1, nlen + 1});
+
+ Kokkos::parallel_for("cell_avg_lib_inner", innerPolicy,
+ KOKKOS_LAMBDA(const int jk, const int jc) {
+
+ p_cell_out_view(jc, jk, jb) =
+ coeff_int_view(jc, 0, jb)*p_vert_in_view(iidx_view(jc, jb, 0), jk, iblk_view(jc, jb, 0)) +
+ coeff_int_view(jc, 1, jb)*p_vert_in_view(iidx_view(jc, jb, 1), jk, iblk_view(jc, jb, 1)) +
+ coeff_int_view(jc, 2, jb)*p_vert_in_view(iidx_view(jc, jb, 2), jk, iblk_view(jc, jb, 2));
+
+ });
+ }
+}
+
+template
+void verts2cells_scalar_lib<double>(
+ const double* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
+ const double* coeff_int, double* p_cell_out, const int nblks_c, const int npromz_c,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblks_v, const bool lacc);
+
+template
+void verts2cells_scalar_lib<float>(
+ const float* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
+ const float* coeff_int, float* p_cell_out, const int nblks_c, const int npromz_c,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblks_v, const bool lacc);
+
+
+//-------------------------------------------------------------------------
+//
+//
+//>
+/// Computes the average of a cell-based variable.
+///
+/// Computes the average of a cell-based variable
+/// over its original location and the neighboring triangles.
+/// Version with variable weighting coefficients, computed such that
+/// linear horizontal gradients are not aliased into a checkerboard noise
+/// input: lives on centers of triangles
+/// output: lives on centers of triangles
+///
+
+template <typename T>
+void cell_avg_lib(
+ const T* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
+ const T* avg_coeff, T* avg_psi_c, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const bool lacc)
+{
+ // Wrap raw pointers in unmanaged Kokkos Views.
+ typedef Kokkos::View<const T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstT3D;
+ typedef Kokkos::View<T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedT3D;
+ typedef Kokkos::View<const int***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstInt3D;
+
+ // cell based variable before averaging, dim: (nproma,nlev,nblks_c)
+ UnmanagedConstT3D psi_c_view(psi_c, nproma, nlev, nblks_c);
+ // line indices of triangles next to each cell, dim: (nproma,nblks_c, 3)
+ UnmanagedConstInt3D iidx_view(cell_neighbor_idx, nproma, nblks_c, 3); // cell_neighbour_idx
+ // block indices of triangles next to each cell, dim: (nproma,nblks_c, 3)
+ UnmanagedConstInt3D iblk_view(cell_neighbor_blk, nproma, nblks_c, 3); // cell_neighbour_blk
+ // averaging coefficients, dim: (nproma,nlev,nblks_c)
+ UnmanagedConstT3D avg_coeff_view(avg_coeff, nproma, nlev, nblks_c);
+
+ // cell based variable after averaging, dim: (nproma,nlev,nblks_c)
+ UnmanagedT3D avg_psi_c_view(avg_psi_c, nproma, nlev, nblks_c);
+
+ int i_startidx, i_endidx;
+
+ for (int jb = i_startblk; jb<i_endblk; ++jb){
+ get_indices_c_lib(i_startidx_in, i_endidx_in, nproma, jb, i_startblk, i_endblk,
+ i_startidx, i_endidx);
+
+ Kokkos::MDRangePolicy<Kokkos::Rank<2>> innerPolicy(
+ {slev, i_startidx}, {elev + 1, i_endidx + 1});
+
+ Kokkos::parallel_for("cell_avg_lib_inner", innerPolicy,
+ KOKKOS_LAMBDA(const int jk, const int jc) {
+ // calculate the weighted average
+
+ avg_psi_c_view(jc, jk, jb) =
+ psi_c_view(jc, jk, jb)*avg_coeff_view(jc, 0, jb) +
+ psi_c_view(iidx_view(jc, jb, 0), jk, iblk_view(jc, jb, 0))*avg_coeff_view(jc, 1, jb) +
+ psi_c_view(iidx_view(jc, jb, 1), jk, iblk_view(jc, jb, 1))*avg_coeff_view(jc, 2, jb) +
+ psi_c_view(iidx_view(jc, jb, 2), jk, iblk_view(jc, jb, 2))*avg_coeff_view(jc, 3, jb);
+ });
+ }
+}
+
+template
+void cell_avg_lib<double>(
+ const double* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
+ const double* avg_coeff, double* avg_psi_c, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c,
+ const bool lacc);
+
+template
+void cell_avg_lib<float>(
+ const float* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
+ const float* avg_coeff, float* avg_psi_c, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c,
+ const bool lacc);
\ No newline at end of file
diff --git a/src/interpolation/mo_lib_interpolation_scalar_Dylan.hpp b/src/interpolation/mo_lib_interpolation_scalar_Dylan.hpp
new file mode 100644
index 0000000..887dd4a
--- /dev/null
+++ b/src/interpolation/mo_lib_interpolation_scalar_Dylan.hpp
@@ -0,0 +1,54 @@
+// ICON
+//
+// ---------------------------------------------------------------
+// Copyright (C) 2004-2024, DWD, MPI-M, DKRZ, KIT, ETH, MeteoSwiss
+// Contact information: icon-model.org
+//
+// See AUTHORS.TXT for a list of authors
+// See LICENSES/ for license information
+// SPDX-License-Identifier: BSD-3-Clause
+// ---------------------------------------------------------------
+
+#pragma once
+
+template <typename T>
+void edges2cells_scalar_lib(
+ const T* p_edge_in, const int* edge_idx, const int* edge_blk,
+ const T* coeff_int, T* p_cell_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_e, const int nblks_c,
+ const bool lacc);
+
+template <typename T, typename S>
+void cells2verts_scalar_lib(
+ const T* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const S* coeff_int, S* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async);
+
+template <typename T, typename S>
+void cells2verts_scalar_ri_lib(
+ const T* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const T* coeff_int, S* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async);
+
+template <typename T>
+void verts2cells_scalar_lib(
+ const T* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
+ const T* coeff_int, T* p_cell_out, const int nblks_c, const int npromz_c,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblks_v, const bool lacc);
+
+template <typename T>
+void cell_avg_lib(
+ const T* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
+ const T* avg_coeff, T* avg_psi_c, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const bool lacc);
\ No newline at end of file
diff --git a/src/interpolation/mo_lib_interpolation_scalar_bindings_Dylan.cpp b/src/interpolation/mo_lib_interpolation_scalar_bindings_Dylan.cpp
new file mode 100644
index 0000000..dc3b509
--- /dev/null
+++ b/src/interpolation/mo_lib_interpolation_scalar_bindings_Dylan.cpp
@@ -0,0 +1,219 @@
+// ICON
+//
+// ---------------------------------------------------------------
+// Copyright (C) 2004-2024, DWD, MPI-M, DKRZ, KIT, ETH, MeteoSwiss
+// Contact information: icon-model.org
+//
+// See AUTHORS.TXT for a list of authors
+// See LICENSES/ for license information
+// SPDX-License-Identifier: BSD-3-Clause
+// ---------------------------------------------------------------
+
+#include "mo_lib_interpolation_scalar_bindings_Dylan.h"
+#include "mo_lib_interpolation_scalar_Dylan.hpp"
+
+void edges2cells_scalar_lib_dp(
+ const double* p_edge_in, const int* edge_idx, const int* edge_blk,
+ const double* coeff_int, double* p_cell_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_e, const int nblks_c,
+ const bool lacc)
+{
+ edges2cells_scalar_lib<double>(
+ p_edge_in, edge_idx, edge_blk,
+ coeff_int, p_cell_out, i_startblk, i_endblk,
+ i_startidx_in, i_endidx_in,
+ slev, elev, nproma,
+ nlev, nblks_e, nblks_c,
+ lacc);
+}
+
+void edges2cells_scalar_lib_sp(
+ const float* p_edge_in, const int* edge_idx, const int* edge_blk,
+ const float* coeff_int, float* p_cell_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_e, const int nblks_c,
+ const bool lacc)
+{
+
+ edges2cells_scalar_lib<float>(
+ p_edge_in, edge_idx, edge_blk,
+ coeff_int, p_cell_out, i_startblk, i_endblk,
+ i_startidx_in, i_endidx_in,
+ slev, elev, nproma,
+ nlev, nblks_e, nblks_c,
+ lacc);
+}
+
+/////////////////////////////////////////////
+
+void cells2verts_scalar_lib_dp(
+ const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const double* coeff_int, double* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async)
+{
+ cells2verts_scalar_lib<double,double>(
+ p_cell_in, vert_cell_idx, vert_cell_blk,
+ coeff_int, p_vert_out, i_startblk, i_endblk,
+ i_startidx_in, i_endidx_in,
+ slev, elev, nproma,
+ nlev, nblks_c, nblks_v,
+ lacc, acc_async);
+}
+
+void cells2verts_scalar_lib_dp2sp(
+ const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async)
+{
+ cells2verts_scalar_lib<double,float>(
+ p_cell_in, vert_cell_idx, vert_cell_blk,
+ coeff_int, p_vert_out, i_startblk, i_endblk,
+ i_startidx_in, i_endidx_in,
+ slev, elev, nproma,
+ nlev, nblks_c, nblks_v,
+ lacc, acc_async);
+}
+
+void cells2verts_scalar_lib_sp(
+ const float* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async)
+{
+ cells2verts_scalar_lib<float,float>(
+ p_cell_in, vert_cell_idx, vert_cell_blk,
+ coeff_int, p_vert_out, i_startblk, i_endblk,
+ i_startidx_in, i_endidx_in,
+ slev, elev, nproma,
+ nlev, nblks_c, nblks_v,
+ lacc, acc_async);
+}
+
+/////////////////////////////////////////////
+
+void cells2verts_scalar_ri_lib_dp(
+ const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const double* coeff_int, double* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async)
+{
+ cells2verts_scalar_ri_lib<double,double>(
+ p_cell_in, vert_cell_idx, vert_cell_blk,
+ coeff_int, p_vert_out, i_startblk, i_endblk,
+ i_startidx_in, i_endidx_in,
+ slev, elev, nproma,
+ nlev, nblks_c, nblks_v,
+ lacc, acc_async);
+}
+
+void cells2verts_scalar_ri_lib_dp2sp(
+ const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const double* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async)
+{
+ cells2verts_scalar_ri_lib<double,float>(
+ p_cell_in, vert_cell_idx, vert_cell_blk,
+ coeff_int, p_vert_out, i_startblk, i_endblk,
+ i_startidx_in, i_endidx_in,
+ slev, elev, nproma,
+ nlev, nblks_c, nblks_v,
+ lacc, acc_async);
+}
+
+void cells2verts_scalar_ri_lib_sp(
+ const float* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async)
+{
+ cells2verts_scalar_ri_lib<float,float>(
+ p_cell_in, vert_cell_idx, vert_cell_blk,
+ coeff_int, p_vert_out, i_startblk, i_endblk,
+ i_startidx_in, i_endidx_in,
+ slev, elev, nproma,
+ nlev, nblks_c, nblks_v,
+ lacc, acc_async);
+}
+
+/////////////////////////////////////////////
+
+void verts2cells_scalar_lib_dp(
+ const double* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
+ const double* coeff_int, double* p_cell_out, const int nblks_c, const int npromz_c,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblks_v, const bool lacc)
+{
+ verts2cells_scalar_lib<double>(
+ p_vert_in, cell_index_idx, cell_vertex_blk,
+ coeff_int, p_cell_out, nblks_c, npromz_c,
+ slev, elev, nproma, nlev,
+ nblks_v, lacc);
+}
+
+void verts2cells_scalar_lib_sp(
+ const float* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
+ const float* coeff_int, float* p_cell_out, const int nblks_c, const int npromz_c,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblks_v, const bool lacc)
+{
+ verts2cells_scalar_lib<float>(
+ p_vert_in, cell_index_idx, cell_vertex_blk,
+ coeff_int, p_cell_out, nblks_c, npromz_c,
+ slev, elev, nproma, nlev,
+ nblks_v, lacc);
+}
+
+/////////////////////////////////////////////
+
+void cell_avg_lib_dp(
+ const double* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
+ const double* avg_coeff, double* avg_psi_c, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c,
+ const bool lacc)
+{
+ cell_avg_lib<double>(
+ psi_c, cell_neighbor_idx, cell_neighbor_blk,
+ avg_coeff, avg_psi_c, i_startblk, i_endblk,
+ i_startidx_in, i_endidx_in,
+ slev, elev, nproma,
+ nlev, nblks_c,
+ lacc);
+}
+
+void cell_avg_lib_sp(
+ const float* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
+ const float* avg_coeff, float* avg_psi_c, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c,
+ const bool lacc)
+{
+ cell_avg_lib<float>(
+ psi_c, cell_neighbor_idx, cell_neighbor_blk,
+ avg_coeff, avg_psi_c, i_startblk, i_endblk,
+ i_startidx_in, i_endidx_in,
+ slev, elev, nproma,
+ nlev, nblks_c,
+ lacc);
+}
+
\ No newline at end of file
diff --git a/src/interpolation/mo_lib_interpolation_scalar_bindings_Dylan.h b/src/interpolation/mo_lib_interpolation_scalar_bindings_Dylan.h
new file mode 100644
index 0000000..4cb399b
--- /dev/null
+++ b/src/interpolation/mo_lib_interpolation_scalar_bindings_Dylan.h
@@ -0,0 +1,109 @@
+// ICON
+//
+// ---------------------------------------------------------------
+// Copyright (C) 2004-2024, DWD, MPI-M, DKRZ, KIT, ETH, MeteoSwiss
+// Contact information: icon-model.org
+//
+// See AUTHORS.TXT for a list of authors
+// See LICENSES/ for license information
+// SPDX-License-Identifier: BSD-3-Clause
+// ---------------------------------------------------------------
+
+#pragma once
+
+void edges2cells_scalar_lib_dp(
+ const double* p_edge_in, const int* edge_idx, const int* edge_blk,
+ const double* coeff_int, double* p_cell_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_e, const int nblks_c,
+ const bool lacc);
+void edges2cells_scalar_lib_sp(
+ const float* p_edge_in, const int* edge_idx, const int* edge_blk,
+ const float* coeff_int, float* p_cell_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_e, const int nblks_c,
+ const bool lacc);
+
+/////////////////////////////////////////////
+
+void cells2verts_scalar_lib_dp(
+ const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const double* coeff_int, double* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async);
+void cells2verts_scalar_lib_dp2sp(
+ const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async);
+void cells2verts_scalar_lib_sp(
+ const float* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async);
+
+/////////////////////////////////////////////
+
+void cells2verts_scalar_ri_lib_dp(
+ const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const double* coeff_int, double* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async);
+
+void cells2verts_scalar_ri_lib_dp2sp(
+ const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const double* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async);
+
+void cells2verts_scalar_ri_lib_sp(
+ const float* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async);
+
+/////////////////////////////////////////////
+
+void verts2cells_scalar_lib_dp(
+ const double* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
+ const double* coeff_int, double* p_cell_out, const int nblks_c, const int npromz_c,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblks_v, const bool lacc);
+
+void verts2cells_scalar_lib_sp(
+ const float* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
+ const float* coeff_int, float* p_cell_out, const int nblks_c, const int npromz_c,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblks_v, const bool lacc);
+
+/////////////////////////////////////////////
+
+void cell_avg_lib_dp(
+ const double* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
+ const double* avg_coeff, double* avg_psi_c, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c,
+ const bool lacc);
+void cell_avg_lib_sp(
+ const float* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
+ const float* avg_coeff, float* avg_psi_c, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c,
+ const bool lacc);
+
\ No newline at end of file
--
GitLab
From fc76537f1b2ed49bfa32cb806f6d2b9b0de70dcf Mon Sep 17 00:00:00 2001
From: Dylan Kierans <kierans@dkrz.de>
Date: Tue, 25 Feb 2025 15:59:46 +0100
Subject: [PATCH 09/36] adding license headers
---
src/interpolation/mo_lib_interpolation_scalar.cpp | 12 +++++++++++-
src/interpolation/mo_lib_interpolation_scalar.hpp | 11 +++++++++++
2 files changed, 22 insertions(+), 1 deletion(-)
diff --git a/src/interpolation/mo_lib_interpolation_scalar.cpp b/src/interpolation/mo_lib_interpolation_scalar.cpp
index 0994ac1..9a5f206 100644
--- a/src/interpolation/mo_lib_interpolation_scalar.cpp
+++ b/src/interpolation/mo_lib_interpolation_scalar.cpp
@@ -1,5 +1,15 @@
+// ICON
+//
+// ---------------------------------------------------------------
+// Copyright (C) 2004-2024, DWD, MPI-M, DKRZ, KIT, ETH, MeteoSwiss
+// Contact information: icon-model.org
+//
+// See AUTHORS.TXT for a list of authors
+// See LICENSES/ for license information
+// SPDX-License-Identifier: BSD-3-Clause
+// ---------------------------------------------------------------
+
#include <Kokkos_Core.hpp>
-#include <vector>
#include <iostream>
#include "mo_lib_loopindices.hpp"
#include "mo_lib_interpolation_scalar.hpp"
diff --git a/src/interpolation/mo_lib_interpolation_scalar.hpp b/src/interpolation/mo_lib_interpolation_scalar.hpp
index 1c578b0..ca52349 100644
--- a/src/interpolation/mo_lib_interpolation_scalar.hpp
+++ b/src/interpolation/mo_lib_interpolation_scalar.hpp
@@ -1,3 +1,14 @@
+// ICON
+//
+// ---------------------------------------------------------------
+// Copyright (C) 2004-2024, DWD, MPI-M, DKRZ, KIT, ETH, MeteoSwiss
+// Contact information: icon-model.org
+//
+// See AUTHORS.TXT for a list of authors
+// See LICENSES/ for license information
+// SPDX-License-Identifier: BSD-3-Clause
+// ---------------------------------------------------------------
+
#pragma once
template <typename T>
--
GitLab
From 00aa94ae6a88437922f65ebef65ce4b1681168e0 Mon Sep 17 00:00:00 2001
From: Dylan Kierans <kierans@dkrz.de>
Date: Tue, 25 Feb 2025 16:07:25 +0100
Subject: [PATCH 10/36] Merging Dylan and Harshada's parts of
mo_lib_interpolation_scalar
---
src/interpolation/CMakeLists.txt | 2 -
src/interpolation/interpolation_bindings.cpp | 202 +++++++++
src/interpolation/interpolation_bindings.h | 95 ++++
.../mo_lib_interpolation_scalar.cpp | 393 +++++++++++++++++
.../mo_lib_interpolation_scalar.hpp | 42 ++
.../mo_lib_interpolation_scalar_Dylan.cpp | 408 ------------------
.../mo_lib_interpolation_scalar_Dylan.hpp | 54 ---
...ib_interpolation_scalar_bindings_Dylan.cpp | 219 ----------
..._lib_interpolation_scalar_bindings_Dylan.h | 109 -----
9 files changed, 732 insertions(+), 792 deletions(-)
delete mode 100644 src/interpolation/mo_lib_interpolation_scalar_Dylan.cpp
delete mode 100644 src/interpolation/mo_lib_interpolation_scalar_Dylan.hpp
delete mode 100644 src/interpolation/mo_lib_interpolation_scalar_bindings_Dylan.cpp
delete mode 100644 src/interpolation/mo_lib_interpolation_scalar_bindings_Dylan.h
diff --git a/src/interpolation/CMakeLists.txt b/src/interpolation/CMakeLists.txt
index 3e8ef1a..eade738 100644
--- a/src/interpolation/CMakeLists.txt
+++ b/src/interpolation/CMakeLists.txt
@@ -12,8 +12,6 @@
add_library(
iconmath-interpolation
mo_lib_interpolation_scalar.F90
- mo_lib_interpolation_scalar_Dylan.cpp
- mo_lib_interpolation_scalar_bindings_Dylan.cpp
mo_lib_interpolation_vector.F90
mo_lib_interpolation_vector.cpp
mo_lib_intp_rbf.F90
diff --git a/src/interpolation/interpolation_bindings.cpp b/src/interpolation/interpolation_bindings.cpp
index 2589edf..4476035 100644
--- a/src/interpolation/interpolation_bindings.cpp
+++ b/src/interpolation/interpolation_bindings.cpp
@@ -154,6 +154,208 @@ void edges2verts_scalar_lib_sp(const float* p_edge_in, const int* vert_edge_idx,
}
+void edges2cells_scalar_lib_dp(
+ const double* p_edge_in, const int* edge_idx, const int* edge_blk,
+ const double* coeff_int, double* p_cell_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_e, const int nblks_c,
+ const bool lacc)
+{
+ edges2cells_scalar_lib<double>(
+ p_edge_in, edge_idx, edge_blk,
+ coeff_int, p_cell_out, i_startblk, i_endblk,
+ i_startidx_in, i_endidx_in,
+ slev, elev, nproma,
+ nlev, nblks_e, nblks_c,
+ lacc);
+}
+
+void edges2cells_scalar_lib_sp(
+ const float* p_edge_in, const int* edge_idx, const int* edge_blk,
+ const float* coeff_int, float* p_cell_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_e, const int nblks_c,
+ const bool lacc)
+{
+
+ edges2cells_scalar_lib<float>(
+ p_edge_in, edge_idx, edge_blk,
+ coeff_int, p_cell_out, i_startblk, i_endblk,
+ i_startidx_in, i_endidx_in,
+ slev, elev, nproma,
+ nlev, nblks_e, nblks_c,
+ lacc);
+}
+
+/////////////////////////////////////////////
+
+void cells2verts_scalar_lib_dp(
+ const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const double* coeff_int, double* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async)
+{
+ cells2verts_scalar_lib<double,double>(
+ p_cell_in, vert_cell_idx, vert_cell_blk,
+ coeff_int, p_vert_out, i_startblk, i_endblk,
+ i_startidx_in, i_endidx_in,
+ slev, elev, nproma,
+ nlev, nblks_c, nblks_v,
+ lacc, acc_async);
+}
+
+void cells2verts_scalar_lib_dp2sp(
+ const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async)
+{
+ cells2verts_scalar_lib<double,float>(
+ p_cell_in, vert_cell_idx, vert_cell_blk,
+ coeff_int, p_vert_out, i_startblk, i_endblk,
+ i_startidx_in, i_endidx_in,
+ slev, elev, nproma,
+ nlev, nblks_c, nblks_v,
+ lacc, acc_async);
+}
+
+void cells2verts_scalar_lib_sp(
+ const float* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async)
+{
+ cells2verts_scalar_lib<float,float>(
+ p_cell_in, vert_cell_idx, vert_cell_blk,
+ coeff_int, p_vert_out, i_startblk, i_endblk,
+ i_startidx_in, i_endidx_in,
+ slev, elev, nproma,
+ nlev, nblks_c, nblks_v,
+ lacc, acc_async);
+}
+
+/////////////////////////////////////////////
+void cells2verts_scalar_ri_lib_dp(
+ const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const double* coeff_int, double* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async)
+{
+ cells2verts_scalar_ri_lib<double,double>(
+ p_cell_in, vert_cell_idx, vert_cell_blk,
+ coeff_int, p_vert_out, i_startblk, i_endblk,
+ i_startidx_in, i_endidx_in,
+ slev, elev, nproma,
+ nlev, nblks_c, nblks_v,
+ lacc, acc_async);
+}
+
+void cells2verts_scalar_ri_lib_dp2sp(
+ const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const double* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async)
+{
+ cells2verts_scalar_ri_lib<double,float>(
+ p_cell_in, vert_cell_idx, vert_cell_blk,
+ coeff_int, p_vert_out, i_startblk, i_endblk,
+ i_startidx_in, i_endidx_in,
+ slev, elev, nproma,
+ nlev, nblks_c, nblks_v,
+ lacc, acc_async);
+}
+
+void cells2verts_scalar_ri_lib_sp(
+ const float* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async)
+{
+ cells2verts_scalar_ri_lib<float,float>(
+ p_cell_in, vert_cell_idx, vert_cell_blk,
+ coeff_int, p_vert_out, i_startblk, i_endblk,
+ i_startidx_in, i_endidx_in,
+ slev, elev, nproma,
+ nlev, nblks_c, nblks_v,
+ lacc, acc_async);
+}
+
+/////////////////////////////////////////////
+void verts2cells_scalar_lib_dp(
+ const double* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
+ const double* coeff_int, double* p_cell_out, const int nblks_c, const int npromz_c,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblks_v, const bool lacc)
+{
+ verts2cells_scalar_lib<double>(
+ p_vert_in, cell_index_idx, cell_vertex_blk,
+ coeff_int, p_cell_out, nblks_c, npromz_c,
+ slev, elev, nproma, nlev,
+ nblks_v, lacc);
+}
+
+void verts2cells_scalar_lib_sp(
+ const float* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
+ const float* coeff_int, float* p_cell_out, const int nblks_c, const int npromz_c,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblks_v, const bool lacc)
+{
+ verts2cells_scalar_lib<float>(
+ p_vert_in, cell_index_idx, cell_vertex_blk,
+ coeff_int, p_cell_out, nblks_c, npromz_c,
+ slev, elev, nproma, nlev,
+ nblks_v, lacc);
+}
+/////////////////////////////////////////////
+
+void cell_avg_lib_dp(
+ const double* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
+ const double* avg_coeff, double* avg_psi_c, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c,
+ const bool lacc)
+{
+ cell_avg_lib<double>(
+ psi_c, cell_neighbor_idx, cell_neighbor_blk,
+ avg_coeff, avg_psi_c, i_startblk, i_endblk,
+ i_startidx_in, i_endidx_in,
+ slev, elev, nproma,
+ nlev, nblks_c,
+ lacc);
+}
+
+void cell_avg_lib_sp(
+ const float* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
+ const float* avg_coeff, float* avg_psi_c, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c,
+ const bool lacc)
+{
+ cell_avg_lib<float>(
+ psi_c, cell_neighbor_idx, cell_neighbor_blk,
+ avg_coeff, avg_psi_c, i_startblk, i_endblk,
+ i_startidx_in, i_endidx_in,
+ slev, elev, nproma,
+ nlev, nblks_c,
+ lacc);
+}
+
diff --git a/src/interpolation/interpolation_bindings.h b/src/interpolation/interpolation_bindings.h
index 8567ef9..123c487 100644
--- a/src/interpolation/interpolation_bindings.h
+++ b/src/interpolation/interpolation_bindings.h
@@ -87,5 +87,100 @@ extern "C"{
const int elev, const int nproma, const int nlev, const int nblks_e,
const int nblks_v, const int cell_type, const bool lacc);
+ void edges2cells_scalar_lib_dp(
+ const double* p_edge_in, const int* edge_idx, const int* edge_blk,
+ const double* coeff_int, double* p_cell_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_e, const int nblks_c,
+ const bool lacc);
+ void edges2cells_scalar_lib_sp(
+ const float* p_edge_in, const int* edge_idx, const int* edge_blk,
+ const float* coeff_int, float* p_cell_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_e, const int nblks_c,
+ const bool lacc);
+
+ /////////////////////////////////////////////
+
+ void cells2verts_scalar_lib_dp(
+ const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const double* coeff_int, double* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async);
+ void cells2verts_scalar_lib_dp2sp(
+ const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async);
+ void cells2verts_scalar_lib_sp(
+ const float* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async);
+
+ /////////////////////////////////////////////
+
+ void cells2verts_scalar_ri_lib_dp(
+ const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const double* coeff_int, double* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async);
+
+ void cells2verts_scalar_ri_lib_dp2sp(
+ const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const double* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async);
+
+ void cells2verts_scalar_ri_lib_sp(
+ const float* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async);
+
+ /////////////////////////////////////////////
+
+ void verts2cells_scalar_lib_dp(
+ const double* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
+ const double* coeff_int, double* p_cell_out, const int nblks_c, const int npromz_c,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblks_v, const bool lacc);
+
+ void verts2cells_scalar_lib_sp(
+ const float* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
+ const float* coeff_int, float* p_cell_out, const int nblks_c, const int npromz_c,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblks_v, const bool lacc);
+
+ /////////////////////////////////////////////
+
+ void cell_avg_lib_dp(
+ const double* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
+ const double* avg_coeff, double* avg_psi_c, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c,
+ const bool lacc);
+ void cell_avg_lib_sp(
+ const float* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
+ const float* avg_coeff, float* avg_psi_c, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c,
+ const bool lacc);
}
\ No newline at end of file
diff --git a/src/interpolation/mo_lib_interpolation_scalar.cpp b/src/interpolation/mo_lib_interpolation_scalar.cpp
index 9a5f206..7f91246 100644
--- a/src/interpolation/mo_lib_interpolation_scalar.cpp
+++ b/src/interpolation/mo_lib_interpolation_scalar.cpp
@@ -229,3 +229,396 @@ void edges2verts_scalar_lib <float> (const float* p_edge_in, const int* vert_edg
const int nblks_v, const int cell_type, const bool lacc);
+//------------------------------------------------------------------------
+//
+//>
+/// Computes interpolation from edges to cells
+///
+/// Computes interpolation of scalar fields from velocity points to
+/// cell centers via given interpolation weights
+///
+template <typename T>
+void edges2cells_scalar_lib(
+ const T* p_edge_in, const int* edge_idx, const int* edge_blk,
+ const T* coeff_int, T* p_cell_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_e, const int nblks_c,
+ const bool lacc)
+{
+ // Wrap raw pointers in unmanaged Kokkos Views.
+ typedef Kokkos::View<const T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstT3D;
+ typedef Kokkos::View<T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedT3D;
+ typedef Kokkos::View<const int***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstInt3D;
+
+ // edge based scalar input field, dim: (nproma,nlev,nblks_e)
+ UnmanagedConstT3D p_edge_in_view(p_edge_in, nproma, nlev, nblks_e);
+
+ // line indices of edges of triangles, dim: (nproma,nblks_c, 3)
+ UnmanagedConstInt3D iidx_view(edge_idx, nproma, nblks_c, 3); // edge_idx_view
+
+ // block indices of edges of triangles, dim: (nproma,nblks_c, 3)
+ UnmanagedConstInt3D iblk_view(edge_blk, nproma, nblks_c, 3); // edge_blk_view
+
+ // coefficients for (area weighted) interpolation, dim: (nproma,3-cell_type,nblks_c)
+ UnmanagedConstT3D coeff_int_view(coeff_int, nproma, 3, nblks_c);
+
+ // cell based scalar output field, dim: (nproma,nlev,nblks_c)
+ UnmanagedT3D p_cell_out_view(p_cell_out, nproma, nlev, nblks_c);
+
+ int i_startidx, i_endidx;
+
+ for (int jb = i_startblk; jb < i_endblk; ++jb){
+ get_indices_c_lib(i_startidx_in, i_endidx_in, nproma, jb, i_startblk, i_endblk,
+ i_startidx, i_endidx);
+
+ Kokkos::MDRangePolicy<Kokkos::Rank<2>> innerPolicy(
+ {slev, i_startidx}, {elev + 1, i_endidx + 1});
+
+ Kokkos::parallel_for("edges2cells_scalar_lib_inner", innerPolicy,
+ KOKKOS_LAMBDA(const int jk, const int jc) {
+ p_cell_out_view(jc, jk, jb) =
+ coeff_int_view(jc, 1, jb)*p_edge_in_view(iidx_view(jc, jb, 1), jk, iblk_view(jc, jb, 1)) +
+ coeff_int_view(jc, 2, jb)*p_edge_in_view(iidx_view(jc, jb, 2), jk, iblk_view(jc, jb, 2)) +
+ coeff_int_view(jc, 3, jb)*p_edge_in_view(iidx_view(jc, jb, 3), jk, iblk_view(jc, jb, 3));
+ });
+ }
+
+}
+
+template
+void edges2cells_scalar_lib<double>(
+ const double* p_edge_in, const int* edge_idx, const int* edge_blk,
+ const double* coeff_int, double* p_cell_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_e, const int nblks_c,
+ const bool lacc);
+
+template
+void edges2cells_scalar_lib<float>(
+ const float* p_edge_in, const int* edge_idx, const int* edge_blk,
+ const float* coeff_int, float* p_cell_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_e, const int nblks_c,
+ const bool lacc);
+
+//------------------------------------------------------------------------
+//>
+//// Computes average of scalar fields from centers of cells to vertices.
+////
+template <typename T, typename S>
+void cells2verts_scalar_lib(
+ const T* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const S* coeff_int, S* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async)
+{
+ // Wrap raw pointers in unmanaged Kokkos Views.
+ typedef Kokkos::View<const T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstT3D;
+ typedef Kokkos::View<const S***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstS3D;
+ typedef Kokkos::View<S***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedS3D;
+ typedef Kokkos::View<const int***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstInt3D;
+
+ // cell based scalar input field, dim: (nproma,nlev,nblks_c)
+ UnmanagedConstT3D p_cell_in_view(p_cell_in, nproma, nlev, nblks_c);
+
+ // line indices of cells around each vertex, dim: (nproma,nblks_v, 6)
+ UnmanagedConstInt3D iidx_view(vert_cell_idx, nproma, nblks_v, 6); // vert_cell_idx_view
+
+ // block indices of cells around each vertex, dim: (nproma,nblks_v, 6)
+ UnmanagedConstInt3D iblk_view(vert_cell_blk, nproma, nblks_v, 6); // vert_cell_blk_view
+
+ // coefficients for interpolation, dim: (nproma,9-cell_type,nblks_v)
+ UnmanagedConstS3D coeff_int_view(coeff_int, nproma, 9, nblks_v);
+
+ // vertex based scalar output field, dim: (nproma,nlev,nblks_c)
+ UnmanagedS3D p_vert_out_view(p_vert_out, nproma, nlev, nblks_c);
+
+ int i_startidx, i_endidx;
+
+ for (int jb = i_startblk; jb < i_endblk; ++jb){
+
+ get_indices_v_lib(i_startidx_in, i_endidx_in, nproma, jb, i_startblk, i_endblk,
+ i_startidx, i_endidx);
+
+ Kokkos::MDRangePolicy<Kokkos::Rank<2>> innerPolicy(
+ {slev, i_startidx}, {elev + 1, i_endidx + 1});
+
+ Kokkos::parallel_for("cells2verts_scalar_lib", innerPolicy,
+ KOKKOS_LAMBDA(const int jk, const int jv) {
+
+ p_vert_out_view(jv, jk, jb) =
+ coeff_int_view(jv, 0, jb)*p_cell_in_view(iidx_view(jv, jb, 0), jk, iblk_view(jv, jb, 0)) +
+ coeff_int_view(jv, 1, jb)*p_cell_in_view(iidx_view(jv, jb, 1), jk, iblk_view(jv, jb, 1)) +
+ coeff_int_view(jv, 2, jb)*p_cell_in_view(iidx_view(jv, jb, 2), jk, iblk_view(jv, jb, 2)) +
+ coeff_int_view(jv, 3, jb)*p_cell_in_view(iidx_view(jv, jb, 3), jk, iblk_view(jv, jb, 3)) +
+ coeff_int_view(jv, 4, jb)*p_cell_in_view(iidx_view(jv, jb, 4), jk, iblk_view(jv, jb, 4)) +
+ coeff_int_view(jv, 5, jb)*p_cell_in_view(iidx_view(jv, jb, 5), jk, iblk_view(jv, jb, 5));
+
+ });
+ }
+}
+
+template
+void cells2verts_scalar_lib<double,double>(
+ const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const double* coeff_int, double* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async);
+
+template
+void cells2verts_scalar_lib<double,float>(
+ const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async);
+
+template
+void cells2verts_scalar_lib<float,float>(
+ const float* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async);
+
+//------------------------------------------------------------------------
+
+
+//>
+/// Same as above, but provides output optionally in single precision and
+/// assumes reversed index order of the output field in loop exchange mode
+///
+template <typename T, typename S>
+void cells2verts_scalar_ri_lib(
+ const T* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const T* coeff_int, S* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async)
+{
+ // Wrap raw pointers in unmanaged Kokkos Views.
+ typedef Kokkos::View<const T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstT3D;
+ typedef Kokkos::View<S***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedS3D;
+ typedef Kokkos::View<const int***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstInt3D;
+
+ // cell based scalar input field, dim: (nproma,nlev,nblks_c)
+ UnmanagedConstT3D p_cell_in_view(p_cell_in, nproma, nlev, nblks_c);
+
+ // line indices of cells around each vertex, dim: (nproma,nblks_v, 6)
+ UnmanagedConstInt3D iidx_view(vert_cell_idx, nproma, nblks_v, 6); // vert_cell_idx_view
+
+ // block indices of cells around each vertex, dim: (nproma,nblks_v, 6)
+ UnmanagedConstInt3D iblk_view(vert_cell_blk, nproma, nblks_v, 6); // vert_cell_blk_view
+
+ // coefficients for interpolation, dim: (nproma,9-cell_type,nblks_v)
+ UnmanagedConstT3D coeff_int_view(coeff_int, nproma, 9, nblks_v);
+
+ // vertex based scalar output field, dim: (nproma,nlev,nblks_c)
+ UnmanagedS3D p_vert_out_view(p_vert_out, nproma, nlev, nblks_c);
+
+ int i_startidx, i_endidx;
+
+ for (int jb = i_startblk; jb < i_endblk; ++jb){
+
+ get_indices_v_lib(i_startidx_in, i_endidx_in, nproma, jb, i_startblk, i_endblk,
+ i_startidx, i_endidx);
+
+ Kokkos::MDRangePolicy<Kokkos::Rank<2>> innerPolicy(
+ {slev, i_startidx}, {elev + 1, i_endidx + 1});
+
+ Kokkos::parallel_for("cells2verts_scalar_ri_lib", innerPolicy,
+ KOKKOS_LAMBDA(const int jk, const int jv) {
+
+ p_vert_out_view(jv, jk, jb) =
+ coeff_int_view(jv, 0, jb)*p_cell_in_view(iidx_view(jv, jb, 0), jk, iblk_view(jv, jb, 0)) +
+ coeff_int_view(jv, 1, jb)*p_cell_in_view(iidx_view(jv, jb, 1), jk, iblk_view(jv, jb, 1)) +
+ coeff_int_view(jv, 2, jb)*p_cell_in_view(iidx_view(jv, jb, 2), jk, iblk_view(jv, jb, 2)) +
+ coeff_int_view(jv, 3, jb)*p_cell_in_view(iidx_view(jv, jb, 3), jk, iblk_view(jv, jb, 3)) +
+ coeff_int_view(jv, 4, jb)*p_cell_in_view(iidx_view(jv, jb, 4), jk, iblk_view(jv, jb, 4)) +
+ coeff_int_view(jv, 5, jb)*p_cell_in_view(iidx_view(jv, jb, 5), jk, iblk_view(jv, jb, 5));
+
+ });
+ }
+
+}
+
+template
+void cells2verts_scalar_ri_lib<double,double>(
+ const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const double* coeff_int, double* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async);
+
+template
+void cells2verts_scalar_ri_lib<double,float>(
+ const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const double* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async);
+
+template
+void cells2verts_scalar_ri_lib<float,float>(
+ const float* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async);
+
+
+//>
+/// Computes average of scalar fields from vertices to centers of cells.
+///
+template <typename T>
+void verts2cells_scalar_lib(
+ const T* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
+ const T* coeff_int, T* p_cell_out, const int nblks_c, const int npromz_c,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblks_v, const bool lacc)
+{
+ // Wrap raw pointers in unmanaged Kokkos Views.
+ typedef Kokkos::View<const T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstT3D;
+ typedef Kokkos::View<T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedT3D;
+ typedef Kokkos::View<const int***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstInt3D;
+
+ // cell based scalar input field, dim: (nproma,nlev,nblks_v)
+ UnmanagedConstT3D p_vert_in_view(p_vert_in, nproma, nlev, nblks_v);
+
+ // line indices of vertices of triangles, dim: (nproma,nblks_c, 3)
+ UnmanagedConstInt3D iidx_view(cell_index_idx, nproma, nblks_c, 3); // cell_vertex_idx
+
+ // block indices of vertices of triangles, dim: (nproma,nblks_c, 3)
+ UnmanagedConstInt3D iblk_view(cell_vertex_blk, nproma, nblks_c, 3); // cell_vertex_blk
+
+ // coefficients for interpolation, dim: (nproma,3-cell_type,nblks_c)
+ UnmanagedConstT3D coeff_int_view(coeff_int, nproma, 3, nblks_c);
+
+ // vertex based scalar output field, dim: (nproma,nlev,nblks_c)
+ UnmanagedT3D p_cell_out_view(p_cell_out, nproma, nlev, nblks_c);
+
+ for (int jb = 1; jb<nblks_c; ++jb){
+
+ int nlen = (jb != nblks_c ? nproma: npromz_c);
+
+ Kokkos::MDRangePolicy<Kokkos::Rank<2>> innerPolicy(
+ {slev, 0}, {elev + 1, nlen + 1});
+
+ Kokkos::parallel_for("cell_avg_lib_inner", innerPolicy,
+ KOKKOS_LAMBDA(const int jk, const int jc) {
+
+ p_cell_out_view(jc, jk, jb) =
+ coeff_int_view(jc, 0, jb)*p_vert_in_view(iidx_view(jc, jb, 0), jk, iblk_view(jc, jb, 0)) +
+ coeff_int_view(jc, 1, jb)*p_vert_in_view(iidx_view(jc, jb, 1), jk, iblk_view(jc, jb, 1)) +
+ coeff_int_view(jc, 2, jb)*p_vert_in_view(iidx_view(jc, jb, 2), jk, iblk_view(jc, jb, 2));
+
+ });
+ }
+}
+
+template
+void verts2cells_scalar_lib<double>(
+ const double* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
+ const double* coeff_int, double* p_cell_out, const int nblks_c, const int npromz_c,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblks_v, const bool lacc);
+
+template
+void verts2cells_scalar_lib<float>(
+ const float* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
+ const float* coeff_int, float* p_cell_out, const int nblks_c, const int npromz_c,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblks_v, const bool lacc);
+
+
+//-------------------------------------------------------------------------
+//
+//
+//>
+/// Computes the average of a cell-based variable.
+///
+/// Computes the average of a cell-based variable
+/// over its original location and the neighboring triangles.
+/// Version with variable weighting coefficients, computed such that
+/// linear horizontal gradients are not aliased into a checkerboard noise
+/// input: lives on centers of triangles
+/// output: lives on centers of triangles
+///
+
+template <typename T>
+void cell_avg_lib(
+ const T* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
+ const T* avg_coeff, T* avg_psi_c, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const bool lacc)
+{
+ // Wrap raw pointers in unmanaged Kokkos Views.
+ typedef Kokkos::View<const T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstT3D;
+ typedef Kokkos::View<T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedT3D;
+ typedef Kokkos::View<const int***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstInt3D;
+
+ // cell based variable before averaging, dim: (nproma,nlev,nblks_c)
+ UnmanagedConstT3D psi_c_view(psi_c, nproma, nlev, nblks_c);
+ // line indices of triangles next to each cell, dim: (nproma,nblks_c, 3)
+ UnmanagedConstInt3D iidx_view(cell_neighbor_idx, nproma, nblks_c, 3); // cell_neighbour_idx
+ // block indices of triangles next to each cell, dim: (nproma,nblks_c, 3)
+ UnmanagedConstInt3D iblk_view(cell_neighbor_blk, nproma, nblks_c, 3); // cell_neighbour_blk
+ // averaging coefficients, dim: (nproma,nlev,nblks_c)
+ UnmanagedConstT3D avg_coeff_view(avg_coeff, nproma, nlev, nblks_c);
+
+ // cell based variable after averaging, dim: (nproma,nlev,nblks_c)
+ UnmanagedT3D avg_psi_c_view(avg_psi_c, nproma, nlev, nblks_c);
+
+ int i_startidx, i_endidx;
+
+ for (int jb = i_startblk; jb<i_endblk; ++jb){
+ get_indices_c_lib(i_startidx_in, i_endidx_in, nproma, jb, i_startblk, i_endblk,
+ i_startidx, i_endidx);
+
+ Kokkos::MDRangePolicy<Kokkos::Rank<2>> innerPolicy(
+ {slev, i_startidx}, {elev + 1, i_endidx + 1});
+
+ Kokkos::parallel_for("cell_avg_lib_inner", innerPolicy,
+ KOKKOS_LAMBDA(const int jk, const int jc) {
+ // calculate the weighted average
+
+ avg_psi_c_view(jc, jk, jb) =
+ psi_c_view(jc, jk, jb)*avg_coeff_view(jc, 0, jb) +
+ psi_c_view(iidx_view(jc, jb, 0), jk, iblk_view(jc, jb, 0))*avg_coeff_view(jc, 1, jb) +
+ psi_c_view(iidx_view(jc, jb, 1), jk, iblk_view(jc, jb, 1))*avg_coeff_view(jc, 2, jb) +
+ psi_c_view(iidx_view(jc, jb, 2), jk, iblk_view(jc, jb, 2))*avg_coeff_view(jc, 3, jb);
+ });
+ }
+}
+
+template
+void cell_avg_lib<double>(
+ const double* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
+ const double* avg_coeff, double* avg_psi_c, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c,
+ const bool lacc);
+
+template
+void cell_avg_lib<float>(
+ const float* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
+ const float* avg_coeff, float* avg_psi_c, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c,
+ const bool lacc);
diff --git a/src/interpolation/mo_lib_interpolation_scalar.hpp b/src/interpolation/mo_lib_interpolation_scalar.hpp
index ca52349..d92cee0 100644
--- a/src/interpolation/mo_lib_interpolation_scalar.hpp
+++ b/src/interpolation/mo_lib_interpolation_scalar.hpp
@@ -35,3 +35,45 @@ void edges2verts_scalar_lib (const T* p_edge_in, const int* vert_edge_idx, const
const int i_startidx_in, const int i_endidx_in, const int slev,
const int elev, const int nproma, const int nlev, const int nblks_e,
const int nblks_v, const int cell_type, const bool lacc);
+
+template <typename T>
+void edges2cells_scalar_lib(
+ const T* p_edge_in, const int* edge_idx, const int* edge_blk,
+ const T* coeff_int, T* p_cell_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_e, const int nblks_c,
+ const bool lacc);
+
+template <typename T, typename S>
+void cells2verts_scalar_lib(
+ const T* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const S* coeff_int, S* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async);
+
+template <typename T, typename S>
+void cells2verts_scalar_ri_lib(
+ const T* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const T* coeff_int, S* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async);
+
+template <typename T>
+void verts2cells_scalar_lib(
+ const T* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
+ const T* coeff_int, T* p_cell_out, const int nblks_c, const int npromz_c,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblks_v, const bool lacc);
+
+template <typename T>
+void cell_avg_lib(
+ const T* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
+ const T* avg_coeff, T* avg_psi_c, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const bool lacc);
\ No newline at end of file
diff --git a/src/interpolation/mo_lib_interpolation_scalar_Dylan.cpp b/src/interpolation/mo_lib_interpolation_scalar_Dylan.cpp
deleted file mode 100644
index 8b133c5..0000000
--- a/src/interpolation/mo_lib_interpolation_scalar_Dylan.cpp
+++ /dev/null
@@ -1,408 +0,0 @@
-// ICON
-//
-// ---------------------------------------------------------------
-// Copyright (C) 2004-2024, DWD, MPI-M, DKRZ, KIT, ETH, MeteoSwiss
-// Contact information: icon-model.org
-//
-// See AUTHORS.TXT for a list of authors
-// See LICENSES/ for license information
-// SPDX-License-Identifier: BSD-3-Clause
-// ---------------------------------------------------------------
-
-#include <Kokkos_Core.hpp>
-#include "mo_lib_loopindices.hpp"
-#include "mo_lib_interpolation_scalar_Dylan.hpp"
-
-//------------------------------------------------------------------------
-//
-//>
-/// Computes interpolation from edges to cells
-///
-/// Computes interpolation of scalar fields from velocity points to
-/// cell centers via given interpolation weights
-///
-template <typename T>
-void edges2cells_scalar_lib(
- const T* p_edge_in, const int* edge_idx, const int* edge_blk,
- const T* coeff_int, T* p_cell_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_e, const int nblks_c,
- const bool lacc)
-{
- // Wrap raw pointers in unmanaged Kokkos Views.
- typedef Kokkos::View<const T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstT3D;
- typedef Kokkos::View<T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedT3D;
- typedef Kokkos::View<const int***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstInt3D;
-
- // edge based scalar input field, dim: (nproma,nlev,nblks_e)
- UnmanagedConstT3D p_edge_in_view(p_edge_in, nproma, nlev, nblks_e);
-
- // line indices of edges of triangles, dim: (nproma,nblks_c, 3)
- UnmanagedConstInt3D iidx_view(edge_idx, nproma, nblks_c, 3); // edge_idx_view
-
- // block indices of edges of triangles, dim: (nproma,nblks_c, 3)
- UnmanagedConstInt3D iblk_view(edge_blk, nproma, nblks_c, 3); // edge_blk_view
-
- // coefficients for (area weighted) interpolation, dim: (nproma,3-cell_type,nblks_c)
- UnmanagedConstT3D coeff_int_view(coeff_int, nproma, 3, nblks_c);
-
- // cell based scalar output field, dim: (nproma,nlev,nblks_c)
- UnmanagedT3D p_cell_out_view(p_cell_out, nproma, nlev, nblks_c);
-
- int i_startidx, i_endidx;
-
- for (int jb = i_startblk; jb < i_endblk; ++jb){
- get_indices_c_lib(i_startidx_in, i_endidx_in, nproma, jb, i_startblk, i_endblk,
- i_startidx, i_endidx);
-
- Kokkos::MDRangePolicy<Kokkos::Rank<2>> innerPolicy(
- {slev, i_startidx}, {elev + 1, i_endidx + 1});
-
- Kokkos::parallel_for("edges2cells_scalar_lib_inner", innerPolicy,
- KOKKOS_LAMBDA(const int jk, const int jc) {
- p_cell_out_view(jc, jk, jb) =
- coeff_int_view(jc, 1, jb)*p_edge_in_view(iidx_view(jc, jb, 1), jk, iblk_view(jc, jb, 1)) +
- coeff_int_view(jc, 2, jb)*p_edge_in_view(iidx_view(jc, jb, 2), jk, iblk_view(jc, jb, 2)) +
- coeff_int_view(jc, 3, jb)*p_edge_in_view(iidx_view(jc, jb, 3), jk, iblk_view(jc, jb, 3));
- });
- }
-
-}
-
-template
-void edges2cells_scalar_lib<double>(
- const double* p_edge_in, const int* edge_idx, const int* edge_blk,
- const double* coeff_int, double* p_cell_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_e, const int nblks_c,
- const bool lacc);
-
-template
-void edges2cells_scalar_lib<float>(
- const float* p_edge_in, const int* edge_idx, const int* edge_blk,
- const float* coeff_int, float* p_cell_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_e, const int nblks_c,
- const bool lacc);
-
-//------------------------------------------------------------------------
-//>
-//// Computes average of scalar fields from centers of cells to vertices.
-////
-template <typename T, typename S>
-void cells2verts_scalar_lib(
- const T* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const S* coeff_int, S* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async)
-{
- // Wrap raw pointers in unmanaged Kokkos Views.
- typedef Kokkos::View<const T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstT3D;
- typedef Kokkos::View<const S***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstS3D;
- typedef Kokkos::View<S***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedS3D;
- typedef Kokkos::View<const int***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstInt3D;
-
- // cell based scalar input field, dim: (nproma,nlev,nblks_c)
- UnmanagedConstT3D p_cell_in_view(p_cell_in, nproma, nlev, nblks_c);
-
- // line indices of cells around each vertex, dim: (nproma,nblks_v, 6)
- UnmanagedConstInt3D iidx_view(vert_cell_idx, nproma, nblks_v, 6); // vert_cell_idx_view
-
- // block indices of cells around each vertex, dim: (nproma,nblks_v, 6)
- UnmanagedConstInt3D iblk_view(vert_cell_blk, nproma, nblks_v, 6); // vert_cell_blk_view
-
- // coefficients for interpolation, dim: (nproma,9-cell_type,nblks_v)
- UnmanagedConstS3D coeff_int_view(coeff_int, nproma, 9, nblks_v);
-
- // vertex based scalar output field, dim: (nproma,nlev,nblks_c)
- UnmanagedS3D p_vert_out_view(p_vert_out, nproma, nlev, nblks_c);
-
- int i_startidx, i_endidx;
-
- for (int jb = i_startblk; jb < i_endblk; ++jb){
-
- get_indices_v_lib(i_startidx_in, i_endidx_in, nproma, jb, i_startblk, i_endblk,
- i_startidx, i_endidx);
-
- Kokkos::MDRangePolicy<Kokkos::Rank<2>> innerPolicy(
- {slev, i_startidx}, {elev + 1, i_endidx + 1});
-
- Kokkos::parallel_for("cells2verts_scalar_lib", innerPolicy,
- KOKKOS_LAMBDA(const int jk, const int jv) {
-
- p_vert_out_view(jv, jk, jb) =
- coeff_int_view(jv, 0, jb)*p_cell_in_view(iidx_view(jv, jb, 0), jk, iblk_view(jv, jb, 0)) +
- coeff_int_view(jv, 1, jb)*p_cell_in_view(iidx_view(jv, jb, 1), jk, iblk_view(jv, jb, 1)) +
- coeff_int_view(jv, 2, jb)*p_cell_in_view(iidx_view(jv, jb, 2), jk, iblk_view(jv, jb, 2)) +
- coeff_int_view(jv, 3, jb)*p_cell_in_view(iidx_view(jv, jb, 3), jk, iblk_view(jv, jb, 3)) +
- coeff_int_view(jv, 4, jb)*p_cell_in_view(iidx_view(jv, jb, 4), jk, iblk_view(jv, jb, 4)) +
- coeff_int_view(jv, 5, jb)*p_cell_in_view(iidx_view(jv, jb, 5), jk, iblk_view(jv, jb, 5));
-
- });
- }
-}
-
-template
-void cells2verts_scalar_lib<double,double>(
- const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const double* coeff_int, double* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async);
-
-template
-void cells2verts_scalar_lib<double,float>(
- const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async);
-
-template
-void cells2verts_scalar_lib<float,float>(
- const float* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async);
-
-//------------------------------------------------------------------------
-
-
-//>
-/// Same as above, but provides output optionally in single precision and
-/// assumes reversed index order of the output field in loop exchange mode
-///
-template <typename T, typename S>
-void cells2verts_scalar_ri_lib(
- const T* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const T* coeff_int, S* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async)
-{
- // Wrap raw pointers in unmanaged Kokkos Views.
- typedef Kokkos::View<const T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstT3D;
- typedef Kokkos::View<S***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedS3D;
- typedef Kokkos::View<const int***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstInt3D;
-
- // cell based scalar input field, dim: (nproma,nlev,nblks_c)
- UnmanagedConstT3D p_cell_in_view(p_cell_in, nproma, nlev, nblks_c);
-
- // line indices of cells around each vertex, dim: (nproma,nblks_v, 6)
- UnmanagedConstInt3D iidx_view(vert_cell_idx, nproma, nblks_v, 6); // vert_cell_idx_view
-
- // block indices of cells around each vertex, dim: (nproma,nblks_v, 6)
- UnmanagedConstInt3D iblk_view(vert_cell_blk, nproma, nblks_v, 6); // vert_cell_blk_view
-
- // coefficients for interpolation, dim: (nproma,9-cell_type,nblks_v)
- UnmanagedConstT3D coeff_int_view(coeff_int, nproma, 9, nblks_v);
-
- // vertex based scalar output field, dim: (nproma,nlev,nblks_c)
- UnmanagedS3D p_vert_out_view(p_vert_out, nproma, nlev, nblks_c);
-
- int i_startidx, i_endidx;
-
- for (int jb = i_startblk; jb < i_endblk; ++jb){
-
- get_indices_v_lib(i_startidx_in, i_endidx_in, nproma, jb, i_startblk, i_endblk,
- i_startidx, i_endidx);
-
- Kokkos::MDRangePolicy<Kokkos::Rank<2>> innerPolicy(
- {slev, i_startidx}, {elev + 1, i_endidx + 1});
-
- Kokkos::parallel_for("cells2verts_scalar_ri_lib", innerPolicy,
- KOKKOS_LAMBDA(const int jk, const int jv) {
-
- p_vert_out_view(jv, jk, jb) =
- coeff_int_view(jv, 0, jb)*p_cell_in_view(iidx_view(jv, jb, 0), jk, iblk_view(jv, jb, 0)) +
- coeff_int_view(jv, 1, jb)*p_cell_in_view(iidx_view(jv, jb, 1), jk, iblk_view(jv, jb, 1)) +
- coeff_int_view(jv, 2, jb)*p_cell_in_view(iidx_view(jv, jb, 2), jk, iblk_view(jv, jb, 2)) +
- coeff_int_view(jv, 3, jb)*p_cell_in_view(iidx_view(jv, jb, 3), jk, iblk_view(jv, jb, 3)) +
- coeff_int_view(jv, 4, jb)*p_cell_in_view(iidx_view(jv, jb, 4), jk, iblk_view(jv, jb, 4)) +
- coeff_int_view(jv, 5, jb)*p_cell_in_view(iidx_view(jv, jb, 5), jk, iblk_view(jv, jb, 5));
-
- });
- }
-
-}
-
-template
-void cells2verts_scalar_ri_lib<double,double>(
- const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const double* coeff_int, double* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async);
-
-template
-void cells2verts_scalar_ri_lib<double,float>(
- const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const double* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async);
-
-template
-void cells2verts_scalar_ri_lib<float,float>(
- const float* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async);
-
-
-//>
-/// Computes average of scalar fields from vertices to centers of cells.
-///
-template <typename T>
-void verts2cells_scalar_lib(
- const T* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
- const T* coeff_int, T* p_cell_out, const int nblks_c, const int npromz_c,
- const int slev, const int elev, const int nproma, const int nlev,
- const int nblks_v, const bool lacc)
-{
- // Wrap raw pointers in unmanaged Kokkos Views.
- typedef Kokkos::View<const T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstT3D;
- typedef Kokkos::View<T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedT3D;
- typedef Kokkos::View<const int***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstInt3D;
-
- // cell based scalar input field, dim: (nproma,nlev,nblks_v)
- UnmanagedConstT3D p_vert_in_view(p_vert_in, nproma, nlev, nblks_v);
-
- // line indices of vertices of triangles, dim: (nproma,nblks_c, 3)
- UnmanagedConstInt3D iidx_view(cell_index_idx, nproma, nblks_c, 3); // cell_vertex_idx
-
- // block indices of vertices of triangles, dim: (nproma,nblks_c, 3)
- UnmanagedConstInt3D iblk_view(cell_vertex_blk, nproma, nblks_c, 3); // cell_vertex_blk
-
- // coefficients for interpolation, dim: (nproma,3-cell_type,nblks_c)
- UnmanagedConstT3D coeff_int_view(coeff_int, nproma, 3, nblks_c);
-
- // vertex based scalar output field, dim: (nproma,nlev,nblks_c)
- UnmanagedT3D p_cell_out_view(p_cell_out, nproma, nlev, nblks_c);
-
- for (int jb = 1; jb<nblks_c; ++jb){
-
- int nlen = (jb != nblks_c ? nproma: npromz_c);
-
- Kokkos::MDRangePolicy<Kokkos::Rank<2>> innerPolicy(
- {slev, 0}, {elev + 1, nlen + 1});
-
- Kokkos::parallel_for("cell_avg_lib_inner", innerPolicy,
- KOKKOS_LAMBDA(const int jk, const int jc) {
-
- p_cell_out_view(jc, jk, jb) =
- coeff_int_view(jc, 0, jb)*p_vert_in_view(iidx_view(jc, jb, 0), jk, iblk_view(jc, jb, 0)) +
- coeff_int_view(jc, 1, jb)*p_vert_in_view(iidx_view(jc, jb, 1), jk, iblk_view(jc, jb, 1)) +
- coeff_int_view(jc, 2, jb)*p_vert_in_view(iidx_view(jc, jb, 2), jk, iblk_view(jc, jb, 2));
-
- });
- }
-}
-
-template
-void verts2cells_scalar_lib<double>(
- const double* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
- const double* coeff_int, double* p_cell_out, const int nblks_c, const int npromz_c,
- const int slev, const int elev, const int nproma, const int nlev,
- const int nblks_v, const bool lacc);
-
-template
-void verts2cells_scalar_lib<float>(
- const float* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
- const float* coeff_int, float* p_cell_out, const int nblks_c, const int npromz_c,
- const int slev, const int elev, const int nproma, const int nlev,
- const int nblks_v, const bool lacc);
-
-
-//-------------------------------------------------------------------------
-//
-//
-//>
-/// Computes the average of a cell-based variable.
-///
-/// Computes the average of a cell-based variable
-/// over its original location and the neighboring triangles.
-/// Version with variable weighting coefficients, computed such that
-/// linear horizontal gradients are not aliased into a checkerboard noise
-/// input: lives on centers of triangles
-/// output: lives on centers of triangles
-///
-
-template <typename T>
-void cell_avg_lib(
- const T* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
- const T* avg_coeff, T* avg_psi_c, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const bool lacc)
-{
- // Wrap raw pointers in unmanaged Kokkos Views.
- typedef Kokkos::View<const T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstT3D;
- typedef Kokkos::View<T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedT3D;
- typedef Kokkos::View<const int***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstInt3D;
-
- // cell based variable before averaging, dim: (nproma,nlev,nblks_c)
- UnmanagedConstT3D psi_c_view(psi_c, nproma, nlev, nblks_c);
- // line indices of triangles next to each cell, dim: (nproma,nblks_c, 3)
- UnmanagedConstInt3D iidx_view(cell_neighbor_idx, nproma, nblks_c, 3); // cell_neighbour_idx
- // block indices of triangles next to each cell, dim: (nproma,nblks_c, 3)
- UnmanagedConstInt3D iblk_view(cell_neighbor_blk, nproma, nblks_c, 3); // cell_neighbour_blk
- // averaging coefficients, dim: (nproma,nlev,nblks_c)
- UnmanagedConstT3D avg_coeff_view(avg_coeff, nproma, nlev, nblks_c);
-
- // cell based variable after averaging, dim: (nproma,nlev,nblks_c)
- UnmanagedT3D avg_psi_c_view(avg_psi_c, nproma, nlev, nblks_c);
-
- int i_startidx, i_endidx;
-
- for (int jb = i_startblk; jb<i_endblk; ++jb){
- get_indices_c_lib(i_startidx_in, i_endidx_in, nproma, jb, i_startblk, i_endblk,
- i_startidx, i_endidx);
-
- Kokkos::MDRangePolicy<Kokkos::Rank<2>> innerPolicy(
- {slev, i_startidx}, {elev + 1, i_endidx + 1});
-
- Kokkos::parallel_for("cell_avg_lib_inner", innerPolicy,
- KOKKOS_LAMBDA(const int jk, const int jc) {
- // calculate the weighted average
-
- avg_psi_c_view(jc, jk, jb) =
- psi_c_view(jc, jk, jb)*avg_coeff_view(jc, 0, jb) +
- psi_c_view(iidx_view(jc, jb, 0), jk, iblk_view(jc, jb, 0))*avg_coeff_view(jc, 1, jb) +
- psi_c_view(iidx_view(jc, jb, 1), jk, iblk_view(jc, jb, 1))*avg_coeff_view(jc, 2, jb) +
- psi_c_view(iidx_view(jc, jb, 2), jk, iblk_view(jc, jb, 2))*avg_coeff_view(jc, 3, jb);
- });
- }
-}
-
-template
-void cell_avg_lib<double>(
- const double* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
- const double* avg_coeff, double* avg_psi_c, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c,
- const bool lacc);
-
-template
-void cell_avg_lib<float>(
- const float* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
- const float* avg_coeff, float* avg_psi_c, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c,
- const bool lacc);
\ No newline at end of file
diff --git a/src/interpolation/mo_lib_interpolation_scalar_Dylan.hpp b/src/interpolation/mo_lib_interpolation_scalar_Dylan.hpp
deleted file mode 100644
index 887dd4a..0000000
--- a/src/interpolation/mo_lib_interpolation_scalar_Dylan.hpp
+++ /dev/null
@@ -1,54 +0,0 @@
-// ICON
-//
-// ---------------------------------------------------------------
-// Copyright (C) 2004-2024, DWD, MPI-M, DKRZ, KIT, ETH, MeteoSwiss
-// Contact information: icon-model.org
-//
-// See AUTHORS.TXT for a list of authors
-// See LICENSES/ for license information
-// SPDX-License-Identifier: BSD-3-Clause
-// ---------------------------------------------------------------
-
-#pragma once
-
-template <typename T>
-void edges2cells_scalar_lib(
- const T* p_edge_in, const int* edge_idx, const int* edge_blk,
- const T* coeff_int, T* p_cell_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_e, const int nblks_c,
- const bool lacc);
-
-template <typename T, typename S>
-void cells2verts_scalar_lib(
- const T* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const S* coeff_int, S* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async);
-
-template <typename T, typename S>
-void cells2verts_scalar_ri_lib(
- const T* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const T* coeff_int, S* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async);
-
-template <typename T>
-void verts2cells_scalar_lib(
- const T* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
- const T* coeff_int, T* p_cell_out, const int nblks_c, const int npromz_c,
- const int slev, const int elev, const int nproma, const int nlev,
- const int nblks_v, const bool lacc);
-
-template <typename T>
-void cell_avg_lib(
- const T* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
- const T* avg_coeff, T* avg_psi_c, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const bool lacc);
\ No newline at end of file
diff --git a/src/interpolation/mo_lib_interpolation_scalar_bindings_Dylan.cpp b/src/interpolation/mo_lib_interpolation_scalar_bindings_Dylan.cpp
deleted file mode 100644
index dc3b509..0000000
--- a/src/interpolation/mo_lib_interpolation_scalar_bindings_Dylan.cpp
+++ /dev/null
@@ -1,219 +0,0 @@
-// ICON
-//
-// ---------------------------------------------------------------
-// Copyright (C) 2004-2024, DWD, MPI-M, DKRZ, KIT, ETH, MeteoSwiss
-// Contact information: icon-model.org
-//
-// See AUTHORS.TXT for a list of authors
-// See LICENSES/ for license information
-// SPDX-License-Identifier: BSD-3-Clause
-// ---------------------------------------------------------------
-
-#include "mo_lib_interpolation_scalar_bindings_Dylan.h"
-#include "mo_lib_interpolation_scalar_Dylan.hpp"
-
-void edges2cells_scalar_lib_dp(
- const double* p_edge_in, const int* edge_idx, const int* edge_blk,
- const double* coeff_int, double* p_cell_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_e, const int nblks_c,
- const bool lacc)
-{
- edges2cells_scalar_lib<double>(
- p_edge_in, edge_idx, edge_blk,
- coeff_int, p_cell_out, i_startblk, i_endblk,
- i_startidx_in, i_endidx_in,
- slev, elev, nproma,
- nlev, nblks_e, nblks_c,
- lacc);
-}
-
-void edges2cells_scalar_lib_sp(
- const float* p_edge_in, const int* edge_idx, const int* edge_blk,
- const float* coeff_int, float* p_cell_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_e, const int nblks_c,
- const bool lacc)
-{
-
- edges2cells_scalar_lib<float>(
- p_edge_in, edge_idx, edge_blk,
- coeff_int, p_cell_out, i_startblk, i_endblk,
- i_startidx_in, i_endidx_in,
- slev, elev, nproma,
- nlev, nblks_e, nblks_c,
- lacc);
-}
-
-/////////////////////////////////////////////
-
-void cells2verts_scalar_lib_dp(
- const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const double* coeff_int, double* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async)
-{
- cells2verts_scalar_lib<double,double>(
- p_cell_in, vert_cell_idx, vert_cell_blk,
- coeff_int, p_vert_out, i_startblk, i_endblk,
- i_startidx_in, i_endidx_in,
- slev, elev, nproma,
- nlev, nblks_c, nblks_v,
- lacc, acc_async);
-}
-
-void cells2verts_scalar_lib_dp2sp(
- const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async)
-{
- cells2verts_scalar_lib<double,float>(
- p_cell_in, vert_cell_idx, vert_cell_blk,
- coeff_int, p_vert_out, i_startblk, i_endblk,
- i_startidx_in, i_endidx_in,
- slev, elev, nproma,
- nlev, nblks_c, nblks_v,
- lacc, acc_async);
-}
-
-void cells2verts_scalar_lib_sp(
- const float* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async)
-{
- cells2verts_scalar_lib<float,float>(
- p_cell_in, vert_cell_idx, vert_cell_blk,
- coeff_int, p_vert_out, i_startblk, i_endblk,
- i_startidx_in, i_endidx_in,
- slev, elev, nproma,
- nlev, nblks_c, nblks_v,
- lacc, acc_async);
-}
-
-/////////////////////////////////////////////
-
-void cells2verts_scalar_ri_lib_dp(
- const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const double* coeff_int, double* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async)
-{
- cells2verts_scalar_ri_lib<double,double>(
- p_cell_in, vert_cell_idx, vert_cell_blk,
- coeff_int, p_vert_out, i_startblk, i_endblk,
- i_startidx_in, i_endidx_in,
- slev, elev, nproma,
- nlev, nblks_c, nblks_v,
- lacc, acc_async);
-}
-
-void cells2verts_scalar_ri_lib_dp2sp(
- const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const double* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async)
-{
- cells2verts_scalar_ri_lib<double,float>(
- p_cell_in, vert_cell_idx, vert_cell_blk,
- coeff_int, p_vert_out, i_startblk, i_endblk,
- i_startidx_in, i_endidx_in,
- slev, elev, nproma,
- nlev, nblks_c, nblks_v,
- lacc, acc_async);
-}
-
-void cells2verts_scalar_ri_lib_sp(
- const float* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async)
-{
- cells2verts_scalar_ri_lib<float,float>(
- p_cell_in, vert_cell_idx, vert_cell_blk,
- coeff_int, p_vert_out, i_startblk, i_endblk,
- i_startidx_in, i_endidx_in,
- slev, elev, nproma,
- nlev, nblks_c, nblks_v,
- lacc, acc_async);
-}
-
-/////////////////////////////////////////////
-
-void verts2cells_scalar_lib_dp(
- const double* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
- const double* coeff_int, double* p_cell_out, const int nblks_c, const int npromz_c,
- const int slev, const int elev, const int nproma, const int nlev,
- const int nblks_v, const bool lacc)
-{
- verts2cells_scalar_lib<double>(
- p_vert_in, cell_index_idx, cell_vertex_blk,
- coeff_int, p_cell_out, nblks_c, npromz_c,
- slev, elev, nproma, nlev,
- nblks_v, lacc);
-}
-
-void verts2cells_scalar_lib_sp(
- const float* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
- const float* coeff_int, float* p_cell_out, const int nblks_c, const int npromz_c,
- const int slev, const int elev, const int nproma, const int nlev,
- const int nblks_v, const bool lacc)
-{
- verts2cells_scalar_lib<float>(
- p_vert_in, cell_index_idx, cell_vertex_blk,
- coeff_int, p_cell_out, nblks_c, npromz_c,
- slev, elev, nproma, nlev,
- nblks_v, lacc);
-}
-
-/////////////////////////////////////////////
-
-void cell_avg_lib_dp(
- const double* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
- const double* avg_coeff, double* avg_psi_c, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c,
- const bool lacc)
-{
- cell_avg_lib<double>(
- psi_c, cell_neighbor_idx, cell_neighbor_blk,
- avg_coeff, avg_psi_c, i_startblk, i_endblk,
- i_startidx_in, i_endidx_in,
- slev, elev, nproma,
- nlev, nblks_c,
- lacc);
-}
-
-void cell_avg_lib_sp(
- const float* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
- const float* avg_coeff, float* avg_psi_c, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c,
- const bool lacc)
-{
- cell_avg_lib<float>(
- psi_c, cell_neighbor_idx, cell_neighbor_blk,
- avg_coeff, avg_psi_c, i_startblk, i_endblk,
- i_startidx_in, i_endidx_in,
- slev, elev, nproma,
- nlev, nblks_c,
- lacc);
-}
-
\ No newline at end of file
diff --git a/src/interpolation/mo_lib_interpolation_scalar_bindings_Dylan.h b/src/interpolation/mo_lib_interpolation_scalar_bindings_Dylan.h
deleted file mode 100644
index 4cb399b..0000000
--- a/src/interpolation/mo_lib_interpolation_scalar_bindings_Dylan.h
+++ /dev/null
@@ -1,109 +0,0 @@
-// ICON
-//
-// ---------------------------------------------------------------
-// Copyright (C) 2004-2024, DWD, MPI-M, DKRZ, KIT, ETH, MeteoSwiss
-// Contact information: icon-model.org
-//
-// See AUTHORS.TXT for a list of authors
-// See LICENSES/ for license information
-// SPDX-License-Identifier: BSD-3-Clause
-// ---------------------------------------------------------------
-
-#pragma once
-
-void edges2cells_scalar_lib_dp(
- const double* p_edge_in, const int* edge_idx, const int* edge_blk,
- const double* coeff_int, double* p_cell_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_e, const int nblks_c,
- const bool lacc);
-void edges2cells_scalar_lib_sp(
- const float* p_edge_in, const int* edge_idx, const int* edge_blk,
- const float* coeff_int, float* p_cell_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_e, const int nblks_c,
- const bool lacc);
-
-/////////////////////////////////////////////
-
-void cells2verts_scalar_lib_dp(
- const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const double* coeff_int, double* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async);
-void cells2verts_scalar_lib_dp2sp(
- const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async);
-void cells2verts_scalar_lib_sp(
- const float* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async);
-
-/////////////////////////////////////////////
-
-void cells2verts_scalar_ri_lib_dp(
- const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const double* coeff_int, double* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async);
-
-void cells2verts_scalar_ri_lib_dp2sp(
- const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const double* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async);
-
-void cells2verts_scalar_ri_lib_sp(
- const float* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async);
-
-/////////////////////////////////////////////
-
-void verts2cells_scalar_lib_dp(
- const double* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
- const double* coeff_int, double* p_cell_out, const int nblks_c, const int npromz_c,
- const int slev, const int elev, const int nproma, const int nlev,
- const int nblks_v, const bool lacc);
-
-void verts2cells_scalar_lib_sp(
- const float* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
- const float* coeff_int, float* p_cell_out, const int nblks_c, const int npromz_c,
- const int slev, const int elev, const int nproma, const int nlev,
- const int nblks_v, const bool lacc);
-
-/////////////////////////////////////////////
-
-void cell_avg_lib_dp(
- const double* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
- const double* avg_coeff, double* avg_psi_c, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c,
- const bool lacc);
-void cell_avg_lib_sp(
- const float* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
- const float* avg_coeff, float* avg_psi_c, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c,
- const bool lacc);
-
\ No newline at end of file
--
GitLab
From d9de153386a5b2dcce632c086c84a987c2c1fe66 Mon Sep 17 00:00:00 2001
From: Dylan Kierans <kierans@dkrz.de>
Date: Wed, 26 Feb 2025 08:57:36 +0100
Subject: [PATCH 11/36] Fix CMakeLists
---
src/interpolation/CMakeLists.txt | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/src/interpolation/CMakeLists.txt b/src/interpolation/CMakeLists.txt
index eade738..1051516 100644
--- a/src/interpolation/CMakeLists.txt
+++ b/src/interpolation/CMakeLists.txt
@@ -12,12 +12,12 @@
add_library(
iconmath-interpolation
mo_lib_interpolation_scalar.F90
+ mo_lib_interpolation_scalar.cpp
mo_lib_interpolation_vector.F90
mo_lib_interpolation_vector.cpp
mo_lib_intp_rbf.F90
mo_lib_intp_rbf-rbf_vec_interpol_vertex_lib.cpp
mo_lib_intp_rbf-rbf_vec_interpol_vertex_lib_bindings.cpp
- interpolation_bindings.h
interpolation_bindings.cpp
)
--
GitLab
From fe2f4edf159cd96d3dd6d2b0bb2ba3b64cb48641 Mon Sep 17 00:00:00 2001
From: Dylan Kierans <kierans@dkrz.de>
Date: Wed, 26 Feb 2025 08:58:15 +0100
Subject: [PATCH 12/36] Reformat mo_lib_interpolation_scalar, and fix end
indexes with +1 in loops
---
.../mo_lib_interpolation_scalar.cpp | 366 ++++++++++--------
1 file changed, 199 insertions(+), 167 deletions(-)
diff --git a/src/interpolation/mo_lib_interpolation_scalar.cpp b/src/interpolation/mo_lib_interpolation_scalar.cpp
index 7f91246..e5369f6 100644
--- a/src/interpolation/mo_lib_interpolation_scalar.cpp
+++ b/src/interpolation/mo_lib_interpolation_scalar.cpp
@@ -14,6 +14,19 @@
#include "mo_lib_loopindices.hpp"
#include "mo_lib_interpolation_scalar.hpp"
+//-----------------------------------------------------------------------
+//
+// ! averaging and interpolation routines and
+// ! routines needed to compute the coefficients therein
+//
+//-----------------------------------------------------------------------
+
+//-----------------------------------------------------------------------
+//>
+/// Performs average of scalar fields from vertices to velocity points.
+///
+/// The coefficients are given by coeff_int.
+///
template <typename T>
void verts2edges_scalar_lib(const T* p_vertex_in,
const int* edge_vertex_idx, const int* edge_vertex_blk,
@@ -34,7 +47,7 @@ void verts2edges_scalar_lib(const T* p_vertex_in,
UnmanagedConstT3D coeff_int_view (coeff_int, nproma,2,nblks_e);
UnmanagedT3D p_edge_out_view (p_edge_out, nproma,nlev,nblks_e);
- for (int jb = i_startblk; jb < i_endblk; ++jb) {
+ for (int jb = i_startblk; jb < i_endblk + 1; ++jb) {
int i_startidx, i_endidx;
get_indices_c_lib(i_startidx_in, i_endidx_in, nproma,
@@ -42,7 +55,7 @@ void verts2edges_scalar_lib(const T* p_vertex_in,
i_startidx, i_endidx);
Kokkos::parallel_for("verts2edges_scalar", Kokkos::MDRangePolicy<Kokkos::Rank<2>>
- ({slev, i_startidx}, {elev, i_endidx}),
+ ({slev, i_startidx}, {elev + 1, i_endidx + 1}),
KOKKOS_LAMBDA(const int jk, const int je) {
p_edge_out_view(je, jk, jb) = coeff_int_view(je, 0, jb) *
@@ -54,6 +67,13 @@ void verts2edges_scalar_lib(const T* p_vertex_in,
}
}
+//------------------------------------------------------------------------
+//>
+/// Computes average of scalar fields from centers of triangular faces to.
+///
+/// Computes average of scalar fields from centers of triangular faces to
+/// velocity points.
+///
template <typename T, typename S>
void cells2edges_scalar_lib(const S* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
const T* coeff_int, T* p_edge_out, const int* i_startblk_in,
@@ -80,14 +100,14 @@ void cells2edges_scalar_lib(const S* p_cell_in, const int* edge_cell_idx, const
i_startblk = i_startblk_in[0];
i_endblk = i_endblk_in[0];
- for (int jb = i_startblk; jb < i_endblk; ++jb) {
+ for (int jb = i_startblk; jb < i_endblk + 1; ++jb) {
int i_startidx, i_endidx;
get_indices_e_lib(i_startidx_in[0], i_endidx_in[0], nproma, jb,
i_startblk, i_endblk, i_startidx, i_endidx);
Kokkos::parallel_for("cells2edges_scalar", Kokkos::MDRangePolicy<Kokkos::Rank<2>>
- ({slev, i_startidx}, {elev, i_endidx}),
+ ({slev, i_startidx}, {elev + 1, i_endidx + 1}),
KOKKOS_LAMBDA(const int jk, const int je) {
if (iidx_view(je, jb, 0) >= 0 && iblk_view(je, jb, 0) >= 0){
@@ -109,14 +129,14 @@ void cells2edges_scalar_lib(const S* p_cell_in, const int* edge_cell_idx, const
i_startblk = i_startblk_in[1];
i_endblk = i_endblk_in[1];
- for (int jb = i_startblk; jb < i_endblk; ++jb) {
+ for (int jb = i_startblk; jb < i_endblk + 1; ++jb) {
int i_startidx, i_endidx;
get_indices_e_lib(i_startidx_in[1], i_endidx_in[1], nproma, jb,
i_startblk, i_endblk, i_startidx, i_endidx);
Kokkos::parallel_for("cells2edges_scalar", Kokkos::MDRangePolicy<Kokkos::Rank<2>>
- ({slev, i_startidx}, {elev, i_endidx}),
+ ({slev, i_startidx}, {elev + 1, i_endidx + 1}),
KOKKOS_LAMBDA(const int jk, const int je) {
p_edge_out_view(je, jk, jb) = coeff_int_view(je, 0, jb) *
@@ -130,6 +150,13 @@ void cells2edges_scalar_lib(const S* p_cell_in, const int* edge_cell_idx, const
}
+//------------------------------------------------------------------------
+//>
+/// Computes average of scalar fields from velocity points to.
+///
+/// Computes average of scalar fields from velocity points to
+/// centers of dual faces.
+///
template <typename T>
void edges2verts_scalar_lib (const T* p_edge_in, const int* vert_edge_idx, const int* vert_edge_blk,
const T* v_int, T* p_vert_out, const int i_startblk, const int i_endblk,
@@ -148,7 +175,7 @@ void edges2verts_scalar_lib (const T* p_edge_in, const int* vert_edge_idx, const
UnmanagedConstT3D v_int_view (v_int, nproma,cell_type,nblks_v);
UnmanagedT3D p_vert_out_view (p_vert_out, nproma,nlev,nblks_v);
- for (int jb = i_startblk; jb < i_endblk; ++jb) {
+ for (int jb = i_startblk; jb < i_endblk + 1; ++jb) {
int i_startidx, i_endidx;
get_indices_v_lib(i_startidx_in, i_endidx_in, nproma,
@@ -156,7 +183,7 @@ void edges2verts_scalar_lib (const T* p_edge_in, const int* vert_edge_idx, const
i_startidx, i_endidx);
Kokkos::parallel_for("edges2verts_scalar", Kokkos::MDRangePolicy<Kokkos::Rank<2>>
- ({slev, i_startidx}, {elev, i_endidx}),
+ ({slev, i_startidx}, {elev + 1, i_endidx + 1}),
KOKKOS_LAMBDA(const int jk, const int jv) {
p_vert_out_view(jv, jk, jb) = v_int_view(jv, 0, jb)*p_edge_in_view(iidx_view(jv, jb, 0), jk, iblk_view(jv, jb, 0))
@@ -171,66 +198,7 @@ void edges2verts_scalar_lib (const T* p_edge_in, const int* vert_edge_idx, const
}
-template
-void verts2edges_scalar_lib<double>(const double* p_vertex_in,
- const int* edge_vertex_idx, const int* edge_vertex_blk,
- const double* coeff_int,
- double* p_edge_out,
- const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
- const int nproma, const int nlev, const int nblks_v, const int nblks_e, const bool lacc);
-
-template
-void verts2edges_scalar_lib<float>(const float* p_vertex_in,
- const int* edge_vertex_idx, const int* edge_vertex_blk,
- const float* coeff_int,
- float* p_edge_out,
- const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
- const int nproma, const int nlev, const int nblks_v, const int nblks_e, const bool lacc);
-
-template
-void cells2edges_scalar_lib<double, double>(const double* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
- const double* coeff_int, double* p_edge_out, const int* i_startblk_in,
- const int* i_endblk_in, const int* i_startidx_in, const int* i_endidx_in,
- const int slev, const int elev, const int nproma, const int nlev,
- const int nblk_c, const int nblks_e, const int patch_id,
- const bool l_limited_area, const bool lfill_latbc, const bool lacc);
-
-template
-void cells2edges_scalar_lib<float, float>(const float* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
- const float* coeff_int, float* p_edge_out, const int* i_startblk_in,
- const int* i_endblk_in, const int* i_startidx_in, const int* i_endidx_in,
- const int slev, const int elev, const int nproma, const int nlev,
- const int nblk_c, const int nblks_e, const int patch_id,
- const bool l_limited_area, const bool lfill_latbc, const bool lacc);
-
-template
-void cells2edges_scalar_lib<double, float>(const float* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
- const double* coeff_int, double* p_edge_out, const int* i_startblk_in,
- const int* i_endblk_in, const int* i_startidx_in, const int* i_endidx_in,
- const int slev, const int elev, const int nproma, const int nlev,
- const int nblk_c, const int nblks_e, const int patch_id,
- const bool l_limited_area, const bool lfill_latbc, const bool lacc);
-
-
-template
-void edges2verts_scalar_lib<double> (const double* p_edge_in, const int* vert_edge_idx, const int* vert_edge_blk,
- const double* v_int, double* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in, const int slev,
- const int elev, const int nproma, const int nlev, const int nblks_e,
- const int nblks_v, const int cell_type, const bool lacc);
-
-template
-void edges2verts_scalar_lib <float> (const float* p_edge_in, const int* vert_edge_idx, const int* vert_edge_blk,
- const float* v_int, float* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in, const int slev,
- const int elev, const int nproma, const int nlev, const int nblks_e,
- const int nblks_v, const int cell_type, const bool lacc);
-
-
//------------------------------------------------------------------------
-//
//>
/// Computes interpolation from edges to cells
///
@@ -268,7 +236,7 @@ void edges2cells_scalar_lib(
int i_startidx, i_endidx;
- for (int jb = i_startblk; jb < i_endblk; ++jb){
+ for (int jb = i_startblk; jb < i_endblk + 1; ++jb){
get_indices_c_lib(i_startidx_in, i_endidx_in, nproma, jb, i_startblk, i_endblk,
i_startidx, i_endidx);
@@ -286,28 +254,10 @@ void edges2cells_scalar_lib(
}
-template
-void edges2cells_scalar_lib<double>(
- const double* p_edge_in, const int* edge_idx, const int* edge_blk,
- const double* coeff_int, double* p_cell_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_e, const int nblks_c,
- const bool lacc);
-
-template
-void edges2cells_scalar_lib<float>(
- const float* p_edge_in, const int* edge_idx, const int* edge_blk,
- const float* coeff_int, float* p_cell_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_e, const int nblks_c,
- const bool lacc);
-
//------------------------------------------------------------------------
//>
-//// Computes average of scalar fields from centers of cells to vertices.
-////
+/// Computes average of scalar fields from centers of cells to vertices.
+///
template <typename T, typename S>
void cells2verts_scalar_lib(
const T* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
@@ -340,7 +290,7 @@ void cells2verts_scalar_lib(
int i_startidx, i_endidx;
- for (int jb = i_startblk; jb < i_endblk; ++jb){
+ for (int jb = i_startblk; jb < i_endblk + 1; ++jb){
get_indices_v_lib(i_startidx_in, i_endidx_in, nproma, jb, i_startblk, i_endblk,
i_startidx, i_endidx);
@@ -363,36 +313,7 @@ void cells2verts_scalar_lib(
}
}
-template
-void cells2verts_scalar_lib<double,double>(
- const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const double* coeff_int, double* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async);
-
-template
-void cells2verts_scalar_lib<double,float>(
- const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async);
-
-template
-void cells2verts_scalar_lib<float,float>(
- const float* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async);
-
-//------------------------------------------------------------------------
-
-
+//-------------------------------------------------------------------------
//>
/// Same as above, but provides output optionally in single precision and
/// assumes reversed index order of the output field in loop exchange mode
@@ -428,7 +349,7 @@ void cells2verts_scalar_ri_lib(
int i_startidx, i_endidx;
- for (int jb = i_startblk; jb < i_endblk; ++jb){
+ for (int jb = i_startblk; jb < i_endblk + 1; ++jb){
get_indices_v_lib(i_startidx_in, i_endidx_in, nproma, jb, i_startblk, i_endblk,
i_startidx, i_endidx);
@@ -452,34 +373,7 @@ void cells2verts_scalar_ri_lib(
}
-template
-void cells2verts_scalar_ri_lib<double,double>(
- const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const double* coeff_int, double* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async);
-
-template
-void cells2verts_scalar_ri_lib<double,float>(
- const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const double* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async);
-
-template
-void cells2verts_scalar_ri_lib<float,float>(
- const float* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async);
-
-
+//-------------------------------------------------------------------------
//>
/// Computes average of scalar fields from vertices to centers of cells.
///
@@ -510,7 +404,7 @@ void verts2cells_scalar_lib(
// vertex based scalar output field, dim: (nproma,nlev,nblks_c)
UnmanagedT3D p_cell_out_view(p_cell_out, nproma, nlev, nblks_c);
- for (int jb = 1; jb<nblks_c; ++jb){
+ for (int jb = 1; jb<nblks_c + 1; ++jb){
int nlen = (jb != nblks_c ? nproma: npromz_c);
@@ -529,24 +423,8 @@ void verts2cells_scalar_lib(
}
}
-template
-void verts2cells_scalar_lib<double>(
- const double* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
- const double* coeff_int, double* p_cell_out, const int nblks_c, const int npromz_c,
- const int slev, const int elev, const int nproma, const int nlev,
- const int nblks_v, const bool lacc);
-
-template
-void verts2cells_scalar_lib<float>(
- const float* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
- const float* coeff_int, float* p_cell_out, const int nblks_c, const int npromz_c,
- const int slev, const int elev, const int nproma, const int nlev,
- const int nblks_v, const bool lacc);
-
//-------------------------------------------------------------------------
-//
-//
//>
/// Computes the average of a cell-based variable.
///
@@ -557,7 +435,6 @@ void verts2cells_scalar_lib<float>(
/// input: lives on centers of triangles
/// output: lives on centers of triangles
///
-
template <typename T>
void cell_avg_lib(
const T* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
@@ -585,7 +462,7 @@ void cell_avg_lib(
int i_startidx, i_endidx;
- for (int jb = i_startblk; jb<i_endblk; ++jb){
+ for (int jb = i_startblk; jb<i_endblk + 1; ++jb){
get_indices_c_lib(i_startidx_in, i_endidx_in, nproma, jb, i_startblk, i_endblk,
i_startidx, i_endidx);
@@ -605,6 +482,161 @@ void cell_avg_lib(
}
}
+//-----------------------------------------------------------------------
+//
+// Explicit Instantiations
+//
+//-----------------------------------------------------------------------
+
+template
+void verts2edges_scalar_lib<double>(const double* p_vertex_in,
+ const int* edge_vertex_idx, const int* edge_vertex_blk,
+ const double* coeff_int,
+ double* p_edge_out,
+ const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
+ const int nproma, const int nlev, const int nblks_v, const int nblks_e, const bool lacc);
+
+template
+void verts2edges_scalar_lib<float>(const float* p_vertex_in,
+ const int* edge_vertex_idx, const int* edge_vertex_blk,
+ const float* coeff_int,
+ float* p_edge_out,
+ const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
+ const int nproma, const int nlev, const int nblks_v, const int nblks_e, const bool lacc);
+
+template
+void cells2edges_scalar_lib<double, double>(const double* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
+ const double* coeff_int, double* p_edge_out, const int* i_startblk_in,
+ const int* i_endblk_in, const int* i_startidx_in, const int* i_endidx_in,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblk_c, const int nblks_e, const int patch_id,
+ const bool l_limited_area, const bool lfill_latbc, const bool lacc);
+
+template
+void cells2edges_scalar_lib<float, float>(const float* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
+ const float* coeff_int, float* p_edge_out, const int* i_startblk_in,
+ const int* i_endblk_in, const int* i_startidx_in, const int* i_endidx_in,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblk_c, const int nblks_e, const int patch_id,
+ const bool l_limited_area, const bool lfill_latbc, const bool lacc);
+
+template
+void cells2edges_scalar_lib<double, float>(const float* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
+ const double* coeff_int, double* p_edge_out, const int* i_startblk_in,
+ const int* i_endblk_in, const int* i_startidx_in, const int* i_endidx_in,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblk_c, const int nblks_e, const int patch_id,
+ const bool l_limited_area, const bool lfill_latbc, const bool lacc);
+
+
+template
+void edges2verts_scalar_lib<double> (const double* p_edge_in, const int* vert_edge_idx, const int* vert_edge_blk,
+ const double* v_int, double* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in, const int slev,
+ const int elev, const int nproma, const int nlev, const int nblks_e,
+ const int nblks_v, const int cell_type, const bool lacc);
+
+template
+void edges2verts_scalar_lib <float> (const float* p_edge_in, const int* vert_edge_idx, const int* vert_edge_blk,
+ const float* v_int, float* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in, const int slev,
+ const int elev, const int nproma, const int nlev, const int nblks_e,
+ const int nblks_v, const int cell_type, const bool lacc);
+
+
+template
+void edges2cells_scalar_lib<double>(
+ const double* p_edge_in, const int* edge_idx, const int* edge_blk,
+ const double* coeff_int, double* p_cell_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_e, const int nblks_c,
+ const bool lacc);
+
+template
+void edges2cells_scalar_lib<float>(
+ const float* p_edge_in, const int* edge_idx, const int* edge_blk,
+ const float* coeff_int, float* p_cell_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_e, const int nblks_c,
+ const bool lacc);
+
+
+template
+void cells2verts_scalar_lib<double,double>(
+ const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const double* coeff_int, double* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async);
+
+template
+void cells2verts_scalar_lib<double,float>(
+ const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async);
+
+template
+void cells2verts_scalar_lib<float,float>(
+ const float* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async);
+
+
+template
+void cells2verts_scalar_ri_lib<double,double>(
+ const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const double* coeff_int, double* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async);
+
+template
+void cells2verts_scalar_ri_lib<double,float>(
+ const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const double* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async);
+
+template
+void cells2verts_scalar_ri_lib<float,float>(
+ const float* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async);
+
+
+
+template
+void verts2cells_scalar_lib<double>(
+ const double* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
+ const double* coeff_int, double* p_cell_out, const int nblks_c, const int npromz_c,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblks_v, const bool lacc);
+
+template
+void verts2cells_scalar_lib<float>(
+ const float* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
+ const float* coeff_int, float* p_cell_out, const int nblks_c, const int npromz_c,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblks_v, const bool lacc);
+
+
template
void cell_avg_lib<double>(
const double* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
--
GitLab
From b8a1df4691d058c3a9ecfc66c497c52b70c14464 Mon Sep 17 00:00:00 2001
From: Dylan Kierans <kierans@dkrz.de>
Date: Wed, 26 Feb 2025 08:58:45 +0100
Subject: [PATCH 13/36] Use +1 for end index in
mo_lib_intp_rbf-rbf_vec_interpol_vertex_lib
---
.../mo_lib_intp_rbf-rbf_vec_interpol_vertex_lib.cpp | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/src/interpolation/mo_lib_intp_rbf-rbf_vec_interpol_vertex_lib.cpp b/src/interpolation/mo_lib_intp_rbf-rbf_vec_interpol_vertex_lib.cpp
index c9b776e..0a41fe8 100644
--- a/src/interpolation/mo_lib_intp_rbf-rbf_vec_interpol_vertex_lib.cpp
+++ b/src/interpolation/mo_lib_intp_rbf-rbf_vec_interpol_vertex_lib.cpp
@@ -101,7 +101,7 @@ void rbf_vec_interpol_vertex_lib(
int i_startidx; // start index
int i_endidx; // end index
- for (jb=i_startblk; jb <= i_endblk; ++jb){
+ for (jb=i_startblk; jb < i_endblk + 1; ++jb){
get_indices_v_lib(i_startidx_in, i_endidx_in, nproma, jb, i_startblk, i_endblk,
i_startidx, i_endidx);
--
GitLab
From 90989fd088dee68a5a5452cd488b6ddcefade465 Mon Sep 17 00:00:00 2001
From: Dylan Kierans <kierans@dkrz.de>
Date: Wed, 26 Feb 2025 09:13:18 +0100
Subject: [PATCH 14/36] Adding first test for
mo_lib_interpolation_scalar::verts2edges_scalar_lib
---
test/c/CMakeLists.txt | 1 +
test/c/test_interpolation_scalar.cpp | 138 +++++++++++++++++++++++++++
2 files changed, 139 insertions(+)
create mode 100644 test/c/test_interpolation_scalar.cpp
diff --git a/test/c/CMakeLists.txt b/test/c/CMakeLists.txt
index 13c5dfe..16371d0 100644
--- a/test/c/CMakeLists.txt
+++ b/test/c/CMakeLists.txt
@@ -26,6 +26,7 @@ set(SOURCES
test_tdma_solver.cpp
test_interpolation_vector.cpp
test_intp_rbf.cpp
+ test_interpolation_scalar.cpp
)
# Create the test executable from your test files, including main.cpp.
add_executable(iconmath_test_c ${SOURCES})
diff --git a/test/c/test_interpolation_scalar.cpp b/test/c/test_interpolation_scalar.cpp
new file mode 100644
index 0000000..efffaef
--- /dev/null
+++ b/test/c/test_interpolation_scalar.cpp
@@ -0,0 +1,138 @@
+// ICON
+//
+// ---------------------------------------------------------------
+// Copyright (C) 2004-2024, DWD, MPI-M, DKRZ, KIT, ETH, MeteoSwiss
+// Contact information: icon-model.org
+//
+// See AUTHORS.TXT for a list of authors
+// See LICENSES/ for license information
+// SPDX-License-Identifier: BSD-3-Clause
+// ---------------------------------------------------------------
+
+#include <gtest/gtest.h>
+#include <Kokkos_Core.hpp>
+#include <vector>
+#include "mo_lib_interpolation_scalar.hpp"
+
+// Free-function helpers for 3D and 4D array sizes (assumed column-major)
+template<typename T>
+size_t num_elements_3d(int d1, int d2, int d3) {
+ return static_cast<size_t>(d1) * d2 * d3;
+}
+
+template<typename T>
+size_t num_elements_4d(int d1, int d2, int d3, int d4) {
+ return static_cast<size_t>(d1) * d2 * d3 * d4;
+}
+
+// Define a helper struct that holds the two types.
+template<typename InT, typename OutT>
+struct MixedPrecision {
+ using in_type = InT;
+ using out_type = OutT;
+};
+
+// Define a helper struct that holds the one type.
+template<typename T>
+struct FullPrecision {
+ using in_type = T;
+};
+
+// Define the list of type pairs we want to test.
+typedef ::testing::Types< MixedPrecision<double, double>,
+ MixedPrecision<double, float>,
+ MixedPrecision<float, float> > MixedTypes;
+
+// Define the list of type pairs we want to test.
+typedef ::testing::Types< FullPrecision<double>,
+ FullPrecision<float > > FullTypes;
+
+
+// Define a typed test fixture.
+template <typename TypePair>
+class Verts2edgesScalarLibFullTestFixture : public ::testing::Test {
+public:
+ using InType = typename TypePair::in_type;
+ using OutType = typename TypePair::in_type;
+
+ // Constant dimensions.
+ static constexpr int nproma = 16; // inner loop length
+ static constexpr int nlev = 7; // number of vertical levels
+ static constexpr int nblks_e = 2; // number of edge blocks (for p_e_in)
+ static constexpr int nblks_v = 2; // number of vertex blocks (for rbf arrays and outputs)
+ static constexpr int rbf_vec_dim = 6; // fixed dimension for rbf vector (stencil points)
+
+ // Parameter values.
+ int i_startblk = 0;
+ int i_endblk = 2; // Test blocks [0, 1]
+ int i_startidx_in = 2;
+ int i_endidx_in = nproma - 3; // Full range: 0 .. nproma-1
+ int slev = 1;
+ int elev = nlev - 1; // Full vertical range (0 .. nlev-1)
+ bool lacc = false; // Not using ACC-specific behavior.
+ bool acc_async = false; // No asynchronous execution.
+
+ // Arrays stored in std::vector.
+ std::vector<InType> p_vertex_in; // Dimensions: (nproma, nlev, nblks_v)
+ std::vector<int> edge_vertex_idx; // Dimensions: (nproma, nblks_e, 4)
+ std::vector<int> edge_vertex_blk; // Dimensions: (nproma, nblks_e, 4)
+ std::vector<InType> coeff_int; // Dimensions: (nproma, 2, nblks_e)
+ std::vector<InType> p_edge_out; // Dimensions: (nproma, nlev, nblks_e)
+
+ Verts2edgesScalarLibFullTestFixture() {
+ // Allocate and initialize inputs.
+ p_vertex_in.resize(num_elements_3d<InType>(nproma, nlev, nblks_v), static_cast<InType>(1));
+ edge_vertex_idx.resize(num_elements_3d<int>(nproma, nblks_e, 4), 1);
+ edge_vertex_blk.resize(num_elements_3d<int>(nproma, nblks_e, 4), 0);
+ coeff_int.resize(num_elements_3d<InType>(nproma, 2, nblks_v), static_cast<InType>(1));
+
+ // Allocate output arrays and initialize to zero.
+ p_edge_out.resize(num_elements_3d<OutType>(nproma, nlev, nblks_e), static_cast<OutType>(0));
+ }
+};
+
+TYPED_TEST_SUITE(Verts2edgesScalarLibFullTestFixture, FullTypes);
+
+TYPED_TEST(Verts2edgesScalarLibFullTestFixture, BasicTest) {
+ using InType = typename TestFixture::InType;
+ using OutType = typename TestFixture::InType;
+
+ // Copy output so it can be restored after
+ auto tmp_p_edge_out = this->p_edge_out;
+
+ // Call the function
+ verts2edges_scalar_lib<InType>(
+ this->p_vertex_in.data(),
+ this->edge_vertex_idx.data(),
+ this->edge_vertex_blk.data(),
+ this->coeff_int.data(),
+ this->p_edge_out.data(),
+ this->i_startblk,
+ this->i_endblk,
+ this->i_startidx_in,
+ this->i_endidx_in,
+ this->slev,
+ this->elev,
+ this->nproma,
+ this->nlev,
+ Verts2edgesScalarLibFullTestFixture< TypeParam >::nblks_v,
+ Verts2edgesScalarLibFullTestFixture< TypeParam >::nblks_e,
+ this->lacc);
+
+ // Check the outputs only for blocks in the range
+ // { [i_startblk, i_endblk], [slev,elev], [i_startidx_in, i_endidx_in] }
+ for (int block = this->i_startblk; block <= this->i_endblk; ++block) {
+ for (int level = this->slev; level < this->elev; ++level) {
+ for (int i = this->i_startidx_in; i < this->i_endidx_in; ++i) {
+ // Compute the linear index for a 3D array in column-major order:
+ size_t idx = i + level * this->nproma + block * this->nproma * this->nlev;
+ // Since every contribution is 1 and there are 6 stencil points, expect 6.
+ EXPECT_NEAR(this->p_edge_out[idx], static_cast<OutType>(2), static_cast<OutType>(1e-5))
+ << "Failure at block " << block << ", level " << level << ", index " << i;
+ }
+ }
+ }
+
+ // Restore output
+ this->p_edge_out = tmp_p_edge_out;
+}
--
GitLab
From ca07f9bf124b0ef0c557e084aa7c366601a6d751 Mon Sep 17 00:00:00 2001
From: Dylan Kierans <kierans@dkrz.de>
Date: Wed, 26 Feb 2025 09:15:56 +0100
Subject: [PATCH 15/36] Restoring proper bounds for test
---
test/c/test_interpolation_scalar.cpp | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/test/c/test_interpolation_scalar.cpp b/test/c/test_interpolation_scalar.cpp
index efffaef..9a47f76 100644
--- a/test/c/test_interpolation_scalar.cpp
+++ b/test/c/test_interpolation_scalar.cpp
@@ -64,11 +64,11 @@ public:
// Parameter values.
int i_startblk = 0;
- int i_endblk = 2; // Test blocks [0, 1]
+ int i_endblk = 1; // Test blocks [0, 1]
int i_startidx_in = 2;
- int i_endidx_in = nproma - 3; // Full range: 0 .. nproma-1
+ int i_endidx_in = nproma - 3; // Partial range: 2 .. nproma-3
int slev = 1;
- int elev = nlev - 1; // Full vertical range (0 .. nlev-1)
+ int elev = nlev - 1; // Partial vertical range (1 .. nlev-1)
bool lacc = false; // Not using ACC-specific behavior.
bool acc_async = false; // No asynchronous execution.
--
GitLab
From 3be29336ca0b63e837dbbd1b1f0d52940e44314d Mon Sep 17 00:00:00 2001
From: Dylan Kierans <kierans@dkrz.de>
Date: Wed, 26 Feb 2025 15:12:49 +0100
Subject: [PATCH 16/36] Formatting
---
.../mo_lib_interpolation_scalar.cpp | 10 ++--
.../mo_lib_interpolation_scalar.hpp | 59 +++++++++----------
2 files changed, 32 insertions(+), 37 deletions(-)
diff --git a/src/interpolation/mo_lib_interpolation_scalar.cpp b/src/interpolation/mo_lib_interpolation_scalar.cpp
index e5369f6..a9b3f46 100644
--- a/src/interpolation/mo_lib_interpolation_scalar.cpp
+++ b/src/interpolation/mo_lib_interpolation_scalar.cpp
@@ -187,11 +187,11 @@ void edges2verts_scalar_lib (const T* p_edge_in, const int* vert_edge_idx, const
KOKKOS_LAMBDA(const int jk, const int jv) {
p_vert_out_view(jv, jk, jb) = v_int_view(jv, 0, jb)*p_edge_in_view(iidx_view(jv, jb, 0), jk, iblk_view(jv, jb, 0))
- + v_int_view(jv, 1, jb)*p_edge_in_view(iidx_view(jv, jb, 1), jk, iblk_view(jv, jb, 1))
- + v_int_view(jv, 2, jb)*p_edge_in_view(iidx_view(jv, jb, 2), jk, iblk_view(jv, jb, 2))
- + v_int_view(jv, 3, jb)*p_edge_in_view(iidx_view(jv, jb, 3), jk, iblk_view(jv, jb, 3))
- + v_int_view(jv, 4, jb)*p_edge_in_view(iidx_view(jv, jb, 4), jk, iblk_view(jv, jb, 4))
- + v_int_view(jv, 5, jb)*p_edge_in_view(iidx_view(jv, jb, 5), jk, iblk_view(jv, jb, 5));
+ + v_int_view(jv, 1, jb)*p_edge_in_view(iidx_view(jv, jb, 1), jk, iblk_view(jv, jb, 1))
+ + v_int_view(jv, 2, jb)*p_edge_in_view(iidx_view(jv, jb, 2), jk, iblk_view(jv, jb, 2))
+ + v_int_view(jv, 3, jb)*p_edge_in_view(iidx_view(jv, jb, 3), jk, iblk_view(jv, jb, 3))
+ + v_int_view(jv, 4, jb)*p_edge_in_view(iidx_view(jv, jb, 4), jk, iblk_view(jv, jb, 4))
+ + v_int_view(jv, 5, jb)*p_edge_in_view(iidx_view(jv, jb, 5), jk, iblk_view(jv, jb, 5));
});
Kokkos::fence();
}
diff --git a/src/interpolation/mo_lib_interpolation_scalar.hpp b/src/interpolation/mo_lib_interpolation_scalar.hpp
index d92cee0..95febf1 100644
--- a/src/interpolation/mo_lib_interpolation_scalar.hpp
+++ b/src/interpolation/mo_lib_interpolation_scalar.hpp
@@ -37,43 +37,38 @@ void edges2verts_scalar_lib (const T* p_edge_in, const int* vert_edge_idx, const
const int nblks_v, const int cell_type, const bool lacc);
template <typename T>
-void edges2cells_scalar_lib(
- const T* p_edge_in, const int* edge_idx, const int* edge_blk,
- const T* coeff_int, T* p_cell_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_e, const int nblks_c,
- const bool lacc);
+void edges2cells_scalar_lib(const T* p_edge_in, const int* edge_idx, const int* edge_blk,
+ const T* coeff_int, T* p_cell_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_e, const int nblks_c,
+ const bool lacc);
template <typename T, typename S>
-void cells2verts_scalar_lib(
- const T* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const S* coeff_int, S* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async);
+void cells2verts_scalar_lib(const T* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const S* coeff_int, S* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async);
template <typename T, typename S>
-void cells2verts_scalar_ri_lib(
- const T* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const T* coeff_int, S* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async);
+void cells2verts_scalar_ri_lib(const T* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const T* coeff_int, S* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async);
template <typename T>
-void verts2cells_scalar_lib(
- const T* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
- const T* coeff_int, T* p_cell_out, const int nblks_c, const int npromz_c,
- const int slev, const int elev, const int nproma, const int nlev,
- const int nblks_v, const bool lacc);
+void verts2cells_scalar_lib(const T* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
+ const T* coeff_int, T* p_cell_out, const int nblks_c, const int npromz_c,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblks_v, const bool lacc);
template <typename T>
-void cell_avg_lib(
- const T* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
- const T* avg_coeff, T* avg_psi_c, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const bool lacc);
\ No newline at end of file
+void cell_avg_lib(const T* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
+ const T* avg_coeff, T* avg_psi_c, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const bool lacc);
\ No newline at end of file
--
GitLab
From 011e68ae08b27c21d215a230e554a4ff30d35388 Mon Sep 17 00:00:00 2001
From: Dylan Kierans <kierans@dkrz.de>
Date: Wed, 26 Feb 2025 15:24:39 +0100
Subject: [PATCH 17/36] index from 0 with statics in edges2cells
---
src/interpolation/mo_lib_interpolation_scalar.cpp | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/src/interpolation/mo_lib_interpolation_scalar.cpp b/src/interpolation/mo_lib_interpolation_scalar.cpp
index a9b3f46..d5f3b8f 100644
--- a/src/interpolation/mo_lib_interpolation_scalar.cpp
+++ b/src/interpolation/mo_lib_interpolation_scalar.cpp
@@ -246,9 +246,9 @@ void edges2cells_scalar_lib(
Kokkos::parallel_for("edges2cells_scalar_lib_inner", innerPolicy,
KOKKOS_LAMBDA(const int jk, const int jc) {
p_cell_out_view(jc, jk, jb) =
+ coeff_int_view(jc, 0, jb)*p_edge_in_view(iidx_view(jc, jb, 0), jk, iblk_view(jc, jb, 0)) +
coeff_int_view(jc, 1, jb)*p_edge_in_view(iidx_view(jc, jb, 1), jk, iblk_view(jc, jb, 1)) +
- coeff_int_view(jc, 2, jb)*p_edge_in_view(iidx_view(jc, jb, 2), jk, iblk_view(jc, jb, 2)) +
- coeff_int_view(jc, 3, jb)*p_edge_in_view(iidx_view(jc, jb, 3), jk, iblk_view(jc, jb, 3));
+ coeff_int_view(jc, 2, jb)*p_edge_in_view(iidx_view(jc, jb, 2), jk, iblk_view(jc, jb, 2));
});
}
--
GitLab
From bf8338be6e3b5f6fd2f4f2ef515d9702bb2d1d49 Mon Sep 17 00:00:00 2001
From: Dylan Kierans <kierans@dkrz.de>
Date: Wed, 26 Feb 2025 15:24:59 +0100
Subject: [PATCH 18/36] Update tests, about halfway mark
---
test/c/test_interpolation_scalar.cpp | 332 +++++++++++++++++++++++----
1 file changed, 293 insertions(+), 39 deletions(-)
diff --git a/test/c/test_interpolation_scalar.cpp b/test/c/test_interpolation_scalar.cpp
index 9a47f76..78c615e 100644
--- a/test/c/test_interpolation_scalar.cpp
+++ b/test/c/test_interpolation_scalar.cpp
@@ -32,54 +32,58 @@ struct MixedPrecision {
using out_type = OutT;
};
-// Define a helper struct that holds the one type.
-template<typename T>
-struct FullPrecision {
- using in_type = T;
-};
-
// Define the list of type pairs we want to test.
typedef ::testing::Types< MixedPrecision<double, double>,
MixedPrecision<double, float>,
MixedPrecision<float, float> > MixedTypes;
// Define the list of type pairs we want to test.
-typedef ::testing::Types< FullPrecision<double>,
- FullPrecision<float > > FullTypes;
+typedef ::testing::Types< MixedPrecision<double, double>,
+ MixedPrecision<float, float > > OneTypes;
-// Define a typed test fixture.
-template <typename TypePair>
-class Verts2edgesScalarLibFullTestFixture : public ::testing::Test {
+// Shared dimensions for all routines and classes
+class interp_dimensions {
public:
- using InType = typename TypePair::in_type;
- using OutType = typename TypePair::in_type;
-
// Constant dimensions.
static constexpr int nproma = 16; // inner loop length
static constexpr int nlev = 7; // number of vertical levels
+ static constexpr int nblks_c = 2; // number of cell blocks
static constexpr int nblks_e = 2; // number of edge blocks (for p_e_in)
static constexpr int nblks_v = 2; // number of vertex blocks (for rbf arrays and outputs)
- static constexpr int rbf_vec_dim = 6; // fixed dimension for rbf vector (stencil points)
// Parameter values.
- int i_startblk = 0;
- int i_endblk = 1; // Test blocks [0, 1]
- int i_startidx_in = 2;
- int i_endidx_in = nproma - 3; // Partial range: 2 .. nproma-3
- int slev = 1;
- int elev = nlev - 1; // Partial vertical range (1 .. nlev-1)
- bool lacc = false; // Not using ACC-specific behavior.
- bool acc_async = false; // No asynchronous execution.
+ const int i_startblk = 0;
+ const int i_endblk = 1; // Test blocks [0, 1]
+ const int i_startidx = 2;
+ const int i_endidx = nproma - 3; // Partial range: 2 .. nproma-3
+ const int slev = 1;
+ const int elev = nlev - 1; // Partial vertical range (1 .. nlev-1)
+ const bool lacc = false; // Not using ACC-specific behavior.
+ const bool acc_async = false; // No asynchronous execution.
+};
+
+////////////////////////////////////////////////////////////////////////////////
+//
+// ! verts2edges
+//
+////////////////////////////////////////////////////////////////////////////////
+
+// Define a test fixture
+template <typename Types>
+class Verts2edgesScalarLibTestFixture : public testing::Test, public interp_dimensions {
+public:
+ using InType = typename Types::in_type;
+ using OutType = typename Types::out_type;
// Arrays stored in std::vector.
std::vector<InType> p_vertex_in; // Dimensions: (nproma, nlev, nblks_v)
std::vector<int> edge_vertex_idx; // Dimensions: (nproma, nblks_e, 4)
std::vector<int> edge_vertex_blk; // Dimensions: (nproma, nblks_e, 4)
std::vector<InType> coeff_int; // Dimensions: (nproma, 2, nblks_e)
- std::vector<InType> p_edge_out; // Dimensions: (nproma, nlev, nblks_e)
+ std::vector<OutType> p_edge_out; // Dimensions: (nproma, nlev, nblks_e)
- Verts2edgesScalarLibFullTestFixture() {
+ Verts2edgesScalarLibTestFixture() {
// Allocate and initialize inputs.
p_vertex_in.resize(num_elements_3d<InType>(nproma, nlev, nblks_v), static_cast<InType>(1));
edge_vertex_idx.resize(num_elements_3d<int>(nproma, nblks_e, 4), 1);
@@ -91,14 +95,13 @@ public:
}
};
-TYPED_TEST_SUITE(Verts2edgesScalarLibFullTestFixture, FullTypes);
+// Add test suite
+TYPED_TEST_SUITE(Verts2edgesScalarLibTestFixture, OneTypes);
-TYPED_TEST(Verts2edgesScalarLibFullTestFixture, BasicTest) {
+// Add test
+TYPED_TEST(Verts2edgesScalarLibTestFixture, verts2edges) {
using InType = typename TestFixture::InType;
- using OutType = typename TestFixture::InType;
-
- // Copy output so it can be restored after
- auto tmp_p_edge_out = this->p_edge_out;
+ using OutType = typename TestFixture::OutType;
// Call the function
verts2edges_scalar_lib<InType>(
@@ -109,30 +112,281 @@ TYPED_TEST(Verts2edgesScalarLibFullTestFixture, BasicTest) {
this->p_edge_out.data(),
this->i_startblk,
this->i_endblk,
- this->i_startidx_in,
- this->i_endidx_in,
+ this->i_startidx,
+ this->i_endidx,
this->slev,
this->elev,
this->nproma,
this->nlev,
- Verts2edgesScalarLibFullTestFixture< TypeParam >::nblks_v,
- Verts2edgesScalarLibFullTestFixture< TypeParam >::nblks_e,
+ this->nblks_v,
+ this->nblks_e,
+ this->lacc);
+
+ // Check the outputs only for blocks in the range
+ // { [i_startblk, i_endblk], [slev,elev], [i_startidx, i_endidx] }
+ for (int block = this->i_startblk; block <= this->i_endblk; ++block) {
+ for (int level = this->slev; level < this->elev; ++level) {
+ for (int i = this->i_startidx; i < this->i_endidx; ++i) {
+ // Compute the linear index for a 3D array in column-major order:
+ size_t idx = i + level * this->nproma + block * this->nproma * this->nlev;
+ // Since every contribution is 1 and there are 2 stencil points, expect 2.
+ EXPECT_NEAR(this->p_edge_out[idx], static_cast<OutType>(2), static_cast<OutType>(1e-5))
+ << "Failure at block " << block << ", level " << level << ", index " << i;
+ }
+ }
+ }
+}
+
+
+////////////////////////////////////////////////////////////////////////////////
+//
+// ! cells2edges
+//
+////////////////////////////////////////////////////////////////////////////////
+
+
+
+// Define a typed test fixture with one precision of data
+template <typename Types>
+class Cells2edgesScalarLibTestFixture : public testing::Test, public interp_dimensions{
+public:
+ using InType = typename Types::in_type;
+ using OutType = typename Types::out_type;
+
+ // Arrays stored in std::vector.
+ std::vector<InType> p_cell_in; // Dimensions: (nproma, nlev, nblks_c)
+ std::vector<int> edge_cell_idx; // Dimensions: (nproma, nblks_e, 2)
+ std::vector<int> edge_cell_blk; // Dimensions: (nproma, nblks_e, 2)
+ std::vector<OutType> coeff_int; // Dimensions: (nproma, 2, nblks_e)
+ std::vector<OutType> p_edge_out; // Dimensions: (nproma, nlev, nblks_e)
+
+ // TODO review these values
+ const int patch_id = 0;
+ const bool l_limited_area = false;
+ const bool lfill_latbc = false;
+ std::vector<int> i_startblk_in; // Dimensions: (2)
+ std::vector<int> i_endblk_in; // Dimensions: (2)
+ std::vector<int> i_startidx_in; // Dimensions: (2)
+ std::vector<int> i_endidx_in; // Dimensions: (2)
+
+ Cells2edgesScalarLibTestFixture() {
+ // Allocate and initialize inputs.
+ p_cell_in.resize(num_elements_3d<InType>(nproma, nlev, nblks_c), static_cast<InType>(1));
+ edge_cell_idx.resize(num_elements_3d<int>(nproma, nblks_e, 2), 1);
+ edge_cell_blk.resize(num_elements_3d<int>(nproma, nblks_e, 2), 0);
+ coeff_int.resize(num_elements_3d<InType>(nproma, 2, nblks_e), static_cast<OutType>(1));
+
+ // Allocate output arrays and initialize to zero.
+ p_edge_out.resize(num_elements_3d<OutType>(nproma, nlev, nblks_e), static_cast<OutType>(0));
+
+ // Allocate neighbour indexes
+ i_startblk_in.resize(2, i_startblk);
+ i_endblk_in.resize(2, i_endblk);
+ i_startidx_in.resize(2, i_startidx);
+ i_endidx_in.resize(2, i_endidx);
+ }
+};
+
+// Add test suite
+TYPED_TEST_SUITE(Cells2edgesScalarLibTestFixture, OneTypes);
+
+// Add test
+TYPED_TEST(Cells2edgesScalarLibTestFixture, cells2edges) {
+ using InType = typename TestFixture::InType;
+ using OutType = typename TestFixture::OutType;
+
+ // Call the function
+ cells2edges_scalar_lib<InType>(
+ this->p_cell_in.data(),
+ this->edge_cell_idx.data(),
+ this->edge_cell_blk.data(),
+ this->coeff_int.data(),
+ this->p_edge_out.data(),
+ this->i_startblk_in.data(),
+ this->i_endblk_in.data(),
+ this->i_startidx_in.data(),
+ this->i_endidx_in.data(),
+ this->slev,
+ this->elev,
+ this->nproma,
+ this->nlev,
+ this->nblks_c,
+ this->nblks_e,
+ this->patch_id,
+ this->l_limited_area,
+ this->lfill_latbc,
this->lacc);
// Check the outputs only for blocks in the range
// { [i_startblk, i_endblk], [slev,elev], [i_startidx_in, i_endidx_in] }
for (int block = this->i_startblk; block <= this->i_endblk; ++block) {
for (int level = this->slev; level < this->elev; ++level) {
- for (int i = this->i_startidx_in; i < this->i_endidx_in; ++i) {
+ for (int i = this->i_startidx; i < this->i_endidx; ++i) {
// Compute the linear index for a 3D array in column-major order:
size_t idx = i + level * this->nproma + block * this->nproma * this->nlev;
- // Since every contribution is 1 and there are 6 stencil points, expect 6.
+ // Since every contribution is 1 and there are 2 stencil points, expect 2.
EXPECT_NEAR(this->p_edge_out[idx], static_cast<OutType>(2), static_cast<OutType>(1e-5))
<< "Failure at block " << block << ", level " << level << ", index " << i;
}
}
}
+}
+
+
+////////////////////////////////////////////////////////////////////////////////
+//
+// ! edges2verts
+//
+////////////////////////////////////////////////////////////////////////////////
- // Restore output
- this->p_edge_out = tmp_p_edge_out;
+// Define a typed test fixture with one precision of data
+template <typename Types>
+class Edges2vertsScalarLibTestFixture : public testing::Test, public interp_dimensions{
+public:
+ using InType = typename Types::in_type;
+ using OutType = typename Types::in_type;
+
+ // Arrays stored in std::vector.
+ std::vector<InType> p_edge_in; // Dimensions: (nproma, nlev, nblks_e)
+ std::vector<int> edge_vert_idx; // Dimensions: (nproma, nblks_e, 6)
+ std::vector<int> edge_vert_blk; // Dimensions: (nproma, nblks_e, 6)
+ std::vector<OutType> v_int; // Dimensions: (nproma, 6-cell_type, nblks_v)
+ std::vector<OutType> p_vert_out; // Dimensions: (nproma, nlev, nblks_v)
+
+ const int cell_type = 6;
+
+ Edges2vertsScalarLibTestFixture() {
+ // Allocate and initialize inputs.
+ p_edge_in.resize(num_elements_3d<InType>(nproma, nlev, nblks_e), static_cast<InType>(1));
+ edge_vert_idx.resize(num_elements_3d<int>(nproma, nblks_e, 6), 1);
+ edge_vert_blk.resize(num_elements_3d<int>(nproma, nblks_e, 6), 0);
+ v_int.resize(num_elements_3d<InType>(nproma, 6, nblks_v), static_cast<OutType>(1));
+
+ // Allocate output arrays and initialize to zero.
+ p_vert_out.resize(num_elements_3d<OutType>(nproma, nlev, nblks_v), static_cast<OutType>(0));
+ }
+};
+
+// Add test suite
+TYPED_TEST_SUITE(Edges2vertsScalarLibTestFixture, OneTypes);
+
+// Add test
+TYPED_TEST(Edges2vertsScalarLibTestFixture, cells2edges) {
+ using InType = typename TestFixture::InType;
+ using OutType = typename TestFixture::OutType;
+
+ // Call the function
+ edges2verts_scalar_lib<InType>(
+ this->p_edge_in.data(),
+ this->edge_vert_idx.data(),
+ this->edge_vert_blk.data(),
+ this->v_int.data(),
+ this->p_vert_out.data(),
+ this->i_startblk,
+ this->i_endblk,
+ this->i_startidx,
+ this->i_endidx,
+ this->slev,
+ this->elev,
+ this->nproma,
+ this->nlev,
+ this->nblks_e,
+ this->nblks_v,
+ this->cell_type,
+ this->lacc);
+
+ // Check the outputs only for blocks in the range
+ // { [i_startblk, i_endblk], [slev,elev], [i_startidx_in, i_endidx_in] }
+ for (int block = this->i_startblk; block <= this->i_endblk; ++block) {
+ for (int level = this->slev; level < this->elev; ++level) {
+ for (int i = this->i_startidx; i < this->i_endidx; ++i) {
+ // Compute the linear index for a 3D array in column-major order:
+ size_t idx = i + level * this->nproma + block * this->nproma * this->nlev;
+ // Since every contribution is 1 and there are 6 stencil points, expect 6.
+ EXPECT_NEAR(this->p_vert_out[idx], static_cast<OutType>(6), static_cast<OutType>(1e-5))
+ << "Failure at block " << block << ", level " << level << ", index " << i;
+ }
+ }
+ }
+}
+
+
+////////////////////////////////////////////////////////////////////////////////
+//
+// ! edges2cells
+//
+////////////////////////////////////////////////////////////////////////////////
+
+
+// Define a typed test fixture with one precision of data
+template <typename Types>
+class Edges2cellsScalarLibTestFixture : public testing::Test, public interp_dimensions{
+public:
+ using InType = typename Types::in_type;
+ using OutType = typename Types::in_type;
+
+ // Arrays stored in std::vector.
+ std::vector<InType> p_edge_in; // Dimensions: (nproma, nlev, nblks_e)
+ std::vector<int> edge_idx; // Dimensions: (nproma, nblks_c, 3)
+ std::vector<int> edge_blk; // Dimensions: (nproma, nblks_c, 3)
+ std::vector<OutType> coeff_int; // Dimensions: (nproma, 3-cell_type, nblks_c)
+ std::vector<OutType> p_cell_out; // Dimensions: (nproma, nlev, nblks_c)
+
+ const int cell_type = 3;
+
+ Edges2cellsScalarLibTestFixture() {
+ // Allocate and initialize inputs.
+ p_edge_in.resize(num_elements_3d<InType>(nproma, nlev, nblks_e), static_cast<InType>(1));
+ edge_idx.resize(num_elements_3d<int>(nproma, nblks_c, cell_type), 1);
+ edge_blk.resize(num_elements_3d<int>(nproma, nblks_c, cell_type), 0);
+ coeff_int.resize(num_elements_3d<InType>(nproma, cell_type, nblks_c), static_cast<OutType>(1));
+
+ // Allocate output arrays and initialize to zero.
+ p_cell_out.resize(num_elements_3d<OutType>(nproma, nlev, nblks_c), static_cast<OutType>(0));
+ }
+};
+
+// Add test suite
+TYPED_TEST_SUITE(Edges2cellsScalarLibTestFixture, OneTypes);
+
+// Add test
+TYPED_TEST(Edges2cellsScalarLibTestFixture, cells2edges) {
+ using InType = typename TestFixture::InType;
+ using OutType = typename TestFixture::OutType;
+
+ // Call the function
+ edges2cells_scalar_lib<InType>(
+ this->p_edge_in.data(),
+ this->edge_idx.data(),
+ this->edge_blk.data(),
+ this->coeff_int.data(),
+ this->p_cell_out.data(),
+ this->i_startblk,
+ this->i_endblk,
+ this->i_startidx,
+ this->i_endidx,
+ this->slev,
+ this->elev,
+ this->nproma,
+ this->nlev,
+ this->nblks_e,
+ this->nblks_c,
+ this->lacc);
+
+ // Check the outputs only for blocks in the range
+ // { [i_startblk, i_endblk], [slev,elev], [i_startidx_in, i_endidx_in] }
+ for (int block = this->i_startblk; block <= this->i_endblk; ++block) {
+ for (int level = this->slev; level < this->elev; ++level) {
+ for (int i = this->i_startidx; i < this->i_endidx; ++i) {
+ // Compute the linear index for a 3D array in column-major order:
+ size_t idx = i + level * this->nproma + block * this->nproma * this->nlev;
+ // Since every contribution is 1 and there are 3 stencil points, expect 3.
+ EXPECT_NEAR(this->p_cell_out[idx], static_cast<OutType>(3), static_cast<OutType>(1e-5))
+ << "Failure at block " << block << ", level " << level << ", index " << i;
+ }
+ }
+ }
}
+
+// TODO:
+// cells2verts, cells2verts ri, verts2cells, cell_avg_lib
\ No newline at end of file
--
GitLab
From e7527f1e94049fa38231332de8575d93469134fb Mon Sep 17 00:00:00 2001
From: Dylan Kierans <kierans@dkrz.de>
Date: Wed, 26 Feb 2025 17:56:00 +0100
Subject: [PATCH 19/36] Fix bounds nproma in verts2cells
---
src/interpolation/mo_lib_interpolation_scalar.cpp | 6 ++++--
1 file changed, 4 insertions(+), 2 deletions(-)
diff --git a/src/interpolation/mo_lib_interpolation_scalar.cpp b/src/interpolation/mo_lib_interpolation_scalar.cpp
index d5f3b8f..9529125 100644
--- a/src/interpolation/mo_lib_interpolation_scalar.cpp
+++ b/src/interpolation/mo_lib_interpolation_scalar.cpp
@@ -404,9 +404,11 @@ void verts2cells_scalar_lib(
// vertex based scalar output field, dim: (nproma,nlev,nblks_c)
UnmanagedT3D p_cell_out_view(p_cell_out, nproma, nlev, nblks_c);
- for (int jb = 1; jb<nblks_c + 1; ++jb){
+ for (int jb = 0; jb<nblks_c; ++jb){
- int nlen = (jb != nblks_c ? nproma: npromz_c);
+ int nlen;
+ if (jb != nblks_c){ nlen = nproma; }
+ else { nlen = npromz_c}
Kokkos::MDRangePolicy<Kokkos::Rank<2>> innerPolicy(
{slev, 0}, {elev + 1, nlen + 1});
--
GitLab
From e0a23eac5464b78bd693581646907f0cf797d78d Mon Sep 17 00:00:00 2001
From: Dylan Kierans <kierans@dkrz.de>
Date: Wed, 26 Feb 2025 17:57:46 +0100
Subject: [PATCH 20/36] All tests for mo_lib_interpolation_scalar
---
test/c/test_interpolation_scalar.cpp | 316 ++++++++++++++++++++++++++-
1 file changed, 311 insertions(+), 5 deletions(-)
diff --git a/test/c/test_interpolation_scalar.cpp b/test/c/test_interpolation_scalar.cpp
index 78c615e..c4e414d 100644
--- a/test/c/test_interpolation_scalar.cpp
+++ b/test/c/test_interpolation_scalar.cpp
@@ -188,7 +188,7 @@ public:
};
// Add test suite
-TYPED_TEST_SUITE(Cells2edgesScalarLibTestFixture, OneTypes);
+TYPED_TEST_SUITE(Cells2edgesScalarLibTestFixture, MixedTypes);
// Add test
TYPED_TEST(Cells2edgesScalarLibTestFixture, cells2edges) {
@@ -244,7 +244,7 @@ template <typename Types>
class Edges2vertsScalarLibTestFixture : public testing::Test, public interp_dimensions{
public:
using InType = typename Types::in_type;
- using OutType = typename Types::in_type;
+ using OutType = typename Types::out_type;
// Arrays stored in std::vector.
std::vector<InType> p_edge_in; // Dimensions: (nproma, nlev, nblks_e)
@@ -323,7 +323,7 @@ template <typename Types>
class Edges2cellsScalarLibTestFixture : public testing::Test, public interp_dimensions{
public:
using InType = typename Types::in_type;
- using OutType = typename Types::in_type;
+ using OutType = typename Types::out_type;
// Arrays stored in std::vector.
std::vector<InType> p_edge_in; // Dimensions: (nproma, nlev, nblks_e)
@@ -388,5 +388,311 @@ TYPED_TEST(Edges2cellsScalarLibTestFixture, cells2edges) {
}
}
-// TODO:
-// cells2verts, cells2verts ri, verts2cells, cell_avg_lib
\ No newline at end of file
+////////////////////////////////////////////////////////////////////////////////
+//
+// ! cells2verts
+//
+////////////////////////////////////////////////////////////////////////////////
+
+
+// Define a typed test fixture with one precision of data
+template <typename Types>
+class Cells2vertsScalarLibTestFixture : public testing::Test, public interp_dimensions{
+public:
+ using InType = typename Types::in_type;
+ using OutType = typename Types::out_type;
+
+ // Arrays stored in std::vector.
+ std::vector<InType> p_cell_in; // Dimensions: (nproma, nlev, nblks_c)
+ std::vector<int> vert_cell_idx; // Dimensions: (nproma, nblks_v, 6)
+ std::vector<int> vert_cell_blk; // Dimensions: (nproma, nblks_v, 6)
+ std::vector<OutType> coeff_int; // Dimensions: (nproma, 9-cell_type, nblks_v)
+ std::vector<OutType> p_vert_out; // Dimensions: (nproma, nlev, nblks_v)
+
+ const int cell_type = 9;
+
+ Cells2vertsScalarLibTestFixture() {
+ // Allocate and initialize inputs.
+ p_cell_in.resize(num_elements_3d<InType>(nproma, nlev, nblks_c), static_cast<InType>(1));
+ vert_cell_idx.resize(num_elements_3d<int>(nproma, nblks_v, 6), 1);
+ vert_cell_blk.resize(num_elements_3d<int>(nproma, nblks_v, 6), 0);
+ coeff_int.resize(num_elements_3d<InType>(nproma, cell_type, nblks_v), static_cast<OutType>(1));
+
+ // Allocate output arrays and initialize to zero.
+ p_vert_out.resize(num_elements_3d<OutType>(nproma, nlev, nblks_v), static_cast<OutType>(0));
+ }
+};
+
+// Add test suite
+TYPED_TEST_SUITE(Cells2vertsScalarLibTestFixture, MixedTypes);
+
+// Add test
+TYPED_TEST(Cells2vertsScalarLibTestFixture, cells2edges) {
+ using InType = typename TestFixture::InType;
+ using OutType = typename TestFixture::OutType;
+
+ // Call the function
+ cells2verts_scalar_lib<InType,OutType>(
+ this->p_cell_in.data(),
+ this->vert_cell_idx.data(),
+ this->vert_cell_blk.data(),
+ this->coeff_int.data(),
+ this->p_vert_out.data(),
+ this->i_startblk,
+ this->i_endblk,
+ this->i_startidx,
+ this->i_endidx,
+ this->slev,
+ this->elev,
+ this->nproma,
+ this->nlev,
+ this->nblks_c,
+ this->nblks_v,
+ this->lacc,
+ this->acc_async);
+
+ // Check the outputs only for blocks in the range
+ // { [i_startblk, i_endblk], [slev,elev], [i_startidx_in, i_endidx_in] }
+ for (int block = this->i_startblk; block <= this->i_endblk; ++block) {
+ for (int level = this->slev; level < this->elev; ++level) {
+ for (int i = this->i_startidx; i < this->i_endidx; ++i) {
+ // Compute the linear index for a 3D array in column-major order:
+ size_t idx = i + level * this->nproma + block * this->nproma * this->nlev;
+ // Since every contribution is 1 and there are 6 stencil points, expect 6.
+ EXPECT_NEAR(this->p_vert_out[idx], static_cast<OutType>(6), static_cast<OutType>(1e-5))
+ << "Failure at block " << block << ", level " << level << ", index " << i;
+ }
+ }
+ }
+}
+
+////////////////////////////////////////////////////////////////////////////////
+//
+// ! cells2verts ri
+//
+////////////////////////////////////////////////////////////////////////////////
+
+
+// Define a typed test fixture with one precision of data
+template <typename Types>
+class Cells2vertsriScalarLibTestFixture : public testing::Test, public interp_dimensions{
+public:
+ using InType = typename Types::in_type;
+ using OutType = typename Types::out_type;
+
+ // Arrays stored in std::vector.
+ std::vector<InType> p_cell_in; // Dimensions: (nproma, nlev, nblks_c)
+ std::vector<int> vert_cell_idx; // Dimensions: (nproma, nblks_v, 6)
+ std::vector<int> vert_cell_blk; // Dimensions: (nproma, nblks_v, 6)
+ std::vector<InType> coeff_int; // Dimensions: (nproma, 9-cell_type, nblks_v)
+ std::vector<OutType> p_vert_out; // Dimensions: (nproma, nlev, nblks_v)
+
+ const int cell_type = 9;
+
+ Cells2vertsriScalarLibTestFixture() {
+ // Allocate and initialize inputs.
+ p_cell_in.resize(num_elements_3d<InType>(nproma, nlev, nblks_c), static_cast<InType>(1));
+ vert_cell_idx.resize(num_elements_3d<int>(nproma, nblks_v, 6), 1);
+ vert_cell_blk.resize(num_elements_3d<int>(nproma, nblks_v, 6), 0);
+ coeff_int.resize(num_elements_3d<InType>(nproma, cell_type, nblks_v), static_cast<InType>(1));
+
+ // Allocate output arrays and initialize to zero.
+ p_vert_out.resize(num_elements_3d<OutType>(nproma, nlev, nblks_v), static_cast<OutType>(0));
+ }
+};
+
+// Add test suite
+TYPED_TEST_SUITE(Cells2vertsriScalarLibTestFixture, MixedTypes);
+
+// Add test
+TYPED_TEST(Cells2vertsriScalarLibTestFixture, cells2edges) {
+ using InType = typename TestFixture::InType;
+ using OutType = typename TestFixture::OutType;
+
+ // Call the function
+ cells2verts_scalar_ri_lib<InType,OutType>(
+ this->p_cell_in.data(),
+ this->vert_cell_idx.data(),
+ this->vert_cell_blk.data(),
+ this->coeff_int.data(),
+ this->p_vert_out.data(),
+ this->i_startblk,
+ this->i_endblk,
+ this->i_startidx,
+ this->i_endidx,
+ this->slev,
+ this->elev,
+ this->nproma,
+ this->nlev,
+ this->nblks_c,
+ this->nblks_v,
+ this->lacc,
+ this->acc_async);
+
+ // Check the outputs only for blocks in the range
+ // { [i_startblk, i_endblk], [slev,elev], [i_startidx_in, i_endidx_in] }
+ for (int block = this->i_startblk; block <= this->i_endblk; ++block) {
+ for (int level = this->slev; level < this->elev; ++level) {
+ for (int i = this->i_startidx; i < this->i_endidx; ++i) {
+ // Compute the linear index for a 3D array in column-major order:
+ size_t idx = i + level * this->nproma + block * this->nproma * this->nlev;
+ // Since every contribution is 1 and there are 6 stencil points, expect 6.
+ EXPECT_NEAR(this->p_vert_out[idx], static_cast<OutType>(6), static_cast<OutType>(1e-5))
+ << "Failure at block " << block << ", level " << level << ", index " << i;
+ }
+ }
+ }
+}
+
+
+////////////////////////////////////////////////////////////////////////////////
+//
+// ! verts2cells
+//
+////////////////////////////////////////////////////////////////////////////////
+
+
+// Define a typed test fixture with one precision of data
+template <typename Types>
+class Verts2cellsScalarLibTestFixture : public testing::Test, public interp_dimensions{
+public:
+ using InType = typename Types::in_type;
+ using OutType = typename Types::out_type;
+
+ // Arrays stored in std::vector.
+ std::vector<InType> p_vert_in; // Dimensions: (nproma, nlev, nblks_v)
+ std::vector<int> cell_index_idx; // Dimensions: (nproma, nblks_c, 3)
+ std::vector<int> cell_index_blk; // Dimensions: (nproma, nblks_c, 3)
+ std::vector<InType> coeff_int; // Dimensions: (nproma, 3-cell_type, nblks_c)
+ std::vector<OutType> p_cell_out; // Dimensions: (nproma, nlev, nblks_c)
+
+ const int cell_type = 3;
+ const int npromz_c = 32;
+
+ Verts2cellsScalarLibTestFixture() {
+ // Allocate and initialize inputs.
+ p_vert_in.resize(num_elements_3d<InType>(nproma, nlev, nblks_v), static_cast<InType>(1));
+ cell_index_idx.resize(num_elements_3d<int>(nproma, nblks_c, 3), 1);
+ cell_index_blk.resize(num_elements_3d<int>(nproma, nblks_c, 3), 0);
+ coeff_int.resize(num_elements_3d<InType>(nproma, cell_type, nblks_c), static_cast<InType>(1));
+
+ // Allocate output arrays and initialize to zero.
+ p_cell_out.resize(num_elements_3d<OutType>(nproma, nlev, nblks_c), static_cast<OutType>(0));
+ }
+};
+
+// Add test suite
+TYPED_TEST_SUITE(Verts2cellsScalarLibTestFixture, OneTypes);
+
+// Add test
+TYPED_TEST(Verts2cellsScalarLibTestFixture, cells2edges) {
+ using InType = typename TestFixture::InType;
+ using OutType = typename TestFixture::OutType;
+
+ // Call the function
+ verts2cells_scalar_lib<InType>(
+ this->p_vert_in.data(),
+ this->cell_index_idx.data(),
+ this->cell_index_blk.data(),
+ this->coeff_int.data(),
+ this->p_cell_out.data(),
+ this->nblks_c,
+ this->npromz_c,
+ this->slev,
+ this->elev,
+ this->nproma,
+ this->nlev,
+ this->nblks_v,
+ this->lacc);
+
+
+ // Check the outputs only for blocks in the range
+ // { [i_startblk, i_endblk], [slev,elev], [i_startidx_in, i_endidx_in] }
+ for (int block = this->i_startblk; block <= this->i_endblk; ++block) {
+ for (int level = this->slev; level < this->elev; ++level) {
+ for (int i = this->i_startidx; i < this->i_endidx; ++i) {
+ // Compute the linear index for a 3D array in column-major order:
+ size_t idx = i + level * this->nproma + block * this->nproma * this->nlev;
+ // Since every contribution is 1 and there are 3 stencil points, expect 3.
+ EXPECT_NEAR(this->p_cell_out[idx], static_cast<OutType>(3), static_cast<OutType>(1e-5))
+ << "Failure at block " << block << ", level " << level << ", index " << i;
+ }
+ }
+ }
+}
+
+
+////////////////////////////////////////////////////////////////////////////////
+//
+// ! cell_avg
+//
+////////////////////////////////////////////////////////////////////////////////
+
+
+// Define a typed test fixture with one precision of data
+template <typename Types>
+class CellAvgScalarLibTestFixture : public testing::Test, public interp_dimensions{
+public:
+ using InType = typename Types::in_type;
+ using OutType = typename Types::out_type;
+
+ // Arrays stored in std::vector.
+ std::vector<InType> psi_c; // Dimensions: (nproma, nlev, nblks_c)
+ std::vector<int> cell_neighbor_idx; // Dimensions: (nproma, nblks_c, 3)
+ std::vector<int> cell_neighbor_blk; // Dimensions: (nproma, nblks_c, 3)
+ std::vector<InType> avg_coeff; // Dimensions: (nproma, nlev, nblks_c)
+ std::vector<OutType> avg_psi_c; // Dimensions: (nproma, nlev, nblks_c)
+
+ CellAvgScalarLibTestFixture() {
+ // Allocate and initialize inputs.
+ psi_c.resize(num_elements_3d<InType>(nproma, nlev, nblks_c), static_cast<InType>(1));
+ cell_neighbor_idx.resize(num_elements_3d<int>(nproma, nblks_c, 3), 1);
+ cell_neighbor_blk.resize(num_elements_3d<int>(nproma, nblks_c, 3), 0);
+ avg_coeff.resize(num_elements_3d<InType>(nproma, nlev, nblks_c), static_cast<InType>(1));
+
+ // Allocate output arrays and initialize to zero.
+ avg_psi_c.resize(num_elements_3d<OutType>(nproma, nlev, nblks_c), static_cast<OutType>(0));
+ }
+};
+
+// Add test suite
+TYPED_TEST_SUITE(CellAvgScalarLibTestFixture, OneTypes);
+
+// Add test
+TYPED_TEST(CellAvgScalarLibTestFixture, cells2edges) {
+ using InType = typename TestFixture::InType;
+ using OutType = typename TestFixture::OutType;
+
+ // Call the function
+ cell_avg_lib<InType>(
+ this->psi_c.data(),
+ this->cell_neighbor_idx.data(),
+ this->cell_neighbor_blk.data(),
+ this->avg_coeff.data(),
+ this->avg_psi_c.data(),
+ this->i_startblk,
+ this->i_endblk,
+ this->i_startidx,
+ this->i_endidx,
+ this->slev,
+ this->elev,
+ this->nproma,
+ this->nlev,
+ this->nblks_c,
+ this->lacc);
+
+
+ // Check the outputs only for blocks in the range
+ // { [i_startblk, i_endblk], [slev,elev], [i_startidx_in, i_endidx_in] }
+ for (int block = this->i_startblk; block <= this->i_endblk; ++block) {
+ for (int level = this->slev; level < this->elev; ++level) {
+ for (int i = this->i_startidx; i < this->i_endidx; ++i) {
+ // Compute the linear index for a 3D array in column-major order:
+ size_t idx = i + level * this->nproma + block * this->nproma * this->nlev;
+ // Since every contribution is 1 and there are 4 stencil points, expect 4.
+ EXPECT_NEAR(this->avg_psi_c[idx], static_cast<OutType>(4), static_cast<OutType>(1e-5))
+ << "Failure at block " << block << ", level " << level << ", index " << i;
+ }
+ }
+ }
+}
--
GitLab
From a051e487a1f105dc2679375688a83a766c6ec8a3 Mon Sep 17 00:00:00 2001
From: Dylan Kierans <kierans@dkrz.de>
Date: Wed, 26 Feb 2025 17:59:32 +0100
Subject: [PATCH 21/36] restoring rbf_vec_interpol_vertex_lib.cpp
---
.../mo_lib_intp_rbf-rbf_vec_interpol_vertex_lib.cpp | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/src/interpolation/mo_lib_intp_rbf-rbf_vec_interpol_vertex_lib.cpp b/src/interpolation/mo_lib_intp_rbf-rbf_vec_interpol_vertex_lib.cpp
index 0a41fe8..c9b776e 100644
--- a/src/interpolation/mo_lib_intp_rbf-rbf_vec_interpol_vertex_lib.cpp
+++ b/src/interpolation/mo_lib_intp_rbf-rbf_vec_interpol_vertex_lib.cpp
@@ -101,7 +101,7 @@ void rbf_vec_interpol_vertex_lib(
int i_startidx; // start index
int i_endidx; // end index
- for (jb=i_startblk; jb < i_endblk + 1; ++jb){
+ for (jb=i_startblk; jb <= i_endblk; ++jb){
get_indices_v_lib(i_startidx_in, i_endidx_in, nproma, jb, i_startblk, i_endblk,
i_startidx, i_endidx);
--
GitLab
From 7fbfbe9f382d22fa91f8b78b9909d9059993d9db Mon Sep 17 00:00:00 2001
From: Dylan Kierans <kierans@dkrz.de>
Date: Wed, 26 Feb 2025 17:34:21 +0000
Subject: [PATCH 22/36] Typo mo_lib_interpolation_scalar.cpp
---
src/interpolation/mo_lib_interpolation_scalar.cpp | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/src/interpolation/mo_lib_interpolation_scalar.cpp b/src/interpolation/mo_lib_interpolation_scalar.cpp
index 9529125..9306728 100644
--- a/src/interpolation/mo_lib_interpolation_scalar.cpp
+++ b/src/interpolation/mo_lib_interpolation_scalar.cpp
@@ -408,7 +408,7 @@ void verts2cells_scalar_lib(
int nlen;
if (jb != nblks_c){ nlen = nproma; }
- else { nlen = npromz_c}
+ else { nlen = npromz_c; }
Kokkos::MDRangePolicy<Kokkos::Rank<2>> innerPolicy(
{slev, 0}, {elev + 1, nlen + 1});
--
GitLab
From 53be884790df4f6af1d0e745ea19010058903a7f Mon Sep 17 00:00:00 2001
From: Dylan Kierans <kierans@dkrz.de>
Date: Thu, 27 Feb 2025 10:49:32 +0100
Subject: [PATCH 23/36] Fixing mixed-precision cells2edges, and loop bounds in
verts2cells
---
src/interpolation/interpolation_bindings.cpp | 2 +-
.../mo_lib_interpolation_scalar.cpp | 18 +++++++++---------
.../mo_lib_interpolation_scalar.hpp | 4 ++--
3 files changed, 12 insertions(+), 12 deletions(-)
diff --git a/src/interpolation/interpolation_bindings.cpp b/src/interpolation/interpolation_bindings.cpp
index 4476035..f986447 100644
--- a/src/interpolation/interpolation_bindings.cpp
+++ b/src/interpolation/interpolation_bindings.cpp
@@ -127,7 +127,7 @@ void cells2edges_scalar_lib_sp2dp(const float* p_cell_in, const int* edge_cell_i
const int nblk_c, const int nblks_e, const int patch_id,
const bool l_limited_area, const bool lfill_latbc, const bool lacc){
- cells2edges_scalar_lib<double, float>(p_cell_in,edge_cell_idx,edge_cell_blk,coeff_int,p_edge_out,i_startblk_in,
+ cells2edges_scalar_lib<float, double>(p_cell_in,edge_cell_idx,edge_cell_blk,coeff_int,p_edge_out,i_startblk_in,
i_endblk_in,i_startidx_in,i_endidx_in,slev,elev,nproma,nlev,
nblk_c,nblks_e,patch_id,l_limited_area,lfill_latbc,lacc);
diff --git a/src/interpolation/mo_lib_interpolation_scalar.cpp b/src/interpolation/mo_lib_interpolation_scalar.cpp
index 9306728..c59f928 100644
--- a/src/interpolation/mo_lib_interpolation_scalar.cpp
+++ b/src/interpolation/mo_lib_interpolation_scalar.cpp
@@ -75,8 +75,8 @@ void verts2edges_scalar_lib(const T* p_vertex_in,
/// velocity points.
///
template <typename T, typename S>
-void cells2edges_scalar_lib(const S* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
- const T* coeff_int, T* p_edge_out, const int* i_startblk_in,
+void cells2edges_scalar_lib(const T* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
+ const S* coeff_int, S* p_edge_out, const int* i_startblk_in,
const int* i_endblk_in, const int* i_startidx_in, const int* i_endidx_in,
const int slev, const int elev, const int nproma, const int nlev,
const int nblk_c, const int nblks_e, const int patch_id,
@@ -85,14 +85,14 @@ void cells2edges_scalar_lib(const S* p_cell_in, const int* edge_cell_idx, const
// Wrap raw pointers in unmanaged Kokkos Views.
typedef Kokkos::View<const T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstT3D;
typedef Kokkos::View<const S***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstS3D;
- typedef Kokkos::View<T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedT3D;
+ typedef Kokkos::View<S***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedS3D;
typedef Kokkos::View<const int***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstInt3D;
- UnmanagedConstS3D p_cell_in_view (p_cell_in, nproma, nlev, nblk_c);
+ UnmanagedConstT3D p_cell_in_view (p_cell_in, nproma, nlev, nblk_c);
UnmanagedConstInt3D iidx_view (edge_cell_idx, nproma, nblks_e, 2);
UnmanagedConstInt3D iblk_view (edge_cell_blk, nproma, nblks_e, 2);
- UnmanagedConstT3D coeff_int_view (coeff_int, nproma, 2, nblks_e);
- UnmanagedT3D p_edge_out_view (p_edge_out, nproma, nlev, nblks_e);
+ UnmanagedConstS3D coeff_int_view (coeff_int, nproma, 2, nblks_e);
+ UnmanagedS3D p_edge_out_view (p_edge_out, nproma, nlev, nblks_e);
//Fill outermost nest boundary
int i_startblk, i_endblk;
@@ -411,7 +411,7 @@ void verts2cells_scalar_lib(
else { nlen = npromz_c; }
Kokkos::MDRangePolicy<Kokkos::Rank<2>> innerPolicy(
- {slev, 0}, {elev + 1, nlen + 1});
+ {slev, 0}, {elev + 1, nlen});
Kokkos::parallel_for("cell_avg_lib_inner", innerPolicy,
KOKKOS_LAMBDA(const int jk, const int jc) {
@@ -524,8 +524,8 @@ void cells2edges_scalar_lib<float, float>(const float* p_cell_in, const int* edg
const int nblk_c, const int nblks_e, const int patch_id,
const bool l_limited_area, const bool lfill_latbc, const bool lacc);
-template
-void cells2edges_scalar_lib<double, float>(const float* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
+template // sp2dp
+void cells2edges_scalar_lib<float, double>(const float* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
const double* coeff_int, double* p_edge_out, const int* i_startblk_in,
const int* i_endblk_in, const int* i_startidx_in, const int* i_endidx_in,
const int slev, const int elev, const int nproma, const int nlev,
diff --git a/src/interpolation/mo_lib_interpolation_scalar.hpp b/src/interpolation/mo_lib_interpolation_scalar.hpp
index 95febf1..16223ec 100644
--- a/src/interpolation/mo_lib_interpolation_scalar.hpp
+++ b/src/interpolation/mo_lib_interpolation_scalar.hpp
@@ -22,8 +22,8 @@ void verts2edges_scalar_lib(const T* p_vertex_in,
;
template <typename T, typename S>
-void cells2edges_scalar_lib(const S* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
- const T* coeff_int, T* p_edge_out, const int* i_startblk_in,
+void cells2edges_scalar_lib(const T* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
+ const S* coeff_int, S* p_edge_out, const int* i_startblk_in,
const int* i_endblk_in, const int* i_startidx_in, const int* i_endidx_in,
const int slev, const int elev, const int nproma, const int nlev,
const int nblk_c, const int nblks_e, const int patch_id,
--
GitLab
From e253335c2b3720813d5e0f6398707fb771e12535 Mon Sep 17 00:00:00 2001
From: Dylan Kierans <kierans@dkrz.de>
Date: Thu, 27 Feb 2025 10:50:29 +0100
Subject: [PATCH 24/36] mixed precision sp2dp in test_interpolation_scalar
---
test/c/test_interpolation_scalar.cpp | 8 ++++++--
1 file changed, 6 insertions(+), 2 deletions(-)
diff --git a/test/c/test_interpolation_scalar.cpp b/test/c/test_interpolation_scalar.cpp
index c4e414d..ac2eddc 100644
--- a/test/c/test_interpolation_scalar.cpp
+++ b/test/c/test_interpolation_scalar.cpp
@@ -37,6 +37,10 @@ typedef ::testing::Types< MixedPrecision<double, double>,
MixedPrecision<double, float>,
MixedPrecision<float, float> > MixedTypes;
+typedef ::testing::Types< MixedPrecision<double, double>,
+ MixedPrecision<float, double>,
+ MixedPrecision<float, float> > MixedTypesSP2DP;
+
// Define the list of type pairs we want to test.
typedef ::testing::Types< MixedPrecision<double, double>,
MixedPrecision<float, float > > OneTypes;
@@ -188,7 +192,7 @@ public:
};
// Add test suite
-TYPED_TEST_SUITE(Cells2edgesScalarLibTestFixture, MixedTypes);
+TYPED_TEST_SUITE(Cells2edgesScalarLibTestFixture, MixedTypesSP2DP);
// Add test
TYPED_TEST(Cells2edgesScalarLibTestFixture, cells2edges) {
@@ -196,7 +200,7 @@ TYPED_TEST(Cells2edgesScalarLibTestFixture, cells2edges) {
using OutType = typename TestFixture::OutType;
// Call the function
- cells2edges_scalar_lib<InType>(
+ cells2edges_scalar_lib<InType,OutType>(
this->p_cell_in.data(),
this->edge_cell_idx.data(),
this->edge_cell_blk.data(),
--
GitLab
From 03f3b980be7cff8bca90c6233c6bcc46a3b24050 Mon Sep 17 00:00:00 2001
From: Pradipta Samanta <samanta@dkrz.de>
Date: Thu, 27 Feb 2025 22:42:07 +0100
Subject: [PATCH 25/36] worked on two of the test cases
---
test/c/test_interpolation_scalar.cpp | 187 ++++++++++++---------------
1 file changed, 81 insertions(+), 106 deletions(-)
diff --git a/test/c/test_interpolation_scalar.cpp b/test/c/test_interpolation_scalar.cpp
index ac2eddc..f46c2b4 100644
--- a/test/c/test_interpolation_scalar.cpp
+++ b/test/c/test_interpolation_scalar.cpp
@@ -67,48 +67,58 @@ public:
const bool acc_async = false; // No asynchronous execution.
};
-////////////////////////////////////////////////////////////////////////////////
-//
-// ! verts2edges
-//
-////////////////////////////////////////////////////////////////////////////////
-
-// Define a test fixture
-template <typename Types>
-class Verts2edgesScalarLibTestFixture : public testing::Test, public interp_dimensions {
+template <typename T>
+class InterpolationScalarTypedTestFixture : public ::testing::Test, public interp_dimensions {
public:
- using InType = typename Types::in_type;
- using OutType = typename Types::out_type;
- // Arrays stored in std::vector.
- std::vector<InType> p_vertex_in; // Dimensions: (nproma, nlev, nblks_v)
+ // Arrays used for verts2edges
+ std::vector<T> p_vertex_in; // Dimensions: (nproma, nlev, nblks_v)
std::vector<int> edge_vertex_idx; // Dimensions: (nproma, nblks_e, 4)
std::vector<int> edge_vertex_blk; // Dimensions: (nproma, nblks_e, 4)
- std::vector<InType> coeff_int; // Dimensions: (nproma, 2, nblks_e)
- std::vector<OutType> p_edge_out; // Dimensions: (nproma, nlev, nblks_e)
+ std::vector<T> coeff_int; // Dimensions: (nproma, 2, nblks_e)
+ std::vector<T> p_edge_out; // Dimensions: (nproma, nlev, nblks_e)
- Verts2edgesScalarLibTestFixture() {
- // Allocate and initialize inputs.
- p_vertex_in.resize(num_elements_3d<InType>(nproma, nlev, nblks_v), static_cast<InType>(1));
+ // Arrays used for edges2verts
+ std::vector<T> p_edge_in; // Dimensions: (nproma, nlev, nblks_e)
+ std::vector<int> edge_vert_idx; // Dimensions: (nproma, nblks_e, 6)
+ std::vector<int> edge_vert_blk; // Dimensions: (nproma, nblks_e, 6)
+ std::vector<T> v_int; // Dimensions: (nproma, 6-cell_type, nblks_v)
+ std::vector<T> p_vert_out; // Dimensions: (nproma, nlev, nblks_v)
+ //
+ const int cell_type = 6;
+
+ InterpolationScalarTypedTestFixture() {
+ // Allocate and initialize arrays needed for verts2edges
+ p_vertex_in.resize(num_elements_3d<T>(nproma, nlev, nblks_v), static_cast<T>(1));
edge_vertex_idx.resize(num_elements_3d<int>(nproma, nblks_e, 4), 1);
edge_vertex_blk.resize(num_elements_3d<int>(nproma, nblks_e, 4), 0);
- coeff_int.resize(num_elements_3d<InType>(nproma, 2, nblks_v), static_cast<InType>(1));
+ coeff_int.resize(num_elements_3d<T>(nproma, 2, nblks_v), static_cast<T>(1));
- // Allocate output arrays and initialize to zero.
- p_edge_out.resize(num_elements_3d<OutType>(nproma, nlev, nblks_e), static_cast<OutType>(0));
+ p_edge_out.resize(num_elements_3d<T>(nproma, nlev, nblks_e), static_cast<T>(0));
+
+ // Allocate & Initialize arrays needed for edges2verts
+ p_edge_in.resize(num_elements_3d<T>(nproma, nlev, nblks_e), static_cast<T>(1));
+ edge_vert_idx.resize(num_elements_3d<int>(nproma, nblks_e, 6), 1);
+ edge_vert_blk.resize(num_elements_3d<int>(nproma, nblks_e, 6), 0);
+ v_int.resize(num_elements_3d<T>(nproma, 6, nblks_v), static_cast<T>(1));
+
+ p_vert_out.resize(num_elements_3d<T>(nproma, nlev, nblks_v), static_cast<T>(0));
}
};
-// Add test suite
-TYPED_TEST_SUITE(Verts2edgesScalarLibTestFixture, OneTypes);
+typedef ::testing::Types<float, double> SingleType;
-// Add test
-TYPED_TEST(Verts2edgesScalarLibTestFixture, verts2edges) {
- using InType = typename TestFixture::InType;
- using OutType = typename TestFixture::OutType;
+TYPED_TEST_SUITE(InterpolationScalarTypedTestFixture, SingleType);
- // Call the function
- verts2edges_scalar_lib<InType>(
+////////////////////////////////////////////////////////////////////////////////
+//
+// ! verts2edges
+//
+////////////////////////////////////////////////////////////////////////////////
+
+TYPED_TEST(InterpolationScalarTypedTestFixture, Verts2Edges) {
+
+ verts2edges_scalar_lib<TypeParam>(
this->p_vertex_in.data(),
this->edge_vertex_idx.data(),
this->edge_vertex_blk.data(),
@@ -134,13 +144,54 @@ TYPED_TEST(Verts2edgesScalarLibTestFixture, verts2edges) {
// Compute the linear index for a 3D array in column-major order:
size_t idx = i + level * this->nproma + block * this->nproma * this->nlev;
// Since every contribution is 1 and there are 2 stencil points, expect 2.
- EXPECT_NEAR(this->p_edge_out[idx], static_cast<OutType>(2), static_cast<OutType>(1e-5))
+ EXPECT_NEAR(this->p_edge_out[idx], static_cast<TypeParam>(2), static_cast<TypeParam>(1e-5))
<< "Failure at block " << block << ", level " << level << ", index " << i;
}
}
}
}
+////////////////////////////////////////////////////////////////////////////////
+//
+// ! edges2verts
+//
+////////////////////////////////////////////////////////////////////////////////
+
+TYPED_TEST(InterpolationScalarTypedTestFixture, Edges2Verts) {
+
+ edges2verts_scalar_lib<TypeParam>(
+ this->p_edge_in.data(),
+ this->edge_vert_idx.data(),
+ this->edge_vert_blk.data(),
+ this->v_int.data(),
+ this->p_vert_out.data(),
+ this->i_startblk,
+ this->i_endblk,
+ this->i_startidx,
+ this->i_endidx,
+ this->slev,
+ this->elev,
+ this->nproma,
+ this->nlev,
+ this->nblks_e,
+ this->nblks_v,
+ this->cell_type,
+ this->lacc);
+
+ // Check the outputs only for blocks in the range
+ // { [i_startblk, i_endblk], [slev,elev], [i_startidx_in, i_endidx_in] }
+ for (int block = this->i_startblk; block <= this->i_endblk; ++block) {
+ for (int level = this->slev; level < this->elev; ++level) {
+ for (int i = this->i_startidx; i < this->i_endidx; ++i) {
+ // Compute the linear index for a 3D array in column-major order:
+ size_t idx = i + level * this->nproma + block * this->nproma * this->nlev;
+ // Since every contribution is 1 and there are 6 stencil points, expect 6.
+ EXPECT_NEAR(this->p_vert_out[idx], static_cast<TypeParam>(6), static_cast<TypeParam>(1e-5))
+ << "Failure at block " << block << ", level " << level << ", index " << i;
+ }
+ }
+ }
+}
////////////////////////////////////////////////////////////////////////////////
//
@@ -237,82 +288,6 @@ TYPED_TEST(Cells2edgesScalarLibTestFixture, cells2edges) {
}
-////////////////////////////////////////////////////////////////////////////////
-//
-// ! edges2verts
-//
-////////////////////////////////////////////////////////////////////////////////
-
-// Define a typed test fixture with one precision of data
-template <typename Types>
-class Edges2vertsScalarLibTestFixture : public testing::Test, public interp_dimensions{
-public:
- using InType = typename Types::in_type;
- using OutType = typename Types::out_type;
-
- // Arrays stored in std::vector.
- std::vector<InType> p_edge_in; // Dimensions: (nproma, nlev, nblks_e)
- std::vector<int> edge_vert_idx; // Dimensions: (nproma, nblks_e, 6)
- std::vector<int> edge_vert_blk; // Dimensions: (nproma, nblks_e, 6)
- std::vector<OutType> v_int; // Dimensions: (nproma, 6-cell_type, nblks_v)
- std::vector<OutType> p_vert_out; // Dimensions: (nproma, nlev, nblks_v)
-
- const int cell_type = 6;
-
- Edges2vertsScalarLibTestFixture() {
- // Allocate and initialize inputs.
- p_edge_in.resize(num_elements_3d<InType>(nproma, nlev, nblks_e), static_cast<InType>(1));
- edge_vert_idx.resize(num_elements_3d<int>(nproma, nblks_e, 6), 1);
- edge_vert_blk.resize(num_elements_3d<int>(nproma, nblks_e, 6), 0);
- v_int.resize(num_elements_3d<InType>(nproma, 6, nblks_v), static_cast<OutType>(1));
-
- // Allocate output arrays and initialize to zero.
- p_vert_out.resize(num_elements_3d<OutType>(nproma, nlev, nblks_v), static_cast<OutType>(0));
- }
-};
-
-// Add test suite
-TYPED_TEST_SUITE(Edges2vertsScalarLibTestFixture, OneTypes);
-
-// Add test
-TYPED_TEST(Edges2vertsScalarLibTestFixture, cells2edges) {
- using InType = typename TestFixture::InType;
- using OutType = typename TestFixture::OutType;
-
- // Call the function
- edges2verts_scalar_lib<InType>(
- this->p_edge_in.data(),
- this->edge_vert_idx.data(),
- this->edge_vert_blk.data(),
- this->v_int.data(),
- this->p_vert_out.data(),
- this->i_startblk,
- this->i_endblk,
- this->i_startidx,
- this->i_endidx,
- this->slev,
- this->elev,
- this->nproma,
- this->nlev,
- this->nblks_e,
- this->nblks_v,
- this->cell_type,
- this->lacc);
-
- // Check the outputs only for blocks in the range
- // { [i_startblk, i_endblk], [slev,elev], [i_startidx_in, i_endidx_in] }
- for (int block = this->i_startblk; block <= this->i_endblk; ++block) {
- for (int level = this->slev; level < this->elev; ++level) {
- for (int i = this->i_startidx; i < this->i_endidx; ++i) {
- // Compute the linear index for a 3D array in column-major order:
- size_t idx = i + level * this->nproma + block * this->nproma * this->nlev;
- // Since every contribution is 1 and there are 6 stencil points, expect 6.
- EXPECT_NEAR(this->p_vert_out[idx], static_cast<OutType>(6), static_cast<OutType>(1e-5))
- << "Failure at block " << block << ", level " << level << ", index " << i;
- }
- }
- }
-}
////////////////////////////////////////////////////////////////////////////////
--
GitLab
From 21af82431c2db969beb0ecb4e36043966a9df47c Mon Sep 17 00:00:00 2001
From: Pradipta Samanta <samanta@dkrz.de>
Date: Tue, 4 Mar 2025 16:48:15 +0100
Subject: [PATCH 26/36] removed cell_type as a function argument
---
src/interpolation/interpolation_bindings.cpp | 8 ++++----
src/interpolation/interpolation_bindings.h | 6 +++---
src/interpolation/mo_lib_interpolation_scalar.cpp | 8 ++++----
src/interpolation/mo_lib_interpolation_scalar.hpp | 4 ++--
test/c/test_interpolation_scalar.cpp | 1 -
5 files changed, 13 insertions(+), 14 deletions(-)
diff --git a/src/interpolation/interpolation_bindings.cpp b/src/interpolation/interpolation_bindings.cpp
index f986447..0535091 100644
--- a/src/interpolation/interpolation_bindings.cpp
+++ b/src/interpolation/interpolation_bindings.cpp
@@ -137,20 +137,20 @@ void edges2verts_scalar_lib_dp (const double* p_edge_in, const int* vert_edge_id
const double* v_int, double* p_vert_out, const int i_startblk, const int i_endblk,
const int i_startidx_in, const int i_endidx_in, const int slev,
const int elev, const int nproma, const int nlev, const int nblks_e,
- const int nblks_v, const int cell_type, const bool lacc){
+ const int nblks_v, const bool lacc){
edges2verts_scalar_lib<double>(p_edge_in, vert_edge_idx, vert_edge_blk, v_int,p_vert_out,i_startblk,
- i_endblk,i_startidx_in,i_endidx_in, slev, elev, nproma, nlev,nblks_e, nblks_v, cell_type,lacc);
+ i_endblk,i_startidx_in,i_endidx_in, slev, elev, nproma, nlev,nblks_e, nblks_v, lacc);
}
void edges2verts_scalar_lib_sp(const float* p_edge_in, const int* vert_edge_idx, const int* vert_edge_blk,
const float* v_int, float* p_vert_out, const int i_startblk, const int i_endblk,
const int i_startidx_in, const int i_endidx_in, const int slev,
const int elev, const int nproma, const int nlev, const int nblks_e,
- const int nblks_v, const int cell_type, const bool lacc){
+ const int nblks_v, const bool lacc){
edges2verts_scalar_lib<float>(p_edge_in, vert_edge_idx, vert_edge_blk, v_int,p_vert_out,i_startblk,
- i_endblk,i_startidx_in,i_endidx_in, slev, elev, nproma, nlev,nblks_e, nblks_v, cell_type,lacc);
+ i_endblk,i_startidx_in,i_endidx_in, slev, elev, nproma, nlev,nblks_e, nblks_v, lacc);
}
diff --git a/src/interpolation/interpolation_bindings.h b/src/interpolation/interpolation_bindings.h
index 123c487..14c1f34 100644
--- a/src/interpolation/interpolation_bindings.h
+++ b/src/interpolation/interpolation_bindings.h
@@ -78,14 +78,14 @@ extern "C"{
const double* v_int, double* p_vert_out, const int i_startblk, const int i_endblk,
const int i_startidx_in, const int i_endidx_in, const int slev,
const int elev, const int nproma, const int nlev, const int nblks_e,
- const int nblks_v, const int cell_type, const bool lacc);
+ const int nblks_v, const bool lacc);
void edges2verts_scalar_lib_sp (const float* p_edge_in, const int* vert_edge_idx, const int* vert_edge_blk,
const float* v_int, float* p_vert_out, const int i_startblk, const int i_endblk,
const int i_startidx_in, const int i_endidx_in, const int slev,
const int elev, const int nproma, const int nlev, const int nblks_e,
- const int nblks_v, const int cell_type, const bool lacc);
+ const int nblks_v, const bool lacc);
void edges2cells_scalar_lib_dp(
const double* p_edge_in, const int* edge_idx, const int* edge_blk,
@@ -183,4 +183,4 @@ extern "C"{
const int nlev, const int nblks_c,
const bool lacc);
-}
\ No newline at end of file
+}
diff --git a/src/interpolation/mo_lib_interpolation_scalar.cpp b/src/interpolation/mo_lib_interpolation_scalar.cpp
index c59f928..07403df 100644
--- a/src/interpolation/mo_lib_interpolation_scalar.cpp
+++ b/src/interpolation/mo_lib_interpolation_scalar.cpp
@@ -162,7 +162,7 @@ void edges2verts_scalar_lib (const T* p_edge_in, const int* vert_edge_idx, const
const T* v_int, T* p_vert_out, const int i_startblk, const int i_endblk,
const int i_startidx_in, const int i_endidx_in, const int slev,
const int elev, const int nproma, const int nlev, const int nblks_e,
- const int nblks_v, const int cell_type, const bool lacc){
+ const int nblks_v, const bool lacc){
// Wrap raw pointers in unmanaged Kokkos Views.
typedef Kokkos::View<const T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstT3D;
@@ -172,8 +172,8 @@ void edges2verts_scalar_lib (const T* p_edge_in, const int* vert_edge_idx, const
UnmanagedConstT3D p_edge_in_view (p_edge_in, nproma, nlev, nblks_e);
UnmanagedConstInt3D iidx_view (vert_edge_idx, nproma, nblks_v, 5);
UnmanagedConstInt3D iblk_view (vert_edge_blk, nproma,nblks_v, 5);
- UnmanagedConstT3D v_int_view (v_int, nproma,cell_type,nblks_v);
UnmanagedT3D p_vert_out_view (p_vert_out, nproma,nlev,nblks_v);
+ UnmanagedConstT3D v_int_view (v_int, nproma, 6, nblks_v);
for (int jb = i_startblk; jb < i_endblk + 1; ++jb) {
@@ -538,14 +538,14 @@ void edges2verts_scalar_lib<double> (const double* p_edge_in, const int* vert_ed
const double* v_int, double* p_vert_out, const int i_startblk, const int i_endblk,
const int i_startidx_in, const int i_endidx_in, const int slev,
const int elev, const int nproma, const int nlev, const int nblks_e,
- const int nblks_v, const int cell_type, const bool lacc);
+ const int nblks_v, const bool lacc);
template
void edges2verts_scalar_lib <float> (const float* p_edge_in, const int* vert_edge_idx, const int* vert_edge_blk,
const float* v_int, float* p_vert_out, const int i_startblk, const int i_endblk,
const int i_startidx_in, const int i_endidx_in, const int slev,
const int elev, const int nproma, const int nlev, const int nblks_e,
- const int nblks_v, const int cell_type, const bool lacc);
+ const int nblks_v, const bool lacc);
template
diff --git a/src/interpolation/mo_lib_interpolation_scalar.hpp b/src/interpolation/mo_lib_interpolation_scalar.hpp
index 16223ec..78f7df4 100644
--- a/src/interpolation/mo_lib_interpolation_scalar.hpp
+++ b/src/interpolation/mo_lib_interpolation_scalar.hpp
@@ -34,7 +34,7 @@ void edges2verts_scalar_lib (const T* p_edge_in, const int* vert_edge_idx, const
const T* v_int, T* p_vert_out, const int i_startblk, const int i_endblk,
const int i_startidx_in, const int i_endidx_in, const int slev,
const int elev, const int nproma, const int nlev, const int nblks_e,
- const int nblks_v, const int cell_type, const bool lacc);
+ const int nblks_v, const bool lacc);
template <typename T>
void edges2cells_scalar_lib(const T* p_edge_in, const int* edge_idx, const int* edge_blk,
@@ -71,4 +71,4 @@ void cell_avg_lib(const T* psi_c, const int* cell_neighbor_idx, const int* cell_
const T* avg_coeff, T* avg_psi_c, const int i_startblk, const int i_endblk,
const int i_startidx_in, const int i_endidx_in,
const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const bool lacc);
\ No newline at end of file
+ const int nlev, const int nblks_c, const bool lacc);
diff --git a/test/c/test_interpolation_scalar.cpp b/test/c/test_interpolation_scalar.cpp
index f46c2b4..bad8277 100644
--- a/test/c/test_interpolation_scalar.cpp
+++ b/test/c/test_interpolation_scalar.cpp
@@ -175,7 +175,6 @@ TYPED_TEST(InterpolationScalarTypedTestFixture, Edges2Verts) {
this->nlev,
this->nblks_e,
this->nblks_v,
- this->cell_type,
this->lacc);
// Check the outputs only for blocks in the range
--
GitLab
From be37e502bacf1c9b542e431a0b25e8e25f297e79 Mon Sep 17 00:00:00 2001
From: Pradipta Samanta <samanta@dkrz.de>
Date: Tue, 4 Mar 2025 16:51:34 +0100
Subject: [PATCH 27/36] corrected dimensions for some of the arrays
---
src/interpolation/mo_lib_interpolation_scalar.cpp | 12 ++++++------
1 file changed, 6 insertions(+), 6 deletions(-)
diff --git a/src/interpolation/mo_lib_interpolation_scalar.cpp b/src/interpolation/mo_lib_interpolation_scalar.cpp
index 07403df..3ed58cb 100644
--- a/src/interpolation/mo_lib_interpolation_scalar.cpp
+++ b/src/interpolation/mo_lib_interpolation_scalar.cpp
@@ -170,10 +170,10 @@ void edges2verts_scalar_lib (const T* p_edge_in, const int* vert_edge_idx, const
typedef Kokkos::View<const int***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstInt3D;
UnmanagedConstT3D p_edge_in_view (p_edge_in, nproma, nlev, nblks_e);
- UnmanagedConstInt3D iidx_view (vert_edge_idx, nproma, nblks_v, 5);
- UnmanagedConstInt3D iblk_view (vert_edge_blk, nproma,nblks_v, 5);
- UnmanagedT3D p_vert_out_view (p_vert_out, nproma,nlev,nblks_v);
+ UnmanagedConstInt3D iidx_view (vert_edge_idx, nproma, nblks_v, 6);
+ UnmanagedConstInt3D iblk_view (vert_edge_blk, nproma,nblks_v, 6);
UnmanagedConstT3D v_int_view (v_int, nproma, 6, nblks_v);
+ UnmanagedT3D p_vert_out_view (p_vert_out, nproma, nlev, nblks_v);
for (int jb = i_startblk; jb < i_endblk + 1; ++jb) {
@@ -283,7 +283,7 @@ void cells2verts_scalar_lib(
UnmanagedConstInt3D iblk_view(vert_cell_blk, nproma, nblks_v, 6); // vert_cell_blk_view
// coefficients for interpolation, dim: (nproma,9-cell_type,nblks_v)
- UnmanagedConstS3D coeff_int_view(coeff_int, nproma, 9, nblks_v);
+ UnmanagedConstS3D coeff_int_view(coeff_int, nproma, 6, nblks_v);
// vertex based scalar output field, dim: (nproma,nlev,nblks_c)
UnmanagedS3D p_vert_out_view(p_vert_out, nproma, nlev, nblks_c);
@@ -342,7 +342,7 @@ void cells2verts_scalar_ri_lib(
UnmanagedConstInt3D iblk_view(vert_cell_blk, nproma, nblks_v, 6); // vert_cell_blk_view
// coefficients for interpolation, dim: (nproma,9-cell_type,nblks_v)
- UnmanagedConstT3D coeff_int_view(coeff_int, nproma, 9, nblks_v);
+ UnmanagedConstT3D coeff_int_view(coeff_int, nproma, 6, nblks_v);
// vertex based scalar output field, dim: (nproma,nlev,nblks_c)
UnmanagedS3D p_vert_out_view(p_vert_out, nproma, nlev, nblks_c);
@@ -398,7 +398,7 @@ void verts2cells_scalar_lib(
// block indices of vertices of triangles, dim: (nproma,nblks_c, 3)
UnmanagedConstInt3D iblk_view(cell_vertex_blk, nproma, nblks_c, 3); // cell_vertex_blk
- // coefficients for interpolation, dim: (nproma,3-cell_type,nblks_c)
+ // coefficients for interpolation, dim: (nproma, 3, nblks_c)
UnmanagedConstT3D coeff_int_view(coeff_int, nproma, 3, nblks_c);
// vertex based scalar output field, dim: (nproma,nlev,nblks_c)
--
GitLab
From e73d605403b5557f7203f32aaaf17b0bac8c6d63 Mon Sep 17 00:00:00 2001
From: Pradipta Samanta <samanta@dkrz.de>
Date: Tue, 4 Mar 2025 16:52:46 +0100
Subject: [PATCH 28/36] added few more tests for mo_interpolation_scalar
---
test/c/test_interpolation_scalar.cpp | 419 +++++++++++----------------
1 file changed, 175 insertions(+), 244 deletions(-)
diff --git a/test/c/test_interpolation_scalar.cpp b/test/c/test_interpolation_scalar.cpp
index bad8277..2307cb8 100644
--- a/test/c/test_interpolation_scalar.cpp
+++ b/test/c/test_interpolation_scalar.cpp
@@ -72,27 +72,49 @@ class InterpolationScalarTypedTestFixture : public ::testing::Test, public inter
public:
// Arrays used for verts2edges
- std::vector<T> p_vertex_in; // Dimensions: (nproma, nlev, nblks_v)
- std::vector<int> edge_vertex_idx; // Dimensions: (nproma, nblks_e, 4)
- std::vector<int> edge_vertex_blk; // Dimensions: (nproma, nblks_e, 4)
- std::vector<T> coeff_int; // Dimensions: (nproma, 2, nblks_e)
+ std::vector<T> p_vertex_in; // Dimensions: (nproma, nlev, nblks_v)
+ std::vector<int> edge_vertex_idx; // Dimensions: (nproma, nblks_e, 4)
+ std::vector<int> edge_vertex_blk; // Dimensions: (nproma, nblks_e, 4)
+ std::vector<T> coeff_int_edges; // Dimensions: (nproma, 2, nblks_e)
std::vector<T> p_edge_out; // Dimensions: (nproma, nlev, nblks_e)
// Arrays used for edges2verts
- std::vector<T> p_edge_in; // Dimensions: (nproma, nlev, nblks_e)
- std::vector<int> edge_vert_idx; // Dimensions: (nproma, nblks_e, 6)
- std::vector<int> edge_vert_blk; // Dimensions: (nproma, nblks_e, 6)
- std::vector<T> v_int; // Dimensions: (nproma, 6-cell_type, nblks_v)
+ std::vector<T> p_edge_in; // Dimensions: (nproma, nlev, nblks_e)
+ std::vector<int> edge_vert_idx; // Dimensions: (nproma, nblks_e, 6)
+ std::vector<int> edge_vert_blk; // Dimensions: (nproma, nblks_e, 6)
+ std::vector<T> v_int; // Dimensions: (nproma, 6, nblks_v)
std::vector<T> p_vert_out; // Dimensions: (nproma, nlev, nblks_v)
- //
+
+ // Arrays used for edges2cells
+ // std::vector<T> p_edge_in; // Dimensions: (nproma, nlev, nblks_e)
+ std::vector<int> edge_idx; // Dimensions: (nproma, nblks_c, 3)
+ std::vector<int> edge_blk; // Dimensions: (nproma, nblks_c, 3)
+ std::vector<T> coeff_int_cells; // Dimensions: (nproma, 3, nblks_c)
+ std::vector<T> p_cell_out; // Dimensions: (nproma, nlev, nblks_c)
+
+ // Arrays used for verts2cells
+ std::vector<T> p_vert_in; // Dimensions: (nproma, nlev, nblks_v)
+ std::vector<int> cell_index_idx; // Dimensions: (nproma, nblks_c, 3)
+ std::vector<int> cell_index_blk; // Dimensions: (nproma, nblks_c, 3)
+ // std::vector<InType> coeff_int; // Dimensions: (nproma, 3, nblks_c)
+ // std::vector<T> p_cell_out; // Dimensions: (nproma, nlev, nblks_c)
+
+ // Arrays used for avg_lib
+ std::vector<T> psi_c; // Dimensions: (nproma, nlev, nblks_c)
+ std::vector<int> cell_neighbor_idx; // Dimensions: (nproma, nblks_c, 3)
+ std::vector<int> cell_neighbor_blk; // Dimensions: (nproma, nblks_c, 3)
+ std::vector<T> avg_coeff; // Dimensions: (nproma, nlev, nblks_c)
+ std::vector<T> avg_psi_c; // Dimensions: (nproma, nlev, nblks_c)
+
const int cell_type = 6;
+ const int npromz_c = 32;
InterpolationScalarTypedTestFixture() {
// Allocate and initialize arrays needed for verts2edges
p_vertex_in.resize(num_elements_3d<T>(nproma, nlev, nblks_v), static_cast<T>(1));
edge_vertex_idx.resize(num_elements_3d<int>(nproma, nblks_e, 4), 1);
edge_vertex_blk.resize(num_elements_3d<int>(nproma, nblks_e, 4), 0);
- coeff_int.resize(num_elements_3d<T>(nproma, 2, nblks_v), static_cast<T>(1));
+ coeff_int_edges.resize(num_elements_3d<T>(nproma, 2, nblks_e), static_cast<T>(1));
p_edge_out.resize(num_elements_3d<T>(nproma, nlev, nblks_e), static_cast<T>(0));
@@ -103,6 +125,33 @@ public:
v_int.resize(num_elements_3d<T>(nproma, 6, nblks_v), static_cast<T>(1));
p_vert_out.resize(num_elements_3d<T>(nproma, nlev, nblks_v), static_cast<T>(0));
+
+ // Allocate & Initialize arrays needed for edges2cells
+ // p_edge_in.resize(num_elements_3d<T>(nproma, nlev, nblks_e), static_cast<T>(1));
+ edge_idx.resize(num_elements_3d<int>(nproma, nblks_c, 3), 1);
+ edge_blk.resize(num_elements_3d<int>(nproma, nblks_c, 3), 0);
+ coeff_int_cells.resize(num_elements_3d<T>(nproma, 3, nblks_c), static_cast<T>(1));
+
+ p_cell_out.resize(num_elements_3d<T>(nproma, nlev, nblks_c), static_cast<T>(0));
+
+ // Allocate and initialize arrays needed for verts2cells
+ p_vert_in.resize(num_elements_3d<T>(nproma, nlev, nblks_v), static_cast<T>(1));
+ cell_index_idx.resize(num_elements_3d<int>(nproma, nblks_c, 3), 1);
+ cell_index_blk.resize(num_elements_3d<int>(nproma, nblks_c, 3), 0);
+ // coeff_int.resize(num_elements_3d<T>(nproma, cell_type, nblks_c), static_cast<T>(1));
+
+ // Allocate output arrays and initialize to zero.
+ // p_cell_out.resize(num_elements_3d<T>(nproma, nlev, nblks_c), static_cast<T>(0));
+
+ // Allocate and initialize arrays needed for avg_lib
+ psi_c.resize(num_elements_3d<T>(nproma, nlev, nblks_c), static_cast<T>(1));
+ cell_neighbor_idx.resize(num_elements_3d<int>(nproma, nblks_c, 3), 1);
+ cell_neighbor_blk.resize(num_elements_3d<int>(nproma, nblks_c, 3), 0);
+ avg_coeff.resize(num_elements_3d<T>(nproma, nlev, nblks_c), static_cast<T>(1));
+
+ // Allocate output arrays and initialize to zero.
+ avg_psi_c.resize(num_elements_3d<T>(nproma, nlev, nblks_c), static_cast<T>(0));
+
}
};
@@ -122,7 +171,7 @@ TYPED_TEST(InterpolationScalarTypedTestFixture, Verts2Edges) {
this->p_vertex_in.data(),
this->edge_vertex_idx.data(),
this->edge_vertex_blk.data(),
- this->coeff_int.data(),
+ this->coeff_int_edges.data(),
this->p_edge_out.data(),
this->i_startblk,
this->i_endblk,
@@ -194,11 +243,125 @@ TYPED_TEST(InterpolationScalarTypedTestFixture, Edges2Verts) {
////////////////////////////////////////////////////////////////////////////////
//
-// ! cells2edges
+// ! edges2cells
+//
+////////////////////////////////////////////////////////////////////////////////
+
+TYPED_TEST(InterpolationScalarTypedTestFixture, Edges2Cells) {
+
+ edges2cells_scalar_lib<TypeParam>(
+ this->p_edge_in.data(),
+ this->edge_idx.data(),
+ this->edge_blk.data(),
+ this->coeff_int_cells.data(),
+ this->p_cell_out.data(),
+ this->i_startblk,
+ this->i_endblk,
+ this->i_startidx,
+ this->i_endidx,
+ this->slev,
+ this->elev,
+ this->nproma,
+ this->nlev,
+ this->nblks_e,
+ this->nblks_c,
+ this->lacc);
+
+ // Check the outputs only for blocks in the range
+ // { [i_startblk, i_endblk], [slev,elev], [i_startidx_in, i_endidx_in] }
+ for (int block = this->i_startblk; block <= this->i_endblk; ++block) {
+ for (int level = this->slev; level < this->elev; ++level) {
+ for (int i = this->i_startidx; i < this->i_endidx; ++i) {
+ // Compute the linear index for a 3D array in column-major order:
+ size_t idx = i + level * this->nproma + block * this->nproma * this->nlev;
+ // Since every contribution is 1 and there are 3 stencil points, expect 3.
+ EXPECT_NEAR(this->p_cell_out[idx], static_cast<TypeParam>(3), static_cast<TypeParam>(1e-5))
+ << "Failure at block " << block << ", level " << level << ", index " << i;
+ }
+ }
+ }
+}
+
+TYPED_TEST(InterpolationScalarTypedTestFixture, Verts2Cells) {
+
+ verts2cells_scalar_lib<TypeParam>(
+ this->p_vert_in.data(),
+ this->cell_index_idx.data(),
+ this->cell_index_blk.data(),
+ this->coeff_int_cells.data(),
+ this->p_cell_out.data(),
+ this->nblks_c,
+ this->npromz_c,
+ this->slev,
+ this->elev,
+ this->nproma,
+ this->nlev,
+ this->nblks_v,
+ this->lacc);
+
+
+ // Check the outputs only for blocks in the range
+ // { [i_startblk, i_endblk], [slev,elev], [i_startidx_in, i_endidx_in] }
+ for (int block = this->i_startblk; block <= this->i_endblk; ++block) {
+ for (int level = this->slev; level < this->elev; ++level) {
+ for (int i = this->i_startidx; i < this->i_endidx; ++i) {
+ // Compute the linear index for a 3D array in column-major order:
+ size_t idx = i + level * this->nproma + block * this->nproma * this->nlev;
+ // Since every contribution is 1 and there are 3 stencil points, expect 3.
+ EXPECT_NEAR(this->p_cell_out[idx], static_cast<TypeParam>(3), static_cast<TypeParam>(1e-5))
+ << "Failure at block " << block << ", level " << level << ", index " << i;
+ }
+ }
+ }
+}
+
+////////////////////////////////////////////////////////////////////////////////
+//
+// ! cell_avg
//
////////////////////////////////////////////////////////////////////////////////
+TYPED_TEST(InterpolationScalarTypedTestFixture, AvgLib) {
+ // Call the function
+ cell_avg_lib<TypeParam>(
+ this->psi_c.data(),
+ this->cell_neighbor_idx.data(),
+ this->cell_neighbor_blk.data(),
+ this->avg_coeff.data(),
+ this->avg_psi_c.data(),
+ this->i_startblk,
+ this->i_endblk,
+ this->i_startidx,
+ this->i_endidx,
+ this->slev,
+ this->elev,
+ this->nproma,
+ this->nlev,
+ this->nblks_c,
+ this->lacc);
+
+
+ // Check the outputs only for blocks in the range
+ // { [i_startblk, i_endblk], [slev,elev], [i_startidx_in, i_endidx_in] }
+ for (int block = this->i_startblk; block <= this->i_endblk; ++block) {
+ for (int level = this->slev; level < this->elev; ++level) {
+ for (int i = this->i_startidx; i < this->i_endidx; ++i) {
+ // Compute the linear index for a 3D array in column-major order:
+ size_t idx = i + level * this->nproma + block * this->nproma * this->nlev;
+ // Since every contribution is 1 and there are 4 stencil points, expect 4.
+ EXPECT_NEAR(this->avg_psi_c[idx], static_cast<TypeParam>(4), static_cast<TypeParam>(1e-5))
+ << "Failure at block " << block << ", level " << level << ", index " << i;
+ }
+ }
+ }
+}
+
+////////////////////////////////////////////////////////////////////////////////
+//
+// ! cells2edges
+//
+////////////////////////////////////////////////////////////////////////////////
// Define a typed test fixture with one precision of data
template <typename Types>
@@ -287,85 +450,6 @@ TYPED_TEST(Cells2edgesScalarLibTestFixture, cells2edges) {
}
-
-
-////////////////////////////////////////////////////////////////////////////////
-//
-// ! edges2cells
-//
-////////////////////////////////////////////////////////////////////////////////
-
-
-// Define a typed test fixture with one precision of data
-template <typename Types>
-class Edges2cellsScalarLibTestFixture : public testing::Test, public interp_dimensions{
-public:
- using InType = typename Types::in_type;
- using OutType = typename Types::out_type;
-
- // Arrays stored in std::vector.
- std::vector<InType> p_edge_in; // Dimensions: (nproma, nlev, nblks_e)
- std::vector<int> edge_idx; // Dimensions: (nproma, nblks_c, 3)
- std::vector<int> edge_blk; // Dimensions: (nproma, nblks_c, 3)
- std::vector<OutType> coeff_int; // Dimensions: (nproma, 3-cell_type, nblks_c)
- std::vector<OutType> p_cell_out; // Dimensions: (nproma, nlev, nblks_c)
-
- const int cell_type = 3;
-
- Edges2cellsScalarLibTestFixture() {
- // Allocate and initialize inputs.
- p_edge_in.resize(num_elements_3d<InType>(nproma, nlev, nblks_e), static_cast<InType>(1));
- edge_idx.resize(num_elements_3d<int>(nproma, nblks_c, cell_type), 1);
- edge_blk.resize(num_elements_3d<int>(nproma, nblks_c, cell_type), 0);
- coeff_int.resize(num_elements_3d<InType>(nproma, cell_type, nblks_c), static_cast<OutType>(1));
-
- // Allocate output arrays and initialize to zero.
- p_cell_out.resize(num_elements_3d<OutType>(nproma, nlev, nblks_c), static_cast<OutType>(0));
- }
-};
-
-// Add test suite
-TYPED_TEST_SUITE(Edges2cellsScalarLibTestFixture, OneTypes);
-
-// Add test
-TYPED_TEST(Edges2cellsScalarLibTestFixture, cells2edges) {
- using InType = typename TestFixture::InType;
- using OutType = typename TestFixture::OutType;
-
- // Call the function
- edges2cells_scalar_lib<InType>(
- this->p_edge_in.data(),
- this->edge_idx.data(),
- this->edge_blk.data(),
- this->coeff_int.data(),
- this->p_cell_out.data(),
- this->i_startblk,
- this->i_endblk,
- this->i_startidx,
- this->i_endidx,
- this->slev,
- this->elev,
- this->nproma,
- this->nlev,
- this->nblks_e,
- this->nblks_c,
- this->lacc);
-
- // Check the outputs only for blocks in the range
- // { [i_startblk, i_endblk], [slev,elev], [i_startidx_in, i_endidx_in] }
- for (int block = this->i_startblk; block <= this->i_endblk; ++block) {
- for (int level = this->slev; level < this->elev; ++level) {
- for (int i = this->i_startidx; i < this->i_endidx; ++i) {
- // Compute the linear index for a 3D array in column-major order:
- size_t idx = i + level * this->nproma + block * this->nproma * this->nlev;
- // Since every contribution is 1 and there are 3 stencil points, expect 3.
- EXPECT_NEAR(this->p_cell_out[idx], static_cast<OutType>(3), static_cast<OutType>(1e-5))
- << "Failure at block " << block << ", level " << level << ", index " << i;
- }
- }
- }
-}
-
////////////////////////////////////////////////////////////////////////////////
//
// ! cells2verts
@@ -521,156 +605,3 @@ TYPED_TEST(Cells2vertsriScalarLibTestFixture, cells2edges) {
}
}
}
-
-
-////////////////////////////////////////////////////////////////////////////////
-//
-// ! verts2cells
-//
-////////////////////////////////////////////////////////////////////////////////
-
-
-// Define a typed test fixture with one precision of data
-template <typename Types>
-class Verts2cellsScalarLibTestFixture : public testing::Test, public interp_dimensions{
-public:
- using InType = typename Types::in_type;
- using OutType = typename Types::out_type;
-
- // Arrays stored in std::vector.
- std::vector<InType> p_vert_in; // Dimensions: (nproma, nlev, nblks_v)
- std::vector<int> cell_index_idx; // Dimensions: (nproma, nblks_c, 3)
- std::vector<int> cell_index_blk; // Dimensions: (nproma, nblks_c, 3)
- std::vector<InType> coeff_int; // Dimensions: (nproma, 3-cell_type, nblks_c)
- std::vector<OutType> p_cell_out; // Dimensions: (nproma, nlev, nblks_c)
-
- const int cell_type = 3;
- const int npromz_c = 32;
-
- Verts2cellsScalarLibTestFixture() {
- // Allocate and initialize inputs.
- p_vert_in.resize(num_elements_3d<InType>(nproma, nlev, nblks_v), static_cast<InType>(1));
- cell_index_idx.resize(num_elements_3d<int>(nproma, nblks_c, 3), 1);
- cell_index_blk.resize(num_elements_3d<int>(nproma, nblks_c, 3), 0);
- coeff_int.resize(num_elements_3d<InType>(nproma, cell_type, nblks_c), static_cast<InType>(1));
-
- // Allocate output arrays and initialize to zero.
- p_cell_out.resize(num_elements_3d<OutType>(nproma, nlev, nblks_c), static_cast<OutType>(0));
- }
-};
-
-// Add test suite
-TYPED_TEST_SUITE(Verts2cellsScalarLibTestFixture, OneTypes);
-
-// Add test
-TYPED_TEST(Verts2cellsScalarLibTestFixture, cells2edges) {
- using InType = typename TestFixture::InType;
- using OutType = typename TestFixture::OutType;
-
- // Call the function
- verts2cells_scalar_lib<InType>(
- this->p_vert_in.data(),
- this->cell_index_idx.data(),
- this->cell_index_blk.data(),
- this->coeff_int.data(),
- this->p_cell_out.data(),
- this->nblks_c,
- this->npromz_c,
- this->slev,
- this->elev,
- this->nproma,
- this->nlev,
- this->nblks_v,
- this->lacc);
-
-
- // Check the outputs only for blocks in the range
- // { [i_startblk, i_endblk], [slev,elev], [i_startidx_in, i_endidx_in] }
- for (int block = this->i_startblk; block <= this->i_endblk; ++block) {
- for (int level = this->slev; level < this->elev; ++level) {
- for (int i = this->i_startidx; i < this->i_endidx; ++i) {
- // Compute the linear index for a 3D array in column-major order:
- size_t idx = i + level * this->nproma + block * this->nproma * this->nlev;
- // Since every contribution is 1 and there are 3 stencil points, expect 3.
- EXPECT_NEAR(this->p_cell_out[idx], static_cast<OutType>(3), static_cast<OutType>(1e-5))
- << "Failure at block " << block << ", level " << level << ", index " << i;
- }
- }
- }
-}
-
-
-////////////////////////////////////////////////////////////////////////////////
-//
-// ! cell_avg
-//
-////////////////////////////////////////////////////////////////////////////////
-
-
-// Define a typed test fixture with one precision of data
-template <typename Types>
-class CellAvgScalarLibTestFixture : public testing::Test, public interp_dimensions{
-public:
- using InType = typename Types::in_type;
- using OutType = typename Types::out_type;
-
- // Arrays stored in std::vector.
- std::vector<InType> psi_c; // Dimensions: (nproma, nlev, nblks_c)
- std::vector<int> cell_neighbor_idx; // Dimensions: (nproma, nblks_c, 3)
- std::vector<int> cell_neighbor_blk; // Dimensions: (nproma, nblks_c, 3)
- std::vector<InType> avg_coeff; // Dimensions: (nproma, nlev, nblks_c)
- std::vector<OutType> avg_psi_c; // Dimensions: (nproma, nlev, nblks_c)
-
- CellAvgScalarLibTestFixture() {
- // Allocate and initialize inputs.
- psi_c.resize(num_elements_3d<InType>(nproma, nlev, nblks_c), static_cast<InType>(1));
- cell_neighbor_idx.resize(num_elements_3d<int>(nproma, nblks_c, 3), 1);
- cell_neighbor_blk.resize(num_elements_3d<int>(nproma, nblks_c, 3), 0);
- avg_coeff.resize(num_elements_3d<InType>(nproma, nlev, nblks_c), static_cast<InType>(1));
-
- // Allocate output arrays and initialize to zero.
- avg_psi_c.resize(num_elements_3d<OutType>(nproma, nlev, nblks_c), static_cast<OutType>(0));
- }
-};
-
-// Add test suite
-TYPED_TEST_SUITE(CellAvgScalarLibTestFixture, OneTypes);
-
-// Add test
-TYPED_TEST(CellAvgScalarLibTestFixture, cells2edges) {
- using InType = typename TestFixture::InType;
- using OutType = typename TestFixture::OutType;
-
- // Call the function
- cell_avg_lib<InType>(
- this->psi_c.data(),
- this->cell_neighbor_idx.data(),
- this->cell_neighbor_blk.data(),
- this->avg_coeff.data(),
- this->avg_psi_c.data(),
- this->i_startblk,
- this->i_endblk,
- this->i_startidx,
- this->i_endidx,
- this->slev,
- this->elev,
- this->nproma,
- this->nlev,
- this->nblks_c,
- this->lacc);
-
-
- // Check the outputs only for blocks in the range
- // { [i_startblk, i_endblk], [slev,elev], [i_startidx_in, i_endidx_in] }
- for (int block = this->i_startblk; block <= this->i_endblk; ++block) {
- for (int level = this->slev; level < this->elev; ++level) {
- for (int i = this->i_startidx; i < this->i_endidx; ++i) {
- // Compute the linear index for a 3D array in column-major order:
- size_t idx = i + level * this->nproma + block * this->nproma * this->nlev;
- // Since every contribution is 1 and there are 4 stencil points, expect 4.
- EXPECT_NEAR(this->avg_psi_c[idx], static_cast<OutType>(4), static_cast<OutType>(1e-5))
- << "Failure at block " << block << ", level " << level << ", index " << i;
- }
- }
- }
-}
--
GitLab
From c6fc4504549773608f791869155da741a2bcb7c0 Mon Sep 17 00:00:00 2001
From: Pradipta Samanta <samanta@dkrz.de>
Date: Wed, 5 Mar 2025 13:27:04 +0100
Subject: [PATCH 29/36] added the use of loop_exchange in
cells2verts_ri_scalar_lib
---
src/interpolation/mo_lib_interpolation_scalar.cpp | 8 ++++++++
test/c/CMakeLists.txt | 4 ++++
test/c/test_interpolation_scalar.cpp | 4 ++++
3 files changed, 16 insertions(+)
diff --git a/src/interpolation/mo_lib_interpolation_scalar.cpp b/src/interpolation/mo_lib_interpolation_scalar.cpp
index 3ed58cb..6372870 100644
--- a/src/interpolation/mo_lib_interpolation_scalar.cpp
+++ b/src/interpolation/mo_lib_interpolation_scalar.cpp
@@ -345,7 +345,11 @@ void cells2verts_scalar_ri_lib(
UnmanagedConstT3D coeff_int_view(coeff_int, nproma, 6, nblks_v);
// vertex based scalar output field, dim: (nproma,nlev,nblks_c)
+#ifdef __LOOP_EXCHANGE
UnmanagedS3D p_vert_out_view(p_vert_out, nproma, nlev, nblks_c);
+#else
+ UnmanagedS3D p_vert_out_view(p_vert_out, nlev, nproma, nblks_c);
+#endif
int i_startidx, i_endidx;
@@ -360,7 +364,11 @@ void cells2verts_scalar_ri_lib(
Kokkos::parallel_for("cells2verts_scalar_ri_lib", innerPolicy,
KOKKOS_LAMBDA(const int jk, const int jv) {
+#ifdef __LOOP_EXCHANGE
p_vert_out_view(jv, jk, jb) =
+#else
+ p_vert_out_view(jk, jv, jb) =
+#endif
coeff_int_view(jv, 0, jb)*p_cell_in_view(iidx_view(jv, jb, 0), jk, iblk_view(jv, jb, 0)) +
coeff_int_view(jv, 1, jb)*p_cell_in_view(iidx_view(jv, jb, 1), jk, iblk_view(jv, jb, 1)) +
coeff_int_view(jv, 2, jb)*p_cell_in_view(iidx_view(jv, jb, 2), jk, iblk_view(jv, jb, 2)) +
diff --git a/test/c/CMakeLists.txt b/test/c/CMakeLists.txt
index 16371d0..c9320cb 100644
--- a/test/c/CMakeLists.txt
+++ b/test/c/CMakeLists.txt
@@ -21,6 +21,10 @@ FetchContent_MakeAvailable(googletest)
# Find Kokkos (or use your existing Kokkos installation)
# find_package(Kokkos REQUIRED)
+if(IM_ENABLE_LOOP_EXCHANGE)
+ target_compile_definitions(iconmath-interpolation PRIVATE __LOOP_EXCHANGE)
+endif()
+
set(SOURCES
main.cpp
test_tdma_solver.cpp
diff --git a/test/c/test_interpolation_scalar.cpp b/test/c/test_interpolation_scalar.cpp
index 2307cb8..2e0be9c 100644
--- a/test/c/test_interpolation_scalar.cpp
+++ b/test/c/test_interpolation_scalar.cpp
@@ -597,7 +597,11 @@ TYPED_TEST(Cells2vertsriScalarLibTestFixture, cells2edges) {
for (int level = this->slev; level < this->elev; ++level) {
for (int i = this->i_startidx; i < this->i_endidx; ++i) {
// Compute the linear index for a 3D array in column-major order:
+#ifdef __LOOP_EXCHANGE
size_t idx = i + level * this->nproma + block * this->nproma * this->nlev;
+#else
+ size_t idx = level + i * this->nlev + block * this->nproma * this->nlev;
+#endif
// Since every contribution is 1 and there are 6 stencil points, expect 6.
EXPECT_NEAR(this->p_vert_out[idx], static_cast<OutType>(6), static_cast<OutType>(1e-5))
<< "Failure at block " << block << ", level " << level << ", index " << i;
--
GitLab
From b4c86e81158123f23645c8f949b1252bd13fc0d8 Mon Sep 17 00:00:00 2001
From: Pradipta Samanta <samanta@dkrz.de>
Date: Wed, 5 Mar 2025 13:28:28 +0100
Subject: [PATCH 30/36] modified the routine cells2verts_scalar_lib
---
src/interpolation/interpolation_bindings.cpp | 6 +++---
src/interpolation/mo_lib_interpolation_scalar.cpp | 8 ++++----
test/c/test_interpolation_scalar.cpp | 2 +-
3 files changed, 8 insertions(+), 8 deletions(-)
diff --git a/src/interpolation/interpolation_bindings.cpp b/src/interpolation/interpolation_bindings.cpp
index 0535091..d113952 100644
--- a/src/interpolation/interpolation_bindings.cpp
+++ b/src/interpolation/interpolation_bindings.cpp
@@ -209,14 +209,14 @@ void cells2verts_scalar_lib_dp(
}
void cells2verts_scalar_lib_dp2sp(
- const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
+ const float* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const double* coeff_int, double* p_vert_out, const int i_startblk, const int i_endblk,
const int i_startidx_in, const int i_endidx_in,
const int slev, const int elev, const int nproma,
const int nlev, const int nblks_c, const int nblks_v,
const bool lacc, const bool acc_async)
{
- cells2verts_scalar_lib<double,float>(
+ cells2verts_scalar_lib<float,double>(
p_cell_in, vert_cell_idx, vert_cell_blk,
coeff_int, p_vert_out, i_startblk, i_endblk,
i_startidx_in, i_endidx_in,
diff --git a/src/interpolation/mo_lib_interpolation_scalar.cpp b/src/interpolation/mo_lib_interpolation_scalar.cpp
index 6372870..26bd876 100644
--- a/src/interpolation/mo_lib_interpolation_scalar.cpp
+++ b/src/interpolation/mo_lib_interpolation_scalar.cpp
@@ -323,7 +323,7 @@ void cells2verts_scalar_ri_lib(
const T* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
const T* coeff_int, S* p_vert_out, const int i_startblk, const int i_endblk,
const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
+ const int slev, const int elev, const int nproma,
const int nlev, const int nblks_c, const int nblks_v,
const bool lacc, const bool acc_async)
{
@@ -585,9 +585,9 @@ void cells2verts_scalar_lib<double,double>(
const bool lacc, const bool acc_async);
template
-void cells2verts_scalar_lib<double,float>(
- const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
+void cells2verts_scalar_lib<float,double>(
+ const float* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
+ const double* coeff_int, double* p_vert_out, const int i_startblk, const int i_endblk,
const int i_startidx_in, const int i_endidx_in,
const int slev, const int elev, const int nproma,
const int nlev, const int nblks_c, const int nblks_v,
diff --git a/test/c/test_interpolation_scalar.cpp b/test/c/test_interpolation_scalar.cpp
index 2e0be9c..bd179c4 100644
--- a/test/c/test_interpolation_scalar.cpp
+++ b/test/c/test_interpolation_scalar.cpp
@@ -486,7 +486,7 @@ public:
};
// Add test suite
-TYPED_TEST_SUITE(Cells2vertsScalarLibTestFixture, MixedTypes);
+TYPED_TEST_SUITE(Cells2vertsScalarLibTestFixture, MixedTypesSP2DP);
// Add test
TYPED_TEST(Cells2vertsScalarLibTestFixture, cells2edges) {
--
GitLab
From aab9f0354f42d178893683cf4576af7008aac0a8 Mon Sep 17 00:00:00 2001
From: Pradipta Samanta <samanta@dkrz.de>
Date: Wed, 5 Mar 2025 15:38:51 +0100
Subject: [PATCH 31/36] added the last set of tests
---
test/c/test_interpolation_scalar.cpp | 134 ++++++++++-----------------
1 file changed, 51 insertions(+), 83 deletions(-)
diff --git a/test/c/test_interpolation_scalar.cpp b/test/c/test_interpolation_scalar.cpp
index bd179c4..6338a36 100644
--- a/test/c/test_interpolation_scalar.cpp
+++ b/test/c/test_interpolation_scalar.cpp
@@ -41,10 +41,6 @@ typedef ::testing::Types< MixedPrecision<double, double>,
MixedPrecision<float, double>,
MixedPrecision<float, float> > MixedTypesSP2DP;
-// Define the list of type pairs we want to test.
-typedef ::testing::Types< MixedPrecision<double, double>,
- MixedPrecision<float, float > > OneTypes;
-
// Shared dimensions for all routines and classes
class interp_dimensions {
@@ -357,67 +353,76 @@ TYPED_TEST(InterpolationScalarTypedTestFixture, AvgLib) {
}
}
-////////////////////////////////////////////////////////////////////////////////
-//
-// ! cells2edges
-//
-////////////////////////////////////////////////////////////////////////////////
-
-// Define a typed test fixture with one precision of data
-template <typename Types>
-class Cells2edgesScalarLibTestFixture : public testing::Test, public interp_dimensions{
+template <typename TypePair>
+class InterpolationScalarMixedTestFixture : public ::testing::Test, public interp_dimensions {
public:
- using InType = typename Types::in_type;
- using OutType = typename Types::out_type;
+ using InType = typename TypePair::in_type;
+ using OutType = typename TypePair::out_type;
- // Arrays stored in std::vector.
- std::vector<InType> p_cell_in; // Dimensions: (nproma, nlev, nblks_c)
- std::vector<int> edge_cell_idx; // Dimensions: (nproma, nblks_e, 2)
- std::vector<int> edge_cell_blk; // Dimensions: (nproma, nblks_e, 2)
- std::vector<OutType> coeff_int; // Dimensions: (nproma, 2, nblks_e)
- std::vector<OutType> p_edge_out; // Dimensions: (nproma, nlev, nblks_e)
+ // Arrays used for cells2edges
+ std::vector<InType> p_cell_in; // Dimensions: (nproma, nlev, nblks_c)
+ std::vector<int> edge_cell_idx; // Dimensions: (nproma, nblks_e, 2)
+ std::vector<int> edge_cell_blk; // Dimensions: (nproma, nblks_e, 2)
+ std::vector<OutType> coeff_int_edges; // Dimensions: (nproma, 2, nblks_e)
+ std::vector<OutType> p_edge_out; // Dimensions: (nproma, nlev, nblks_e)
- // TODO review these values
+ // Further parameters for cells2edges
const int patch_id = 0;
const bool l_limited_area = false;
const bool lfill_latbc = false;
- std::vector<int> i_startblk_in; // Dimensions: (2)
- std::vector<int> i_endblk_in; // Dimensions: (2)
- std::vector<int> i_startidx_in; // Dimensions: (2)
- std::vector<int> i_endidx_in; // Dimensions: (2)
-
- Cells2edgesScalarLibTestFixture() {
- // Allocate and initialize inputs.
+ std::vector<int> i_startblk_in; // Dimensions: (2)
+ std::vector<int> i_endblk_in; // Dimensions: (2)
+ std::vector<int> i_startidx_in; // Dimensions: (2)
+ std::vector<int> i_endidx_in; // Dimensions: (2)
+
+ // Arrays used for cells2verts
+ std::vector<int> vert_cell_idx; // Dimensions: (nproma, nblks_v, 6)
+ std::vector<int> vert_cell_blk; // Dimensions: (nproma, nblks_v, 6)
+ std::vector<OutType> coeff_int_verts; // Dimensions: (nproma, 6, nblks_v)
+ std::vector<OutType> p_vert_out; // Dimensions: (nproma, nlev, nblks_v)
+
+ InterpolationScalarMixedTestFixture() {
+ // Allocate and initialize arrays needed for cells2edges
p_cell_in.resize(num_elements_3d<InType>(nproma, nlev, nblks_c), static_cast<InType>(1));
edge_cell_idx.resize(num_elements_3d<int>(nproma, nblks_e, 2), 1);
edge_cell_blk.resize(num_elements_3d<int>(nproma, nblks_e, 2), 0);
- coeff_int.resize(num_elements_3d<InType>(nproma, 2, nblks_e), static_cast<OutType>(1));
+ coeff_int_edges.resize(num_elements_3d<InType>(nproma, 2, nblks_e), static_cast<OutType>(1));
- // Allocate output arrays and initialize to zero.
p_edge_out.resize(num_elements_3d<OutType>(nproma, nlev, nblks_e), static_cast<OutType>(0));
- // Allocate neighbour indexes
+ // Allocate neighbour indexes for cells2edges
i_startblk_in.resize(2, i_startblk);
i_endblk_in.resize(2, i_endblk);
i_startidx_in.resize(2, i_startidx);
i_endidx_in.resize(2, i_endidx);
+
+ // Allocate & Initialize arrays needed for cells2verts
+ vert_cell_idx.resize(num_elements_3d<int>(nproma, nblks_v, 6), 1);
+ vert_cell_blk.resize(num_elements_3d<int>(nproma, nblks_v, 6), 0);
+ coeff_int_verts.resize(num_elements_3d<InType>(nproma, 6, nblks_v), static_cast<OutType>(1));
+
+ p_vert_out.resize(num_elements_3d<OutType>(nproma, nlev, nblks_v), static_cast<OutType>(0));
}
};
-// Add test suite
-TYPED_TEST_SUITE(Cells2edgesScalarLibTestFixture, MixedTypesSP2DP);
+TYPED_TEST_SUITE(InterpolationScalarMixedTestFixture, MixedTypesSP2DP);
-// Add test
-TYPED_TEST(Cells2edgesScalarLibTestFixture, cells2edges) {
+////////////////////////////////////////////////////////////////////////////////
+//
+// ! cells2edges
+//
+////////////////////////////////////////////////////////////////////////////////
+
+TYPED_TEST(InterpolationScalarMixedTestFixture, cells2edges) {
using InType = typename TestFixture::InType;
- using OutType = typename TestFixture::OutType;
+ using OutType = typename TestFixture::OutType;
// Call the function
cells2edges_scalar_lib<InType,OutType>(
this->p_cell_in.data(),
this->edge_cell_idx.data(),
this->edge_cell_blk.data(),
- this->coeff_int.data(),
+ this->coeff_int_edges.data(),
this->p_edge_out.data(),
this->i_startblk_in.data(),
this->i_endblk_in.data(),
@@ -449,56 +454,21 @@ TYPED_TEST(Cells2edgesScalarLibTestFixture, cells2edges) {
}
}
-
////////////////////////////////////////////////////////////////////////////////
//
// ! cells2verts
//
////////////////////////////////////////////////////////////////////////////////
-
-// Define a typed test fixture with one precision of data
-template <typename Types>
-class Cells2vertsScalarLibTestFixture : public testing::Test, public interp_dimensions{
-public:
- using InType = typename Types::in_type;
- using OutType = typename Types::out_type;
-
- // Arrays stored in std::vector.
- std::vector<InType> p_cell_in; // Dimensions: (nproma, nlev, nblks_c)
- std::vector<int> vert_cell_idx; // Dimensions: (nproma, nblks_v, 6)
- std::vector<int> vert_cell_blk; // Dimensions: (nproma, nblks_v, 6)
- std::vector<OutType> coeff_int; // Dimensions: (nproma, 9-cell_type, nblks_v)
- std::vector<OutType> p_vert_out; // Dimensions: (nproma, nlev, nblks_v)
-
- const int cell_type = 9;
-
- Cells2vertsScalarLibTestFixture() {
- // Allocate and initialize inputs.
- p_cell_in.resize(num_elements_3d<InType>(nproma, nlev, nblks_c), static_cast<InType>(1));
- vert_cell_idx.resize(num_elements_3d<int>(nproma, nblks_v, 6), 1);
- vert_cell_blk.resize(num_elements_3d<int>(nproma, nblks_v, 6), 0);
- coeff_int.resize(num_elements_3d<InType>(nproma, cell_type, nblks_v), static_cast<OutType>(1));
-
- // Allocate output arrays and initialize to zero.
- p_vert_out.resize(num_elements_3d<OutType>(nproma, nlev, nblks_v), static_cast<OutType>(0));
- }
-};
-
-// Add test suite
-TYPED_TEST_SUITE(Cells2vertsScalarLibTestFixture, MixedTypesSP2DP);
-
-// Add test
-TYPED_TEST(Cells2vertsScalarLibTestFixture, cells2edges) {
+TYPED_TEST(InterpolationScalarMixedTestFixture, cells2verts) {
using InType = typename TestFixture::InType;
- using OutType = typename TestFixture::OutType;
+ using OutType = typename TestFixture::OutType;
- // Call the function
cells2verts_scalar_lib<InType,OutType>(
this->p_cell_in.data(),
this->vert_cell_idx.data(),
this->vert_cell_blk.data(),
- this->coeff_int.data(),
+ this->coeff_int_verts.data(),
this->p_vert_out.data(),
this->i_startblk,
this->i_endblk,
@@ -534,8 +504,8 @@ TYPED_TEST(Cells2vertsScalarLibTestFixture, cells2edges) {
//
////////////////////////////////////////////////////////////////////////////////
-
-// Define a typed test fixture with one precision of data
+// The test for cells2verts_ri is similar to cells2verts, but is done here separtely
+// to avoid as a differebt template instantiation is needed for the function call
template <typename Types>
class Cells2vertsriScalarLibTestFixture : public testing::Test, public interp_dimensions{
public:
@@ -546,17 +516,15 @@ public:
std::vector<InType> p_cell_in; // Dimensions: (nproma, nlev, nblks_c)
std::vector<int> vert_cell_idx; // Dimensions: (nproma, nblks_v, 6)
std::vector<int> vert_cell_blk; // Dimensions: (nproma, nblks_v, 6)
- std::vector<InType> coeff_int; // Dimensions: (nproma, 9-cell_type, nblks_v)
+ std::vector<InType> coeff_int; // Dimensions: (nproma, 6, nblks_v)
std::vector<OutType> p_vert_out; // Dimensions: (nproma, nlev, nblks_v)
- const int cell_type = 9;
-
Cells2vertsriScalarLibTestFixture() {
// Allocate and initialize inputs.
p_cell_in.resize(num_elements_3d<InType>(nproma, nlev, nblks_c), static_cast<InType>(1));
vert_cell_idx.resize(num_elements_3d<int>(nproma, nblks_v, 6), 1);
vert_cell_blk.resize(num_elements_3d<int>(nproma, nblks_v, 6), 0);
- coeff_int.resize(num_elements_3d<InType>(nproma, cell_type, nblks_v), static_cast<InType>(1));
+ coeff_int.resize(num_elements_3d<InType>(nproma, 6, nblks_v), static_cast<InType>(1));
// Allocate output arrays and initialize to zero.
p_vert_out.resize(num_elements_3d<OutType>(nproma, nlev, nblks_v), static_cast<OutType>(0));
@@ -567,7 +535,7 @@ public:
TYPED_TEST_SUITE(Cells2vertsriScalarLibTestFixture, MixedTypes);
// Add test
-TYPED_TEST(Cells2vertsriScalarLibTestFixture, cells2edges) {
+TYPED_TEST(Cells2vertsriScalarLibTestFixture, cells2verts_ri) {
using InType = typename TestFixture::InType;
using OutType = typename TestFixture::OutType;
--
GitLab
From 6e479502ea0799f4b9546c105ebc78e2028dc67d Mon Sep 17 00:00:00 2001
From: Pradipta Samanta <samanta@dkrz.de>
Date: Wed, 5 Mar 2025 15:52:34 +0100
Subject: [PATCH 32/36] removed whitespaces
---
src/interpolation/interpolation_bindings.cpp | 178 +++++++--------
src/interpolation/interpolation_bindings.h | 96 ++++----
.../mo_lib_interpolation_scalar.cpp | 214 +++++++++---------
.../mo_lib_interpolation_scalar.hpp | 38 ++--
.../mo_lib_interpolation_vector.cpp | 4 +-
test/c/test_interpolation_scalar.cpp | 20 +-
6 files changed, 275 insertions(+), 275 deletions(-)
diff --git a/src/interpolation/interpolation_bindings.cpp b/src/interpolation/interpolation_bindings.cpp
index d113952..fc39ccd 100644
--- a/src/interpolation/interpolation_bindings.cpp
+++ b/src/interpolation/interpolation_bindings.cpp
@@ -46,7 +46,7 @@ void edges2cells_vector_lib_sp(
int slev, int elev,
int nproma,
int nlev, int nblks_e, int nblks_c){
-
+
edges2cells_vector_lib<float>(p_vn_in, p_vt_in,
cell_edge_idx, cell_edge_blk,
e_bln_c_u, e_bln_c_v,
@@ -63,15 +63,15 @@ void verts2edges_scalar_lib_dp(const double* p_vertex_in,
const int* edge_vertex_idx, const int* edge_vertex_blk,
const double* coeff_int,
double* p_edge_out,
- const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
+ const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
const int nproma, const int nlev, const int nblks_v, const int nblks_e, const bool lacc){
verts2edges_scalar_lib<double>(p_vertex_in,
edge_vertex_idx, edge_vertex_blk, coeff_int,
p_edge_out,
- i_startblk, i_endblk,
- i_startidx_in, i_endidx_in, slev, elev,
+ i_startblk, i_endblk,
+ i_startidx_in, i_endidx_in, slev, elev,
nproma, nlev, nblks_v, nblks_e, lacc);
}
@@ -80,76 +80,76 @@ void verts2edges_scalar_lib_sp(const float* p_vertex_in,
const int* edge_vertex_idx, const int* edge_vertex_blk,
const float* coeff_int,
float* p_edge_out,
- const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
+ const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
const int nproma, const int nlev, const int nblks_v, const int nblks_e, const bool lacc){
verts2edges_scalar_lib<float>(p_vertex_in,
edge_vertex_idx, edge_vertex_blk, coeff_int,
p_edge_out,
- i_startblk, i_endblk,
- i_startidx_in, i_endidx_in, slev, elev,
+ i_startblk, i_endblk,
+ i_startidx_in, i_endidx_in, slev, elev,
nproma, nlev, nblks_v, nblks_e, lacc);
}
-void cells2edges_scalar_lib_dp(const double* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
- const double* coeff_int, double* p_edge_out, const int* i_startblk_in,
+void cells2edges_scalar_lib_dp(const double* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
+ const double* coeff_int, double* p_edge_out, const int* i_startblk_in,
const int* i_endblk_in, const int* i_startidx_in, const int* i_endidx_in,
const int slev, const int elev, const int nproma, const int nlev,
- const int nblk_c, const int nblks_e, const int patch_id,
+ const int nblk_c, const int nblks_e, const int patch_id,
const bool l_limited_area, const bool lfill_latbc, const bool lacc){
-
- cells2edges_scalar_lib<double, double>(p_cell_in,edge_cell_idx,edge_cell_blk,coeff_int,p_edge_out,i_startblk_in,
+
+ cells2edges_scalar_lib<double, double>(p_cell_in,edge_cell_idx,edge_cell_blk,coeff_int,p_edge_out,i_startblk_in,
i_endblk_in,i_startidx_in,i_endidx_in,slev,elev,nproma,nlev,
nblk_c,nblks_e,patch_id,l_limited_area,lfill_latbc,lacc);
}
-void cells2edges_scalar_lib_sp(const float* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
- const float* coeff_int, float* p_edge_out, const int* i_startblk_in,
+void cells2edges_scalar_lib_sp(const float* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
+ const float* coeff_int, float* p_edge_out, const int* i_startblk_in,
const int* i_endblk_in, const int* i_startidx_in, const int* i_endidx_in,
const int slev, const int elev, const int nproma, const int nlev,
- const int nblk_c, const int nblks_e, const int patch_id,
+ const int nblk_c, const int nblks_e, const int patch_id,
const bool l_limited_area, const bool lfill_latbc, const bool lacc){
-
- cells2edges_scalar_lib<float, float>(p_cell_in,edge_cell_idx,edge_cell_blk,coeff_int,p_edge_out,i_startblk_in,
+
+ cells2edges_scalar_lib<float, float>(p_cell_in,edge_cell_idx,edge_cell_blk,coeff_int,p_edge_out,i_startblk_in,
i_endblk_in,i_startidx_in,i_endidx_in,slev,elev,nproma,nlev,
nblk_c,nblks_e,patch_id,l_limited_area,lfill_latbc,lacc);
}
-void cells2edges_scalar_lib_sp2dp(const float* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
- const double* coeff_int, double* p_edge_out, const int* i_startblk_in,
+void cells2edges_scalar_lib_sp2dp(const float* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
+ const double* coeff_int, double* p_edge_out, const int* i_startblk_in,
const int* i_endblk_in, const int* i_startidx_in, const int* i_endidx_in,
const int slev, const int elev, const int nproma, const int nlev,
- const int nblk_c, const int nblks_e, const int patch_id,
+ const int nblk_c, const int nblks_e, const int patch_id,
const bool l_limited_area, const bool lfill_latbc, const bool lacc){
- cells2edges_scalar_lib<float, double>(p_cell_in,edge_cell_idx,edge_cell_blk,coeff_int,p_edge_out,i_startblk_in,
+ cells2edges_scalar_lib<float, double>(p_cell_in,edge_cell_idx,edge_cell_blk,coeff_int,p_edge_out,i_startblk_in,
i_endblk_in,i_startidx_in,i_endidx_in,slev,elev,nproma,nlev,
nblk_c,nblks_e,patch_id,l_limited_area,lfill_latbc,lacc);
}
-void edges2verts_scalar_lib_dp (const double* p_edge_in, const int* vert_edge_idx, const int* vert_edge_blk,
- const double* v_int, double* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in, const int slev,
+void edges2verts_scalar_lib_dp (const double* p_edge_in, const int* vert_edge_idx, const int* vert_edge_blk,
+ const double* v_int, double* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in, const int slev,
const int elev, const int nproma, const int nlev, const int nblks_e,
const int nblks_v, const bool lacc){
- edges2verts_scalar_lib<double>(p_edge_in, vert_edge_idx, vert_edge_blk, v_int,p_vert_out,i_startblk,
+ edges2verts_scalar_lib<double>(p_edge_in, vert_edge_idx, vert_edge_blk, v_int,p_vert_out,i_startblk,
i_endblk,i_startidx_in,i_endidx_in, slev, elev, nproma, nlev,nblks_e, nblks_v, lacc);
}
-void edges2verts_scalar_lib_sp(const float* p_edge_in, const int* vert_edge_idx, const int* vert_edge_blk,
- const float* v_int, float* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in, const int slev,
+void edges2verts_scalar_lib_sp(const float* p_edge_in, const int* vert_edge_idx, const int* vert_edge_blk,
+ const float* v_int, float* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in, const int slev,
const int elev, const int nproma, const int nlev, const int nblks_e,
const int nblks_v, const bool lacc){
- edges2verts_scalar_lib<float>(p_edge_in, vert_edge_idx, vert_edge_blk, v_int,p_vert_out,i_startblk,
+ edges2verts_scalar_lib<float>(p_edge_in, vert_edge_idx, vert_edge_blk, v_int,p_vert_out,i_startblk,
i_endblk,i_startidx_in,i_endidx_in, slev, elev, nproma, nlev,nblks_e, nblks_v, lacc);
}
@@ -157,16 +157,16 @@ void edges2verts_scalar_lib_sp(const float* p_edge_in, const int* vert_edge_idx,
void edges2cells_scalar_lib_dp(
const double* p_edge_in, const int* edge_idx, const int* edge_blk,
const double* coeff_int, double* p_cell_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
const int nlev, const int nblks_e, const int nblks_c,
const bool lacc)
{
edges2cells_scalar_lib<double>(
p_edge_in, edge_idx, edge_blk,
coeff_int, p_cell_out, i_startblk, i_endblk,
- i_startidx_in, i_endidx_in,
- slev, elev, nproma,
+ i_startidx_in, i_endidx_in,
+ slev, elev, nproma,
nlev, nblks_e, nblks_c,
lacc);
}
@@ -174,8 +174,8 @@ void edges2cells_scalar_lib_dp(
void edges2cells_scalar_lib_sp(
const float* p_edge_in, const int* edge_idx, const int* edge_blk,
const float* coeff_int, float* p_cell_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
const int nlev, const int nblks_e, const int nblks_c,
const bool lacc)
{
@@ -183,8 +183,8 @@ void edges2cells_scalar_lib_sp(
edges2cells_scalar_lib<float>(
p_edge_in, edge_idx, edge_blk,
coeff_int, p_cell_out, i_startblk, i_endblk,
- i_startidx_in, i_endidx_in,
- slev, elev, nproma,
+ i_startidx_in, i_endidx_in,
+ slev, elev, nproma,
nlev, nblks_e, nblks_c,
lacc);
}
@@ -194,16 +194,16 @@ void edges2cells_scalar_lib_sp(
void cells2verts_scalar_lib_dp(
const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
const double* coeff_int, double* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
const int nlev, const int nblks_c, const int nblks_v,
const bool lacc, const bool acc_async)
{
cells2verts_scalar_lib<double,double>(
p_cell_in, vert_cell_idx, vert_cell_blk,
coeff_int, p_vert_out, i_startblk, i_endblk,
- i_startidx_in, i_endidx_in,
- slev, elev, nproma,
+ i_startidx_in, i_endidx_in,
+ slev, elev, nproma,
nlev, nblks_c, nblks_v,
lacc, acc_async);
}
@@ -211,16 +211,16 @@ void cells2verts_scalar_lib_dp(
void cells2verts_scalar_lib_dp2sp(
const float* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
const double* coeff_int, double* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
const int nlev, const int nblks_c, const int nblks_v,
const bool lacc, const bool acc_async)
{
cells2verts_scalar_lib<float,double>(
p_cell_in, vert_cell_idx, vert_cell_blk,
coeff_int, p_vert_out, i_startblk, i_endblk,
- i_startidx_in, i_endidx_in,
- slev, elev, nproma,
+ i_startidx_in, i_endidx_in,
+ slev, elev, nproma,
nlev, nblks_c, nblks_v,
lacc, acc_async);
}
@@ -228,16 +228,16 @@ void cells2verts_scalar_lib_dp2sp(
void cells2verts_scalar_lib_sp(
const float* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
const int nlev, const int nblks_c, const int nblks_v,
const bool lacc, const bool acc_async)
{
cells2verts_scalar_lib<float,float>(
p_cell_in, vert_cell_idx, vert_cell_blk,
coeff_int, p_vert_out, i_startblk, i_endblk,
- i_startidx_in, i_endidx_in,
- slev, elev, nproma,
+ i_startidx_in, i_endidx_in,
+ slev, elev, nproma,
nlev, nblks_c, nblks_v,
lacc, acc_async);
}
@@ -247,16 +247,16 @@ void cells2verts_scalar_lib_sp(
void cells2verts_scalar_ri_lib_dp(
const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
const double* coeff_int, double* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
const int nlev, const int nblks_c, const int nblks_v,
const bool lacc, const bool acc_async)
{
cells2verts_scalar_ri_lib<double,double>(
p_cell_in, vert_cell_idx, vert_cell_blk,
coeff_int, p_vert_out, i_startblk, i_endblk,
- i_startidx_in, i_endidx_in,
- slev, elev, nproma,
+ i_startidx_in, i_endidx_in,
+ slev, elev, nproma,
nlev, nblks_c, nblks_v,
lacc, acc_async);
}
@@ -264,16 +264,16 @@ void cells2verts_scalar_ri_lib_dp(
void cells2verts_scalar_ri_lib_dp2sp(
const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
const double* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
const int nlev, const int nblks_c, const int nblks_v,
const bool lacc, const bool acc_async)
{
cells2verts_scalar_ri_lib<double,float>(
p_cell_in, vert_cell_idx, vert_cell_blk,
coeff_int, p_vert_out, i_startblk, i_endblk,
- i_startidx_in, i_endidx_in,
- slev, elev, nproma,
+ i_startidx_in, i_endidx_in,
+ slev, elev, nproma,
nlev, nblks_c, nblks_v,
lacc, acc_async);
}
@@ -281,16 +281,16 @@ void cells2verts_scalar_ri_lib_dp2sp(
void cells2verts_scalar_ri_lib_sp(
const float* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
const int nlev, const int nblks_c, const int nblks_v,
const bool lacc, const bool acc_async)
{
cells2verts_scalar_ri_lib<float,float>(
p_cell_in, vert_cell_idx, vert_cell_blk,
coeff_int, p_vert_out, i_startblk, i_endblk,
- i_startidx_in, i_endidx_in,
- slev, elev, nproma,
+ i_startidx_in, i_endidx_in,
+ slev, elev, nproma,
nlev, nblks_c, nblks_v,
lacc, acc_async);
}
@@ -298,64 +298,64 @@ void cells2verts_scalar_ri_lib_sp(
/////////////////////////////////////////////
void verts2cells_scalar_lib_dp(
- const double* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
+ const double* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
const double* coeff_int, double* p_cell_out, const int nblks_c, const int npromz_c,
- const int slev, const int elev, const int nproma, const int nlev,
+ const int slev, const int elev, const int nproma, const int nlev,
const int nblks_v, const bool lacc)
{
verts2cells_scalar_lib<double>(
- p_vert_in, cell_index_idx, cell_vertex_blk,
+ p_vert_in, cell_index_idx, cell_vertex_blk,
coeff_int, p_cell_out, nblks_c, npromz_c,
- slev, elev, nproma, nlev,
+ slev, elev, nproma, nlev,
nblks_v, lacc);
}
void verts2cells_scalar_lib_sp(
- const float* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
+ const float* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
const float* coeff_int, float* p_cell_out, const int nblks_c, const int npromz_c,
- const int slev, const int elev, const int nproma, const int nlev,
+ const int slev, const int elev, const int nproma, const int nlev,
const int nblks_v, const bool lacc)
{
verts2cells_scalar_lib<float>(
- p_vert_in, cell_index_idx, cell_vertex_blk,
+ p_vert_in, cell_index_idx, cell_vertex_blk,
coeff_int, p_cell_out, nblks_c, npromz_c,
- slev, elev, nproma, nlev,
+ slev, elev, nproma, nlev,
nblks_v, lacc);
}
/////////////////////////////////////////////
void cell_avg_lib_dp(
- const double* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
- const double* avg_coeff, double* avg_psi_c, const int i_startblk, const int i_endblk,
+ const double* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
+ const double* avg_coeff, double* avg_psi_c, const int i_startblk, const int i_endblk,
const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c,
const bool lacc)
{
cell_avg_lib<double>(
- psi_c, cell_neighbor_idx, cell_neighbor_blk,
- avg_coeff, avg_psi_c, i_startblk, i_endblk,
+ psi_c, cell_neighbor_idx, cell_neighbor_blk,
+ avg_coeff, avg_psi_c, i_startblk, i_endblk,
i_startidx_in, i_endidx_in,
- slev, elev, nproma,
- nlev, nblks_c,
+ slev, elev, nproma,
+ nlev, nblks_c,
lacc);
}
-
+
void cell_avg_lib_sp(
- const float* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
- const float* avg_coeff, float* avg_psi_c, const int i_startblk, const int i_endblk,
+ const float* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
+ const float* avg_coeff, float* avg_psi_c, const int i_startblk, const int i_endblk,
const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c,
const bool lacc)
{
cell_avg_lib<float>(
- psi_c, cell_neighbor_idx, cell_neighbor_blk,
- avg_coeff, avg_psi_c, i_startblk, i_endblk,
+ psi_c, cell_neighbor_idx, cell_neighbor_blk,
+ avg_coeff, avg_psi_c, i_startblk, i_endblk,
i_startidx_in, i_endidx_in,
- slev, elev, nproma,
- nlev, nblks_c,
+ slev, elev, nproma,
+ nlev, nblks_c,
lacc);
}
-
+
diff --git a/src/interpolation/interpolation_bindings.h b/src/interpolation/interpolation_bindings.h
index 14c1f34..7c6b125 100644
--- a/src/interpolation/interpolation_bindings.h
+++ b/src/interpolation/interpolation_bindings.h
@@ -40,65 +40,65 @@ extern "C"{
const int* edge_vertex_idx, const int* edge_vertex_blk,
const double* coeff_int,
double* p_edge_out,
- const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
+ const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
const int nproma, const int nlev, const int nblks_v, const int nblks_e, const bool lacc);
void verts2edges_scalar_lib_sp(const float* p_vertex_in,
const int* edge_vertex_idx, const int* edge_vertex_blk,
const float* coeff_int,
float* p_edge_out,
- const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
+ const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
const int nproma, const int nlev, const int nblks_v, const int nblks_e, const bool lacc);
-
- void cells2edges_scalar_lib_dp(const double* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
- const double* coeff_int, double* p_edge_out, const int* i_startblk_in,
+
+ void cells2edges_scalar_lib_dp(const double* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
+ const double* coeff_int, double* p_edge_out, const int* i_startblk_in,
const int* i_endblk_in, const int* i_startidx_in, const int* i_endidx_in,
const int slev, const int elev, const int nproma, const int nlev,
- const int nblk_c, const int nblks_e, const int patch_id,
+ const int nblk_c, const int nblks_e, const int patch_id,
const bool l_limited_area, const bool lfill_latbc, const bool lacc);
- void cells2edges_scalar_lib_sp(const float* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
- const float* coeff_int, float* p_edge_out, const int* i_startblk_in,
+ void cells2edges_scalar_lib_sp(const float* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
+ const float* coeff_int, float* p_edge_out, const int* i_startblk_in,
const int* i_endblk_in, const int* i_startidx_in, const int* i_endidx_in,
const int slev, const int elev, const int nproma, const int nlev,
- const int nblk_c, const int nblks_e, const int patch_id,
+ const int nblk_c, const int nblks_e, const int patch_id,
const bool l_limited_area, const bool lfill_latbc, const bool lacc);
- void cells2edges_scalar_lib_sp2dp(const float* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
- const double* coeff_int, double* p_edge_out, const int* i_startblk_in,
+ void cells2edges_scalar_lib_sp2dp(const float* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
+ const double* coeff_int, double* p_edge_out, const int* i_startblk_in,
const int* i_endblk_in, const int* i_startidx_in, const int* i_endidx_in,
const int slev, const int elev, const int nproma, const int nlev,
- const int nblk_c, const int nblks_e, const int patch_id,
+ const int nblk_c, const int nblks_e, const int patch_id,
const bool l_limited_area, const bool lfill_latbc, const bool lacc);
- void edges2verts_scalar_lib_dp (const double* p_edge_in, const int* vert_edge_idx, const int* vert_edge_blk,
- const double* v_int, double* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in, const int slev,
+ void edges2verts_scalar_lib_dp (const double* p_edge_in, const int* vert_edge_idx, const int* vert_edge_blk,
+ const double* v_int, double* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in, const int slev,
const int elev, const int nproma, const int nlev, const int nblks_e,
const int nblks_v, const bool lacc);
- void edges2verts_scalar_lib_sp (const float* p_edge_in, const int* vert_edge_idx, const int* vert_edge_blk,
- const float* v_int, float* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in, const int slev,
+ void edges2verts_scalar_lib_sp (const float* p_edge_in, const int* vert_edge_idx, const int* vert_edge_blk,
+ const float* v_int, float* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in, const int slev,
const int elev, const int nproma, const int nlev, const int nblks_e,
const int nblks_v, const bool lacc);
void edges2cells_scalar_lib_dp(
const double* p_edge_in, const int* edge_idx, const int* edge_blk,
const double* coeff_int, double* p_cell_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
const int nlev, const int nblks_e, const int nblks_c,
const bool lacc);
void edges2cells_scalar_lib_sp(
const float* p_edge_in, const int* edge_idx, const int* edge_blk,
const float* coeff_int, float* p_cell_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
const int nlev, const int nblks_e, const int nblks_c,
const bool lacc);
@@ -107,22 +107,22 @@ extern "C"{
void cells2verts_scalar_lib_dp(
const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
const double* coeff_int, double* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
const int nlev, const int nblks_c, const int nblks_v,
const bool lacc, const bool acc_async);
void cells2verts_scalar_lib_dp2sp(
const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
const int nlev, const int nblks_c, const int nblks_v,
const bool lacc, const bool acc_async);
void cells2verts_scalar_lib_sp(
const float* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
const int nlev, const int nblks_c, const int nblks_v,
const bool lacc, const bool acc_async);
@@ -131,56 +131,56 @@ extern "C"{
void cells2verts_scalar_ri_lib_dp(
const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
const double* coeff_int, double* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
const int nlev, const int nblks_c, const int nblks_v,
const bool lacc, const bool acc_async);
void cells2verts_scalar_ri_lib_dp2sp(
const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
const double* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
const int nlev, const int nblks_c, const int nblks_v,
const bool lacc, const bool acc_async);
void cells2verts_scalar_ri_lib_sp(
const float* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
const int nlev, const int nblks_c, const int nblks_v,
const bool lacc, const bool acc_async);
/////////////////////////////////////////////
void verts2cells_scalar_lib_dp(
- const double* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
+ const double* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
const double* coeff_int, double* p_cell_out, const int nblks_c, const int npromz_c,
- const int slev, const int elev, const int nproma, const int nlev,
+ const int slev, const int elev, const int nproma, const int nlev,
const int nblks_v, const bool lacc);
void verts2cells_scalar_lib_sp(
- const float* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
+ const float* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
const float* coeff_int, float* p_cell_out, const int nblks_c, const int npromz_c,
- const int slev, const int elev, const int nproma, const int nlev,
+ const int slev, const int elev, const int nproma, const int nlev,
const int nblks_v, const bool lacc);
/////////////////////////////////////////////
void cell_avg_lib_dp(
- const double* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
- const double* avg_coeff, double* avg_psi_c, const int i_startblk, const int i_endblk,
+ const double* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
+ const double* avg_coeff, double* avg_psi_c, const int i_startblk, const int i_endblk,
const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c,
const bool lacc);
void cell_avg_lib_sp(
- const float* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
- const float* avg_coeff, float* avg_psi_c, const int i_startblk, const int i_endblk,
+ const float* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
+ const float* avg_coeff, float* avg_psi_c, const int i_startblk, const int i_endblk,
const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c,
const bool lacc);
}
diff --git a/src/interpolation/mo_lib_interpolation_scalar.cpp b/src/interpolation/mo_lib_interpolation_scalar.cpp
index 26bd876..7f37448 100644
--- a/src/interpolation/mo_lib_interpolation_scalar.cpp
+++ b/src/interpolation/mo_lib_interpolation_scalar.cpp
@@ -29,13 +29,13 @@
///
template <typename T>
void verts2edges_scalar_lib(const T* p_vertex_in,
- const int* edge_vertex_idx, const int* edge_vertex_blk,
+ const int* edge_vertex_idx, const int* edge_vertex_blk,
const T* coeff_int,
T* p_edge_out,
- const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
+ const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
const int nproma, const int nlev, const int nblks_v, const int nblks_e, const bool lacc){
-
+
// Wrap raw pointers in unmanaged Kokkos Views.
typedef Kokkos::View<const T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstT3D;
typedef Kokkos::View<T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedT3D;
@@ -46,7 +46,7 @@ void verts2edges_scalar_lib(const T* p_vertex_in,
UnmanagedConstInt3D iblk_view (edge_vertex_blk, nproma,nblks_e, 4);
UnmanagedConstT3D coeff_int_view (coeff_int, nproma,2,nblks_e);
UnmanagedT3D p_edge_out_view (p_edge_out, nproma,nlev,nblks_e);
-
+
for (int jb = i_startblk; jb < i_endblk + 1; ++jb) {
int i_startidx, i_endidx;
@@ -54,17 +54,17 @@ void verts2edges_scalar_lib(const T* p_vertex_in,
jb, i_startblk, i_endblk,
i_startidx, i_endidx);
- Kokkos::parallel_for("verts2edges_scalar", Kokkos::MDRangePolicy<Kokkos::Rank<2>>
+ Kokkos::parallel_for("verts2edges_scalar", Kokkos::MDRangePolicy<Kokkos::Rank<2>>
({slev, i_startidx}, {elev + 1, i_endidx + 1}),
KOKKOS_LAMBDA(const int jk, const int je) {
-
- p_edge_out_view(je, jk, jb) = coeff_int_view(je, 0, jb) *
- p_vertex_in_view(iidx_view(je, jb, 0), jk, iblk_view(je, jb, 0)) +
+
+ p_edge_out_view(je, jk, jb) = coeff_int_view(je, 0, jb) *
+ p_vertex_in_view(iidx_view(je, jb, 0), jk, iblk_view(je, jb, 0)) +
coeff_int_view(je, 1, jb)*p_vertex_in_view(iidx_view(je, jb, 1), jk, iblk_view(je, jb, 1));
-
+
});
Kokkos::fence();
- }
+ }
}
//------------------------------------------------------------------------
@@ -75,12 +75,12 @@ void verts2edges_scalar_lib(const T* p_vertex_in,
/// velocity points.
///
template <typename T, typename S>
-void cells2edges_scalar_lib(const T* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
- const S* coeff_int, S* p_edge_out, const int* i_startblk_in,
+void cells2edges_scalar_lib(const T* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
+ const S* coeff_int, S* p_edge_out, const int* i_startblk_in,
const int* i_endblk_in, const int* i_startidx_in, const int* i_endidx_in,
const int slev, const int elev, const int nproma, const int nlev,
- const int nblk_c, const int nblks_e, const int patch_id,
- const bool l_limited_area, const bool lfill_latbc, const bool lacc){
+ const int nblk_c, const int nblks_e, const int patch_id,
+ const bool l_limited_area, const bool lfill_latbc, const bool lacc){
// Wrap raw pointers in unmanaged Kokkos Views.
typedef Kokkos::View<const T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstT3D;
@@ -93,7 +93,7 @@ void cells2edges_scalar_lib(const T* p_cell_in, const int* edge_cell_idx, const
UnmanagedConstInt3D iblk_view (edge_cell_blk, nproma, nblks_e, 2);
UnmanagedConstS3D coeff_int_view (coeff_int, nproma, 2, nblks_e);
UnmanagedS3D p_edge_out_view (p_edge_out, nproma, nlev, nblks_e);
-
+
//Fill outermost nest boundary
int i_startblk, i_endblk;
if ((l_limited_area || patch_id > 0) && (lfill_latbc)){
@@ -103,13 +103,13 @@ void cells2edges_scalar_lib(const T* p_cell_in, const int* edge_cell_idx, const
for (int jb = i_startblk; jb < i_endblk + 1; ++jb) {
int i_startidx, i_endidx;
- get_indices_e_lib(i_startidx_in[0], i_endidx_in[0], nproma, jb,
+ get_indices_e_lib(i_startidx_in[0], i_endidx_in[0], nproma, jb,
i_startblk, i_endblk, i_startidx, i_endidx);
- Kokkos::parallel_for("cells2edges_scalar", Kokkos::MDRangePolicy<Kokkos::Rank<2>>
+ Kokkos::parallel_for("cells2edges_scalar", Kokkos::MDRangePolicy<Kokkos::Rank<2>>
({slev, i_startidx}, {elev + 1, i_endidx + 1}),
KOKKOS_LAMBDA(const int jk, const int je) {
-
+
if (iidx_view(je, jb, 0) >= 0 && iblk_view(je, jb, 0) >= 0){
p_edge_out_view(je, jk, jb) = p_cell_in_view(iidx_view(je, jb, 0), jk, iblk_view(je, jb, 0));
}
@@ -119,7 +119,7 @@ void cells2edges_scalar_lib(const T* p_cell_in, const int* edge_cell_idx, const
else{
std::cerr << "mo_interpolation:cells2edges_scalar_lib: error in lateral boundary filling" << std::endl;
std::exit(EXIT_FAILURE);
- }
+ }
});
Kokkos::fence();
}
@@ -132,20 +132,20 @@ void cells2edges_scalar_lib(const T* p_cell_in, const int* edge_cell_idx, const
for (int jb = i_startblk; jb < i_endblk + 1; ++jb) {
int i_startidx, i_endidx;
- get_indices_e_lib(i_startidx_in[1], i_endidx_in[1], nproma, jb,
+ get_indices_e_lib(i_startidx_in[1], i_endidx_in[1], nproma, jb,
i_startblk, i_endblk, i_startidx, i_endidx);
- Kokkos::parallel_for("cells2edges_scalar", Kokkos::MDRangePolicy<Kokkos::Rank<2>>
+ Kokkos::parallel_for("cells2edges_scalar", Kokkos::MDRangePolicy<Kokkos::Rank<2>>
({slev, i_startidx}, {elev + 1, i_endidx + 1}),
KOKKOS_LAMBDA(const int jk, const int je) {
-
- p_edge_out_view(je, jk, jb) = coeff_int_view(je, 0, jb) *
+
+ p_edge_out_view(je, jk, jb) = coeff_int_view(je, 0, jb) *
p_cell_in_view(iidx_view(je, jb, 0), jk, iblk_view(je, jb, 0)) +
coeff_int_view(je, 1, jb) * p_cell_in_view(iidx_view(je, jb, 1), jk, iblk_view(je, jb, 1));
-
+
});
Kokkos::fence();
- }
+ }
}
}
@@ -158,9 +158,9 @@ void cells2edges_scalar_lib(const T* p_cell_in, const int* edge_cell_idx, const
/// centers of dual faces.
///
template <typename T>
-void edges2verts_scalar_lib (const T* p_edge_in, const int* vert_edge_idx, const int* vert_edge_blk,
- const T* v_int, T* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in, const int slev,
+void edges2verts_scalar_lib (const T* p_edge_in, const int* vert_edge_idx, const int* vert_edge_blk,
+ const T* v_int, T* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in, const int slev,
const int elev, const int nproma, const int nlev, const int nblks_e,
const int nblks_v, const bool lacc){
@@ -174,7 +174,7 @@ void edges2verts_scalar_lib (const T* p_edge_in, const int* vert_edge_idx, const
UnmanagedConstInt3D iblk_view (vert_edge_blk, nproma,nblks_v, 6);
UnmanagedConstT3D v_int_view (v_int, nproma, 6, nblks_v);
UnmanagedT3D p_vert_out_view (p_vert_out, nproma, nlev, nblks_v);
-
+
for (int jb = i_startblk; jb < i_endblk + 1; ++jb) {
int i_startidx, i_endidx;
@@ -182,16 +182,16 @@ void edges2verts_scalar_lib (const T* p_edge_in, const int* vert_edge_idx, const
jb, i_startblk, i_endblk,
i_startidx, i_endidx);
- Kokkos::parallel_for("edges2verts_scalar", Kokkos::MDRangePolicy<Kokkos::Rank<2>>
+ Kokkos::parallel_for("edges2verts_scalar", Kokkos::MDRangePolicy<Kokkos::Rank<2>>
({slev, i_startidx}, {elev + 1, i_endidx + 1}),
KOKKOS_LAMBDA(const int jk, const int jv) {
-
- p_vert_out_view(jv, jk, jb) = v_int_view(jv, 0, jb)*p_edge_in_view(iidx_view(jv, jb, 0), jk, iblk_view(jv, jb, 0))
- + v_int_view(jv, 1, jb)*p_edge_in_view(iidx_view(jv, jb, 1), jk, iblk_view(jv, jb, 1))
- + v_int_view(jv, 2, jb)*p_edge_in_view(iidx_view(jv, jb, 2), jk, iblk_view(jv, jb, 2))
- + v_int_view(jv, 3, jb)*p_edge_in_view(iidx_view(jv, jb, 3), jk, iblk_view(jv, jb, 3))
- + v_int_view(jv, 4, jb)*p_edge_in_view(iidx_view(jv, jb, 4), jk, iblk_view(jv, jb, 4))
- + v_int_view(jv, 5, jb)*p_edge_in_view(iidx_view(jv, jb, 5), jk, iblk_view(jv, jb, 5));
+
+ p_vert_out_view(jv, jk, jb) = v_int_view(jv, 0, jb)*p_edge_in_view(iidx_view(jv, jb, 0), jk, iblk_view(jv, jb, 0))
+ + v_int_view(jv, 1, jb)*p_edge_in_view(iidx_view(jv, jb, 1), jk, iblk_view(jv, jb, 1))
+ + v_int_view(jv, 2, jb)*p_edge_in_view(iidx_view(jv, jb, 2), jk, iblk_view(jv, jb, 2))
+ + v_int_view(jv, 3, jb)*p_edge_in_view(iidx_view(jv, jb, 3), jk, iblk_view(jv, jb, 3))
+ + v_int_view(jv, 4, jb)*p_edge_in_view(iidx_view(jv, jb, 4), jk, iblk_view(jv, jb, 4))
+ + v_int_view(jv, 5, jb)*p_edge_in_view(iidx_view(jv, jb, 5), jk, iblk_view(jv, jb, 5));
});
Kokkos::fence();
}
@@ -209,8 +209,8 @@ template <typename T>
void edges2cells_scalar_lib(
const T* p_edge_in, const int* edge_idx, const int* edge_blk,
const T* coeff_int, T* p_cell_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
const int nlev, const int nblks_e, const int nblks_c,
const bool lacc)
{
@@ -262,8 +262,8 @@ template <typename T, typename S>
void cells2verts_scalar_lib(
const T* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
const S* coeff_int, S* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
const int nlev, const int nblks_c, const int nblks_v,
const bool lacc, const bool acc_async)
{
@@ -297,7 +297,7 @@ void cells2verts_scalar_lib(
Kokkos::MDRangePolicy<Kokkos::Rank<2>> innerPolicy(
{slev, i_startidx}, {elev + 1, i_endidx + 1});
-
+
Kokkos::parallel_for("cells2verts_scalar_lib", innerPolicy,
KOKKOS_LAMBDA(const int jk, const int jv) {
@@ -322,7 +322,7 @@ template <typename T, typename S>
void cells2verts_scalar_ri_lib(
const T* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
const T* coeff_int, S* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
+ const int i_startidx_in, const int i_endidx_in,
const int slev, const int elev, const int nproma,
const int nlev, const int nblks_c, const int nblks_v,
const bool lacc, const bool acc_async)
@@ -360,7 +360,7 @@ void cells2verts_scalar_ri_lib(
Kokkos::MDRangePolicy<Kokkos::Rank<2>> innerPolicy(
{slev, i_startidx}, {elev + 1, i_endidx + 1});
-
+
Kokkos::parallel_for("cells2verts_scalar_ri_lib", innerPolicy,
KOKKOS_LAMBDA(const int jk, const int jv) {
@@ -387,9 +387,9 @@ void cells2verts_scalar_ri_lib(
///
template <typename T>
void verts2cells_scalar_lib(
- const T* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
+ const T* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
const T* coeff_int, T* p_cell_out, const int nblks_c, const int npromz_c,
- const int slev, const int elev, const int nproma, const int nlev,
+ const int slev, const int elev, const int nproma, const int nlev,
const int nblks_v, const bool lacc)
{
// Wrap raw pointers in unmanaged Kokkos Views.
@@ -447,8 +447,8 @@ void verts2cells_scalar_lib(
///
template <typename T>
void cell_avg_lib(
- const T* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
- const T* avg_coeff, T* avg_psi_c, const int i_startblk, const int i_endblk,
+ const T* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
+ const T* avg_coeff, T* avg_psi_c, const int i_startblk, const int i_endblk,
const int i_startidx_in, const int i_endidx_in,
const int slev, const int elev, const int nproma,
const int nlev, const int nblks_c, const bool lacc)
@@ -482,8 +482,8 @@ void cell_avg_lib(
Kokkos::parallel_for("cell_avg_lib_inner", innerPolicy,
KOKKOS_LAMBDA(const int jk, const int jc) {
// calculate the weighted average
-
- avg_psi_c_view(jc, jk, jb) =
+
+ avg_psi_c_view(jc, jk, jb) =
psi_c_view(jc, jk, jb)*avg_coeff_view(jc, 0, jb) +
psi_c_view(iidx_view(jc, jb, 0), jk, iblk_view(jc, jb, 0))*avg_coeff_view(jc, 1, jb) +
psi_c_view(iidx_view(jc, jb, 1), jk, iblk_view(jc, jb, 1))*avg_coeff_view(jc, 2, jb) +
@@ -503,8 +503,8 @@ void verts2edges_scalar_lib<double>(const double* p_vertex_in,
const int* edge_vertex_idx, const int* edge_vertex_blk,
const double* coeff_int,
double* p_edge_out,
- const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
+ const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
const int nproma, const int nlev, const int nblks_v, const int nblks_e, const bool lacc);
template
@@ -512,46 +512,46 @@ void verts2edges_scalar_lib<float>(const float* p_vertex_in,
const int* edge_vertex_idx, const int* edge_vertex_blk,
const float* coeff_int,
float* p_edge_out,
- const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
+ const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
const int nproma, const int nlev, const int nblks_v, const int nblks_e, const bool lacc);
template
-void cells2edges_scalar_lib<double, double>(const double* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
- const double* coeff_int, double* p_edge_out, const int* i_startblk_in,
+void cells2edges_scalar_lib<double, double>(const double* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
+ const double* coeff_int, double* p_edge_out, const int* i_startblk_in,
const int* i_endblk_in, const int* i_startidx_in, const int* i_endidx_in,
const int slev, const int elev, const int nproma, const int nlev,
- const int nblk_c, const int nblks_e, const int patch_id,
+ const int nblk_c, const int nblks_e, const int patch_id,
const bool l_limited_area, const bool lfill_latbc, const bool lacc);
template
-void cells2edges_scalar_lib<float, float>(const float* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
- const float* coeff_int, float* p_edge_out, const int* i_startblk_in,
+void cells2edges_scalar_lib<float, float>(const float* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
+ const float* coeff_int, float* p_edge_out, const int* i_startblk_in,
const int* i_endblk_in, const int* i_startidx_in, const int* i_endidx_in,
const int slev, const int elev, const int nproma, const int nlev,
- const int nblk_c, const int nblks_e, const int patch_id,
+ const int nblk_c, const int nblks_e, const int patch_id,
const bool l_limited_area, const bool lfill_latbc, const bool lacc);
template // sp2dp
-void cells2edges_scalar_lib<float, double>(const float* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
- const double* coeff_int, double* p_edge_out, const int* i_startblk_in,
+void cells2edges_scalar_lib<float, double>(const float* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
+ const double* coeff_int, double* p_edge_out, const int* i_startblk_in,
const int* i_endblk_in, const int* i_startidx_in, const int* i_endidx_in,
const int slev, const int elev, const int nproma, const int nlev,
- const int nblk_c, const int nblks_e, const int patch_id,
+ const int nblk_c, const int nblks_e, const int patch_id,
const bool l_limited_area, const bool lfill_latbc, const bool lacc);
template
-void edges2verts_scalar_lib<double> (const double* p_edge_in, const int* vert_edge_idx, const int* vert_edge_blk,
- const double* v_int, double* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in, const int slev,
+void edges2verts_scalar_lib<double> (const double* p_edge_in, const int* vert_edge_idx, const int* vert_edge_blk,
+ const double* v_int, double* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in, const int slev,
const int elev, const int nproma, const int nlev, const int nblks_e,
const int nblks_v, const bool lacc);
template
-void edges2verts_scalar_lib <float> (const float* p_edge_in, const int* vert_edge_idx, const int* vert_edge_blk,
- const float* v_int, float* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in, const int slev,
+void edges2verts_scalar_lib <float> (const float* p_edge_in, const int* vert_edge_idx, const int* vert_edge_blk,
+ const float* v_int, float* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in, const int slev,
const int elev, const int nproma, const int nlev, const int nblks_e,
const int nblks_v, const bool lacc);
@@ -560,8 +560,8 @@ template
void edges2cells_scalar_lib<double>(
const double* p_edge_in, const int* edge_idx, const int* edge_blk,
const double* coeff_int, double* p_cell_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
const int nlev, const int nblks_e, const int nblks_c,
const bool lacc);
@@ -569,98 +569,98 @@ template
void edges2cells_scalar_lib<float>(
const float* p_edge_in, const int* edge_idx, const int* edge_blk,
const float* coeff_int, float* p_cell_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
const int nlev, const int nblks_e, const int nblks_c,
const bool lacc);
-template
+template
void cells2verts_scalar_lib<double,double>(
const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
const double* coeff_int, double* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
const int nlev, const int nblks_c, const int nblks_v,
const bool lacc, const bool acc_async);
-template
+template
void cells2verts_scalar_lib<float,double>(
const float* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
const double* coeff_int, double* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
const int nlev, const int nblks_c, const int nblks_v,
const bool lacc, const bool acc_async);
-template
+template
void cells2verts_scalar_lib<float,float>(
const float* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
const int nlev, const int nblks_c, const int nblks_v,
const bool lacc, const bool acc_async);
-template
+template
void cells2verts_scalar_ri_lib<double,double>(
const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
const double* coeff_int, double* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
const int nlev, const int nblks_c, const int nblks_v,
const bool lacc, const bool acc_async);
-template
+template
void cells2verts_scalar_ri_lib<double,float>(
const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
const double* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
const int nlev, const int nblks_c, const int nblks_v,
const bool lacc, const bool acc_async);
-template
+template
void cells2verts_scalar_ri_lib<float,float>(
const float* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
const int nlev, const int nblks_c, const int nblks_v,
const bool lacc, const bool acc_async);
-template
+template
void verts2cells_scalar_lib<double>(
- const double* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
+ const double* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
const double* coeff_int, double* p_cell_out, const int nblks_c, const int npromz_c,
- const int slev, const int elev, const int nproma, const int nlev,
+ const int slev, const int elev, const int nproma, const int nlev,
const int nblks_v, const bool lacc);
-template
+template
void verts2cells_scalar_lib<float>(
- const float* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
+ const float* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
const float* coeff_int, float* p_cell_out, const int nblks_c, const int npromz_c,
- const int slev, const int elev, const int nproma, const int nlev,
+ const int slev, const int elev, const int nproma, const int nlev,
const int nblks_v, const bool lacc);
template
void cell_avg_lib<double>(
- const double* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
- const double* avg_coeff, double* avg_psi_c, const int i_startblk, const int i_endblk,
+ const double* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
+ const double* avg_coeff, double* avg_psi_c, const int i_startblk, const int i_endblk,
const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c,
const bool lacc);
-
+
template
void cell_avg_lib<float>(
- const float* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
- const float* avg_coeff, float* avg_psi_c, const int i_startblk, const int i_endblk,
+ const float* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
+ const float* avg_coeff, float* avg_psi_c, const int i_startblk, const int i_endblk,
const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c,
const bool lacc);
diff --git a/src/interpolation/mo_lib_interpolation_scalar.hpp b/src/interpolation/mo_lib_interpolation_scalar.hpp
index 78f7df4..85c8fc5 100644
--- a/src/interpolation/mo_lib_interpolation_scalar.hpp
+++ b/src/interpolation/mo_lib_interpolation_scalar.hpp
@@ -13,62 +13,62 @@
template <typename T>
void verts2edges_scalar_lib(const T* p_vertex_in,
- const int* edge_vertex_idx, const int* edge_vertex_blk,
+ const int* edge_vertex_idx, const int* edge_vertex_blk,
const T* coeff_int,
T* p_edge_out,
- const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
+ const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
const int nproma, const int nlev, const int nblks_v, const int nblks_e, const bool lacc);
;
template <typename T, typename S>
-void cells2edges_scalar_lib(const T* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
- const S* coeff_int, S* p_edge_out, const int* i_startblk_in,
+void cells2edges_scalar_lib(const T* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
+ const S* coeff_int, S* p_edge_out, const int* i_startblk_in,
const int* i_endblk_in, const int* i_startidx_in, const int* i_endidx_in,
const int slev, const int elev, const int nproma, const int nlev,
- const int nblk_c, const int nblks_e, const int patch_id,
+ const int nblk_c, const int nblks_e, const int patch_id,
const bool l_limited_area, const bool lfill_latbc, const bool lacc);
template <typename T>
-void edges2verts_scalar_lib (const T* p_edge_in, const int* vert_edge_idx, const int* vert_edge_blk,
- const T* v_int, T* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in, const int slev,
+void edges2verts_scalar_lib (const T* p_edge_in, const int* vert_edge_idx, const int* vert_edge_blk,
+ const T* v_int, T* p_vert_out, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in, const int slev,
const int elev, const int nproma, const int nlev, const int nblks_e,
const int nblks_v, const bool lacc);
template <typename T>
void edges2cells_scalar_lib(const T* p_edge_in, const int* edge_idx, const int* edge_blk,
const T* coeff_int, T* p_cell_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
const int nlev, const int nblks_e, const int nblks_c,
const bool lacc);
template <typename T, typename S>
void cells2verts_scalar_lib(const T* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
const S* coeff_int, S* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
const int nlev, const int nblks_c, const int nblks_v,
const bool lacc, const bool acc_async);
template <typename T, typename S>
void cells2verts_scalar_ri_lib(const T* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
const T* coeff_int, S* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
const int nlev, const int nblks_c, const int nblks_v,
const bool lacc, const bool acc_async);
template <typename T>
-void verts2cells_scalar_lib(const T* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
+void verts2cells_scalar_lib(const T* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
const T* coeff_int, T* p_cell_out, const int nblks_c, const int npromz_c,
- const int slev, const int elev, const int nproma, const int nlev,
+ const int slev, const int elev, const int nproma, const int nlev,
const int nblks_v, const bool lacc);
template <typename T>
-void cell_avg_lib(const T* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
- const T* avg_coeff, T* avg_psi_c, const int i_startblk, const int i_endblk,
+void cell_avg_lib(const T* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
+ const T* avg_coeff, T* avg_psi_c, const int i_startblk, const int i_endblk,
const int i_startidx_in, const int i_endidx_in,
const int slev, const int elev, const int nproma,
const int nlev, const int nblks_c, const bool lacc);
diff --git a/src/interpolation/mo_lib_interpolation_vector.cpp b/src/interpolation/mo_lib_interpolation_vector.cpp
index 74c2be7..fa1ed32 100644
--- a/src/interpolation/mo_lib_interpolation_vector.cpp
+++ b/src/interpolation/mo_lib_interpolation_vector.cpp
@@ -31,7 +31,7 @@ void edges2cells_vector_lib(
typedef Kokkos::View<T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedT3D;
typedef Kokkos::View<const int***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstInt3D;
-
+
UnmanagedConstT3D p_vn_in_view(p_vn_in, nproma, nlev, nblks_e);
UnmanagedConstT3D p_vt_in_view(p_vt_in, nproma, nlev, nblks_e);
@@ -90,7 +90,7 @@ void edges2cells_vector_lib(
}
}
-template
+template
void edges2cells_vector_lib<double>(
const double* p_vn_in, const double* p_vt_in,
const int* cell_edge_idx, const int* cell_edge_blk,
diff --git a/test/c/test_interpolation_scalar.cpp b/test/c/test_interpolation_scalar.cpp
index 6338a36..606d1e9 100644
--- a/test/c/test_interpolation_scalar.cpp
+++ b/test/c/test_interpolation_scalar.cpp
@@ -181,7 +181,7 @@ TYPED_TEST(InterpolationScalarTypedTestFixture, Verts2Edges) {
this->nblks_e,
this->lacc);
- // Check the outputs only for blocks in the range
+ // Check the outputs only for blocks in the range
// { [i_startblk, i_endblk], [slev,elev], [i_startidx, i_endidx] }
for (int block = this->i_startblk; block <= this->i_endblk; ++block) {
for (int level = this->slev; level < this->elev; ++level) {
@@ -222,7 +222,7 @@ TYPED_TEST(InterpolationScalarTypedTestFixture, Edges2Verts) {
this->nblks_v,
this->lacc);
- // Check the outputs only for blocks in the range
+ // Check the outputs only for blocks in the range
// { [i_startblk, i_endblk], [slev,elev], [i_startidx_in, i_endidx_in] }
for (int block = this->i_startblk; block <= this->i_endblk; ++block) {
for (int level = this->slev; level < this->elev; ++level) {
@@ -263,7 +263,7 @@ TYPED_TEST(InterpolationScalarTypedTestFixture, Edges2Cells) {
this->nblks_c,
this->lacc);
- // Check the outputs only for blocks in the range
+ // Check the outputs only for blocks in the range
// { [i_startblk, i_endblk], [slev,elev], [i_startidx_in, i_endidx_in] }
for (int block = this->i_startblk; block <= this->i_endblk; ++block) {
for (int level = this->slev; level < this->elev; ++level) {
@@ -294,9 +294,9 @@ TYPED_TEST(InterpolationScalarTypedTestFixture, Verts2Cells) {
this->nlev,
this->nblks_v,
this->lacc);
-
- // Check the outputs only for blocks in the range
+
+ // Check the outputs only for blocks in the range
// { [i_startblk, i_endblk], [slev,elev], [i_startidx_in, i_endidx_in] }
for (int block = this->i_startblk; block <= this->i_endblk; ++block) {
for (int level = this->slev; level < this->elev; ++level) {
@@ -336,9 +336,9 @@ TYPED_TEST(InterpolationScalarTypedTestFixture, AvgLib) {
this->nlev,
this->nblks_c,
this->lacc);
-
- // Check the outputs only for blocks in the range
+
+ // Check the outputs only for blocks in the range
// { [i_startblk, i_endblk], [slev,elev], [i_startidx_in, i_endidx_in] }
for (int block = this->i_startblk; block <= this->i_endblk; ++block) {
for (int level = this->slev; level < this->elev; ++level) {
@@ -439,7 +439,7 @@ TYPED_TEST(InterpolationScalarMixedTestFixture, cells2edges) {
this->lfill_latbc,
this->lacc);
- // Check the outputs only for blocks in the range
+ // Check the outputs only for blocks in the range
// { [i_startblk, i_endblk], [slev,elev], [i_startidx_in, i_endidx_in] }
for (int block = this->i_startblk; block <= this->i_endblk; ++block) {
for (int level = this->slev; level < this->elev; ++level) {
@@ -483,7 +483,7 @@ TYPED_TEST(InterpolationScalarMixedTestFixture, cells2verts) {
this->lacc,
this->acc_async);
- // Check the outputs only for blocks in the range
+ // Check the outputs only for blocks in the range
// { [i_startblk, i_endblk], [slev,elev], [i_startidx_in, i_endidx_in] }
for (int block = this->i_startblk; block <= this->i_endblk; ++block) {
for (int level = this->slev; level < this->elev; ++level) {
@@ -559,7 +559,7 @@ TYPED_TEST(Cells2vertsriScalarLibTestFixture, cells2verts_ri) {
this->lacc,
this->acc_async);
- // Check the outputs only for blocks in the range
+ // Check the outputs only for blocks in the range
// { [i_startblk, i_endblk], [slev,elev], [i_startidx_in, i_endidx_in] }
for (int block = this->i_startblk; block <= this->i_endblk; ++block) {
for (int level = this->slev; level < this->elev; ++level) {
--
GitLab
From 8eecaebb14a552b5d968996bf7da67b5d4bf2da0 Mon Sep 17 00:00:00 2001
From: Pradipta Samanta <samanta@dkrz.de>
Date: Wed, 5 Mar 2025 17:09:48 +0100
Subject: [PATCH 33/36] added some comments in the binding file
---
src/interpolation/interpolation_bindings.cpp | 32 +++++++++++++-------
1 file changed, 21 insertions(+), 11 deletions(-)
diff --git a/src/interpolation/interpolation_bindings.cpp b/src/interpolation/interpolation_bindings.cpp
index fc39ccd..dda576c 100644
--- a/src/interpolation/interpolation_bindings.cpp
+++ b/src/interpolation/interpolation_bindings.cpp
@@ -13,7 +13,7 @@
#include "mo_lib_interpolation_vector.hpp"
#include "mo_lib_interpolation_scalar.hpp"
-//mo_lib_interpolation_vector.F90
+// This is the binding for mo_interpolation_vector::edges2cells_vector_lib (wp=dp)
void edges2cells_vector_lib_dp(
const double* p_vn_in, const double* p_vt_in,
const int* cell_edge_idx, const int* cell_edge_blk,
@@ -36,6 +36,7 @@ void edges2cells_vector_lib_dp(
nlev, nblks_e, nblks_c);
}
+// This is the binding for mo_interpolation_vector::edges2cells_vector_lib (wp=sp)
void edges2cells_vector_lib_sp(
const float* p_vn_in, const float* p_vt_in,
const int* cell_edge_idx, const int* cell_edge_blk,
@@ -58,7 +59,7 @@ void edges2cells_vector_lib_sp(
nlev, nblks_e, nblks_c);
}
-//mo_lib_interpolation_scalar.F90
+// This is the binding for mo_interpolation_scalar::verts2edges_scalar_lib (wp=dp)
void verts2edges_scalar_lib_dp(const double* p_vertex_in,
const int* edge_vertex_idx, const int* edge_vertex_blk,
const double* coeff_int,
@@ -76,6 +77,7 @@ void verts2edges_scalar_lib_dp(const double* p_vertex_in,
}
+// This is the binding for mo_interpolation_scalar::verts2edges_scalar_lib (wp=sp)
void verts2edges_scalar_lib_sp(const float* p_vertex_in,
const int* edge_vertex_idx, const int* edge_vertex_blk,
const float* coeff_int,
@@ -93,6 +95,7 @@ void verts2edges_scalar_lib_sp(const float* p_vertex_in,
}
+// This is the binding for mo_interpolation_scalar::cells2edges_scalar_dp_lib
void cells2edges_scalar_lib_dp(const double* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
const double* coeff_int, double* p_edge_out, const int* i_startblk_in,
const int* i_endblk_in, const int* i_startidx_in, const int* i_endidx_in,
@@ -106,7 +109,7 @@ void cells2edges_scalar_lib_dp(const double* p_cell_in, const int* edge_cell_idx
}
-
+// This is the binding for mo_interpolation_scalar::cells2edges_scalar_sp_lib
void cells2edges_scalar_lib_sp(const float* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
const float* coeff_int, float* p_edge_out, const int* i_startblk_in,
const int* i_endblk_in, const int* i_startidx_in, const int* i_endidx_in,
@@ -120,6 +123,7 @@ void cells2edges_scalar_lib_sp(const float* p_cell_in, const int* edge_cell_idx,
}
+// This is the binding for mo_interpolation_scalar::cells2edges_scalar_sp2dp_lib
void cells2edges_scalar_lib_sp2dp(const float* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
const double* coeff_int, double* p_edge_out, const int* i_startblk_in,
const int* i_endblk_in, const int* i_startidx_in, const int* i_endidx_in,
@@ -133,6 +137,7 @@ void cells2edges_scalar_lib_sp2dp(const float* p_cell_in, const int* edge_cell_i
}
+// This is the binding for mo_interpolation_scalar::edges2verts_scalar_lib (wp=dp)
void edges2verts_scalar_lib_dp (const double* p_edge_in, const int* vert_edge_idx, const int* vert_edge_blk,
const double* v_int, double* p_vert_out, const int i_startblk, const int i_endblk,
const int i_startidx_in, const int i_endidx_in, const int slev,
@@ -143,6 +148,7 @@ void edges2verts_scalar_lib_dp (const double* p_edge_in, const int* vert_edge_id
i_endblk,i_startidx_in,i_endidx_in, slev, elev, nproma, nlev,nblks_e, nblks_v, lacc);
}
+// This is the binding for mo_interpolation_scalar::edges2verts_scalar_lib (wp=sp)
void edges2verts_scalar_lib_sp(const float* p_edge_in, const int* vert_edge_idx, const int* vert_edge_blk,
const float* v_int, float* p_vert_out, const int i_startblk, const int i_endblk,
const int i_startidx_in, const int i_endidx_in, const int slev,
@@ -154,6 +160,7 @@ void edges2verts_scalar_lib_sp(const float* p_edge_in, const int* vert_edge_idx,
}
+// This is the binding for mo_interpolation_scalar::edges2cells_scalar_dp_lib
void edges2cells_scalar_lib_dp(
const double* p_edge_in, const int* edge_idx, const int* edge_blk,
const double* coeff_int, double* p_cell_out, const int i_startblk, const int i_endblk,
@@ -171,6 +178,7 @@ void edges2cells_scalar_lib_dp(
lacc);
}
+// This is the binding for mo_interpolation_scalar::edges2cells_scalar_sp_lib
void edges2cells_scalar_lib_sp(
const float* p_edge_in, const int* edge_idx, const int* edge_blk,
const float* coeff_int, float* p_cell_out, const int i_startblk, const int i_endblk,
@@ -189,8 +197,7 @@ void edges2cells_scalar_lib_sp(
lacc);
}
-/////////////////////////////////////////////
-
+// This is the binding for mo_interpolation_scalar::cells2verts_scalar_dp_lib
void cells2verts_scalar_lib_dp(
const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
const double* coeff_int, double* p_vert_out, const int i_startblk, const int i_endblk,
@@ -208,6 +215,7 @@ void cells2verts_scalar_lib_dp(
lacc, acc_async);
}
+// This is the binding for mo_interpolation_scalar::cells2verts_scalar_dp2sp_lib
void cells2verts_scalar_lib_dp2sp(
const float* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
const double* coeff_int, double* p_vert_out, const int i_startblk, const int i_endblk,
@@ -225,6 +233,7 @@ void cells2verts_scalar_lib_dp2sp(
lacc, acc_async);
}
+// This is the binding for mo_interpolation_scalar::cells2verts_scalar_sp_lib
void cells2verts_scalar_lib_sp(
const float* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
@@ -242,8 +251,7 @@ void cells2verts_scalar_lib_sp(
lacc, acc_async);
}
-/////////////////////////////////////////////
-
+// This is the binding for mo_interpolation_scalar::cells2verts_scalar_ri_lib (wp=dp, vp=dp)
void cells2verts_scalar_ri_lib_dp(
const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
const double* coeff_int, double* p_vert_out, const int i_startblk, const int i_endblk,
@@ -261,6 +269,7 @@ void cells2verts_scalar_ri_lib_dp(
lacc, acc_async);
}
+// This is the binding for mo_interpolation_scalar::cells2verts_scalar_ri_lib (wp=dp, vp=sp)
void cells2verts_scalar_ri_lib_dp2sp(
const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
const double* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
@@ -278,6 +287,7 @@ void cells2verts_scalar_ri_lib_dp2sp(
lacc, acc_async);
}
+// This is the binding for mo_interpolation_scalar::cells2verts_scalar_ri_lib (wp=sp, vp=sp)
void cells2verts_scalar_ri_lib_sp(
const float* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
@@ -295,8 +305,7 @@ void cells2verts_scalar_ri_lib_sp(
lacc, acc_async);
}
-/////////////////////////////////////////////
-
+// This is the binding for mo_interpolation_scalar::verts2cells_scalar_lib (wp=dp)
void verts2cells_scalar_lib_dp(
const double* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
const double* coeff_int, double* p_cell_out, const int nblks_c, const int npromz_c,
@@ -310,6 +319,7 @@ void verts2cells_scalar_lib_dp(
nblks_v, lacc);
}
+// This is the binding for mo_interpolation_scalar::verts2cells_scalar_lib (wp=sp)
void verts2cells_scalar_lib_sp(
const float* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
const float* coeff_int, float* p_cell_out, const int nblks_c, const int npromz_c,
@@ -323,8 +333,7 @@ void verts2cells_scalar_lib_sp(
nblks_v, lacc);
}
-/////////////////////////////////////////////
-
+// This is the binding for mo_interpolation_scalar::cell_avg_lib (wp=dp)
void cell_avg_lib_dp(
const double* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
const double* avg_coeff, double* avg_psi_c, const int i_startblk, const int i_endblk,
@@ -342,6 +351,7 @@ void cell_avg_lib_dp(
lacc);
}
+// This is the binding for mo_interpolation_scalar::cell_avg_lib (wp=sp)
void cell_avg_lib_sp(
const float* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
const float* avg_coeff, float* avg_psi_c, const int i_startblk, const int i_endblk,
--
GitLab
From 60308a94a76d9669319856a53309b9688f345e62 Mon Sep 17 00:00:00 2001
From: Pradipta Samanta <samanta@dkrz.de>
Date: Wed, 5 Mar 2025 17:35:41 +0100
Subject: [PATCH 34/36] formatted the cpp code in interpolation_scalar_lib a
little
---
.../mo_lib_interpolation_scalar.cpp | 103 ++++++++++--------
1 file changed, 58 insertions(+), 45 deletions(-)
diff --git a/src/interpolation/mo_lib_interpolation_scalar.cpp b/src/interpolation/mo_lib_interpolation_scalar.cpp
index 7f37448..b2b51ac 100644
--- a/src/interpolation/mo_lib_interpolation_scalar.cpp
+++ b/src/interpolation/mo_lib_interpolation_scalar.cpp
@@ -54,15 +54,17 @@ void verts2edges_scalar_lib(const T* p_vertex_in,
jb, i_startblk, i_endblk,
i_startidx, i_endidx);
- Kokkos::parallel_for("verts2edges_scalar", Kokkos::MDRangePolicy<Kokkos::Rank<2>>
- ({slev, i_startidx}, {elev + 1, i_endidx + 1}),
- KOKKOS_LAMBDA(const int jk, const int je) {
+ Kokkos::MDRangePolicy<Kokkos::Rank<2>> innerPolicy(
+ {slev, i_startidx}, {elev + 1, i_endidx + 1});
+
+ Kokkos::parallel_for("verts2edges_scalar", innerPolicy,
+ KOKKOS_LAMBDA(const int jk, const int je) {
p_edge_out_view(je, jk, jb) = coeff_int_view(je, 0, jb) *
p_vertex_in_view(iidx_view(je, jb, 0), jk, iblk_view(je, jb, 0)) +
coeff_int_view(je, 1, jb)*p_vertex_in_view(iidx_view(je, jb, 1), jk, iblk_view(je, jb, 1));
-
- });
+ }
+ );
Kokkos::fence();
}
}
@@ -106,21 +108,24 @@ void cells2edges_scalar_lib(const T* p_cell_in, const int* edge_cell_idx, const
get_indices_e_lib(i_startidx_in[0], i_endidx_in[0], nproma, jb,
i_startblk, i_endblk, i_startidx, i_endidx);
- Kokkos::parallel_for("cells2edges_scalar", Kokkos::MDRangePolicy<Kokkos::Rank<2>>
- ({slev, i_startidx}, {elev + 1, i_endidx + 1}),
- KOKKOS_LAMBDA(const int jk, const int je) {
-
- if (iidx_view(je, jb, 0) >= 0 && iblk_view(je, jb, 0) >= 0){
- p_edge_out_view(je, jk, jb) = p_cell_in_view(iidx_view(je, jb, 0), jk, iblk_view(je, jb, 0));
- }
- else if (iidx_view(je, jb, 1) >= 0 && iblk_view(je, jb, 1) >= 0){
- p_edge_out_view(je, jk, jb) = p_cell_in_view(iidx_view(je, jb, 1), jk, iblk_view(je, jb, 1));
+ Kokkos::MDRangePolicy<Kokkos::Rank<2>> innerPolicy(
+ {slev, i_startidx}, {elev + 1, i_endidx + 1});
+
+ Kokkos::parallel_for("cells2edges_scalar", innerPolicy,
+ KOKKOS_LAMBDA(const int jk, const int je) {
+
+ if (iidx_view(je, jb, 0) >= 0 && iblk_view(je, jb, 0) >= 0){
+ p_edge_out_view(je, jk, jb) = p_cell_in_view(iidx_view(je, jb, 0), jk, iblk_view(je, jb, 0));
+ }
+ else if (iidx_view(je, jb, 1) >= 0 && iblk_view(je, jb, 1) >= 0){
+ p_edge_out_view(je, jk, jb) = p_cell_in_view(iidx_view(je, jb, 1), jk, iblk_view(je, jb, 1));
+ }
+ else{
+ std::cerr << "mo_interpolation:cells2edges_scalar_lib: error in lateral boundary filling" << std::endl;
+ std::exit(EXIT_FAILURE);
+ }
}
- else{
- std::cerr << "mo_interpolation:cells2edges_scalar_lib: error in lateral boundary filling" << std::endl;
- std::exit(EXIT_FAILURE);
- }
- });
+ );
Kokkos::fence();
}
}
@@ -135,15 +140,16 @@ void cells2edges_scalar_lib(const T* p_cell_in, const int* edge_cell_idx, const
get_indices_e_lib(i_startidx_in[1], i_endidx_in[1], nproma, jb,
i_startblk, i_endblk, i_startidx, i_endidx);
- Kokkos::parallel_for("cells2edges_scalar", Kokkos::MDRangePolicy<Kokkos::Rank<2>>
- ({slev, i_startidx}, {elev + 1, i_endidx + 1}),
- KOKKOS_LAMBDA(const int jk, const int je) {
+ Kokkos::MDRangePolicy<Kokkos::Rank<2>> innerPolicy(
+ {slev, i_startidx}, {elev + 1, i_endidx + 1});
+ Kokkos::parallel_for("cells2edges_scalar", innerPolicy,
+ KOKKOS_LAMBDA(const int jk, const int je) {
p_edge_out_view(je, jk, jb) = coeff_int_view(je, 0, jb) *
p_cell_in_view(iidx_view(je, jb, 0), jk, iblk_view(je, jb, 0)) +
coeff_int_view(je, 1, jb) * p_cell_in_view(iidx_view(je, jb, 1), jk, iblk_view(je, jb, 1));
-
- });
+ }
+ );
Kokkos::fence();
}
}
@@ -182,17 +188,19 @@ void edges2verts_scalar_lib (const T* p_edge_in, const int* vert_edge_idx, const
jb, i_startblk, i_endblk,
i_startidx, i_endidx);
- Kokkos::parallel_for("edges2verts_scalar", Kokkos::MDRangePolicy<Kokkos::Rank<2>>
- ({slev, i_startidx}, {elev + 1, i_endidx + 1}),
- KOKKOS_LAMBDA(const int jk, const int jv) {
-
- p_vert_out_view(jv, jk, jb) = v_int_view(jv, 0, jb)*p_edge_in_view(iidx_view(jv, jb, 0), jk, iblk_view(jv, jb, 0))
- + v_int_view(jv, 1, jb)*p_edge_in_view(iidx_view(jv, jb, 1), jk, iblk_view(jv, jb, 1))
- + v_int_view(jv, 2, jb)*p_edge_in_view(iidx_view(jv, jb, 2), jk, iblk_view(jv, jb, 2))
- + v_int_view(jv, 3, jb)*p_edge_in_view(iidx_view(jv, jb, 3), jk, iblk_view(jv, jb, 3))
- + v_int_view(jv, 4, jb)*p_edge_in_view(iidx_view(jv, jb, 4), jk, iblk_view(jv, jb, 4))
- + v_int_view(jv, 5, jb)*p_edge_in_view(iidx_view(jv, jb, 5), jk, iblk_view(jv, jb, 5));
- });
+ Kokkos::MDRangePolicy<Kokkos::Rank<2>> innerPolicy(
+ {slev, i_startidx}, {elev + 1, i_endidx + 1});
+
+ Kokkos::parallel_for("edges2verts_scalar", innerPolicy,
+ KOKKOS_LAMBDA(const int jk, const int jv) {
+ p_vert_out_view(jv, jk, jb) = v_int_view(jv, 0, jb)*p_edge_in_view(iidx_view(jv, jb, 0), jk, iblk_view(jv, jb, 0))
+ + v_int_view(jv, 1, jb)*p_edge_in_view(iidx_view(jv, jb, 1), jk, iblk_view(jv, jb, 1))
+ + v_int_view(jv, 2, jb)*p_edge_in_view(iidx_view(jv, jb, 2), jk, iblk_view(jv, jb, 2))
+ + v_int_view(jv, 3, jb)*p_edge_in_view(iidx_view(jv, jb, 3), jk, iblk_view(jv, jb, 3))
+ + v_int_view(jv, 4, jb)*p_edge_in_view(iidx_view(jv, jb, 4), jk, iblk_view(jv, jb, 4))
+ + v_int_view(jv, 5, jb)*p_edge_in_view(iidx_view(jv, jb, 5), jk, iblk_view(jv, jb, 5));
+ }
+ );
Kokkos::fence();
}
@@ -245,11 +253,12 @@ void edges2cells_scalar_lib(
Kokkos::parallel_for("edges2cells_scalar_lib_inner", innerPolicy,
KOKKOS_LAMBDA(const int jk, const int jc) {
- p_cell_out_view(jc, jk, jb) =
- coeff_int_view(jc, 0, jb)*p_edge_in_view(iidx_view(jc, jb, 0), jk, iblk_view(jc, jb, 0)) +
- coeff_int_view(jc, 1, jb)*p_edge_in_view(iidx_view(jc, jb, 1), jk, iblk_view(jc, jb, 1)) +
- coeff_int_view(jc, 2, jb)*p_edge_in_view(iidx_view(jc, jb, 2), jk, iblk_view(jc, jb, 2));
- });
+ p_cell_out_view(jc, jk, jb) =
+ coeff_int_view(jc, 0, jb)*p_edge_in_view(iidx_view(jc, jb, 0), jk, iblk_view(jc, jb, 0)) +
+ coeff_int_view(jc, 1, jb)*p_edge_in_view(iidx_view(jc, jb, 1), jk, iblk_view(jc, jb, 1)) +
+ coeff_int_view(jc, 2, jb)*p_edge_in_view(iidx_view(jc, jb, 2), jk, iblk_view(jc, jb, 2));
+ }
+ );
}
}
@@ -309,7 +318,8 @@ void cells2verts_scalar_lib(
coeff_int_view(jv, 4, jb)*p_cell_in_view(iidx_view(jv, jb, 4), jk, iblk_view(jv, jb, 4)) +
coeff_int_view(jv, 5, jb)*p_cell_in_view(iidx_view(jv, jb, 5), jk, iblk_view(jv, jb, 5));
- });
+ }
+ );
}
}
@@ -376,7 +386,8 @@ void cells2verts_scalar_ri_lib(
coeff_int_view(jv, 4, jb)*p_cell_in_view(iidx_view(jv, jb, 4), jk, iblk_view(jv, jb, 4)) +
coeff_int_view(jv, 5, jb)*p_cell_in_view(iidx_view(jv, jb, 5), jk, iblk_view(jv, jb, 5));
- });
+ }
+ );
}
}
@@ -429,7 +440,8 @@ void verts2cells_scalar_lib(
coeff_int_view(jc, 1, jb)*p_vert_in_view(iidx_view(jc, jb, 1), jk, iblk_view(jc, jb, 1)) +
coeff_int_view(jc, 2, jb)*p_vert_in_view(iidx_view(jc, jb, 2), jk, iblk_view(jc, jb, 2));
- });
+ }
+ );
}
}
@@ -488,8 +500,9 @@ void cell_avg_lib(
psi_c_view(iidx_view(jc, jb, 0), jk, iblk_view(jc, jb, 0))*avg_coeff_view(jc, 1, jb) +
psi_c_view(iidx_view(jc, jb, 1), jk, iblk_view(jc, jb, 1))*avg_coeff_view(jc, 2, jb) +
psi_c_view(iidx_view(jc, jb, 2), jk, iblk_view(jc, jb, 2))*avg_coeff_view(jc, 3, jb);
- });
- }
+ }
+ );
+ }
}
//-----------------------------------------------------------------------
--
GitLab
From 853b97cec3f0240b932f9a3dbcb7d132ab9d204c Mon Sep 17 00:00:00 2001
From: Pradipta Samanta <samanta@dkrz.de>
Date: Wed, 5 Mar 2025 17:37:31 +0100
Subject: [PATCH 35/36] added the missing kokkos::fence
---
src/interpolation/mo_lib_interpolation_scalar.cpp | 5 +++++
1 file changed, 5 insertions(+)
diff --git a/src/interpolation/mo_lib_interpolation_scalar.cpp b/src/interpolation/mo_lib_interpolation_scalar.cpp
index b2b51ac..386ce0c 100644
--- a/src/interpolation/mo_lib_interpolation_scalar.cpp
+++ b/src/interpolation/mo_lib_interpolation_scalar.cpp
@@ -259,6 +259,7 @@ void edges2cells_scalar_lib(
coeff_int_view(jc, 2, jb)*p_edge_in_view(iidx_view(jc, jb, 2), jk, iblk_view(jc, jb, 2));
}
);
+ Kokkos::fence();
}
}
@@ -320,6 +321,7 @@ void cells2verts_scalar_lib(
}
);
+ Kokkos::fence();
}
}
@@ -388,6 +390,7 @@ void cells2verts_scalar_ri_lib(
}
);
+ Kokkos::fence();
}
}
@@ -442,6 +445,7 @@ void verts2cells_scalar_lib(
}
);
+ Kokkos::fence();
}
}
@@ -502,6 +506,7 @@ void cell_avg_lib(
psi_c_view(iidx_view(jc, jb, 2), jk, iblk_view(jc, jb, 2))*avg_coeff_view(jc, 3, jb);
}
);
+ Kokkos::fence();
}
}
--
GitLab
From baab213a5a00d74a58c42696760e1fa8d8cef548 Mon Sep 17 00:00:00 2001
From: Pradipta Samanta <samanta@dkrz.de>
Date: Wed, 5 Mar 2025 17:57:07 +0100
Subject: [PATCH 36/36] formatted the cpp codes using clang-format
---
src/interpolation/interpolation_bindings.cpp | 553 ++++----
src/interpolation/interpolation_bindings.h | 348 ++---
.../mo_lib_interpolation_scalar.cpp | 1169 +++++++++--------
.../mo_lib_interpolation_scalar.hpp | 94 +-
.../mo_lib_interpolation_vector.cpp | 175 +--
.../mo_lib_interpolation_vector.hpp | 21 +-
test/c/test_interpolation_scalar.cpp | 493 ++++---
test/c/test_interpolation_vector.cpp | 70 +-
8 files changed, 1456 insertions(+), 1467 deletions(-)
diff --git a/src/interpolation/interpolation_bindings.cpp b/src/interpolation/interpolation_bindings.cpp
index dda576c..628f411 100644
--- a/src/interpolation/interpolation_bindings.cpp
+++ b/src/interpolation/interpolation_bindings.cpp
@@ -10,362 +10,319 @@
// ---------------------------------------------------------------
#include "interpolation_bindings.h"
-#include "mo_lib_interpolation_vector.hpp"
#include "mo_lib_interpolation_scalar.hpp"
+#include "mo_lib_interpolation_vector.hpp"
-// This is the binding for mo_interpolation_vector::edges2cells_vector_lib (wp=dp)
-void edges2cells_vector_lib_dp(
- const double* p_vn_in, const double* p_vt_in,
- const int* cell_edge_idx, const int* cell_edge_blk,
- const double* e_bln_c_u, const double* e_bln_c_v,
- double* p_u_out, double* p_v_out,
- int i_startblk, int i_endblk,
- int i_startidx_in, int i_endidx_in,
- int slev, int elev,
- int nproma,
- int nlev, int nblks_e, int nblks_c){
-
- edges2cells_vector_lib<double>(p_vn_in, p_vt_in,
- cell_edge_idx, cell_edge_blk,
- e_bln_c_u, e_bln_c_v,
- p_u_out, p_v_out,
- i_startblk, i_endblk,
- i_startidx_in, i_endidx_in,
- slev, elev,
- nproma,
- nlev, nblks_e, nblks_c);
+// This is the binding for mo_interpolation_vector::edges2cells_vector_lib
+// (wp=dp)
+void edges2cells_vector_lib_dp(const double *p_vn_in, const double *p_vt_in,
+ const int *cell_edge_idx,
+ const int *cell_edge_blk,
+ const double *e_bln_c_u, const double *e_bln_c_v,
+ double *p_u_out, double *p_v_out, int i_startblk,
+ int i_endblk, int i_startidx_in, int i_endidx_in,
+ int slev, int elev, int nproma, int nlev,
+ int nblks_e, int nblks_c) {
+
+ edges2cells_vector_lib<double>(
+ p_vn_in, p_vt_in, cell_edge_idx, cell_edge_blk, e_bln_c_u, e_bln_c_v,
+ p_u_out, p_v_out, i_startblk, i_endblk, i_startidx_in, i_endidx_in, slev,
+ elev, nproma, nlev, nblks_e, nblks_c);
}
-// This is the binding for mo_interpolation_vector::edges2cells_vector_lib (wp=sp)
-void edges2cells_vector_lib_sp(
- const float* p_vn_in, const float* p_vt_in,
- const int* cell_edge_idx, const int* cell_edge_blk,
- const float* e_bln_c_u, const float* e_bln_c_v,
- float* p_u_out, float* p_v_out,
- int i_startblk, int i_endblk,
- int i_startidx_in, int i_endidx_in,
- int slev, int elev,
- int nproma,
- int nlev, int nblks_e, int nblks_c){
-
- edges2cells_vector_lib<float>(p_vn_in, p_vt_in,
- cell_edge_idx, cell_edge_blk,
- e_bln_c_u, e_bln_c_v,
- p_u_out, p_v_out,
- i_startblk, i_endblk,
- i_startidx_in, i_endidx_in,
- slev, elev,
- nproma,
- nlev, nblks_e, nblks_c);
+// This is the binding for mo_interpolation_vector::edges2cells_vector_lib
+// (wp=sp)
+void edges2cells_vector_lib_sp(const float *p_vn_in, const float *p_vt_in,
+ const int *cell_edge_idx,
+ const int *cell_edge_blk, const float *e_bln_c_u,
+ const float *e_bln_c_v, float *p_u_out,
+ float *p_v_out, int i_startblk, int i_endblk,
+ int i_startidx_in, int i_endidx_in, int slev,
+ int elev, int nproma, int nlev, int nblks_e,
+ int nblks_c) {
+
+ edges2cells_vector_lib<float>(
+ p_vn_in, p_vt_in, cell_edge_idx, cell_edge_blk, e_bln_c_u, e_bln_c_v,
+ p_u_out, p_v_out, i_startblk, i_endblk, i_startidx_in, i_endidx_in, slev,
+ elev, nproma, nlev, nblks_e, nblks_c);
}
-// This is the binding for mo_interpolation_scalar::verts2edges_scalar_lib (wp=dp)
-void verts2edges_scalar_lib_dp(const double* p_vertex_in,
- const int* edge_vertex_idx, const int* edge_vertex_blk,
- const double* coeff_int,
- double* p_edge_out,
- const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
- const int nproma, const int nlev, const int nblks_v, const int nblks_e, const bool lacc){
-
- verts2edges_scalar_lib<double>(p_vertex_in,
- edge_vertex_idx, edge_vertex_blk, coeff_int,
- p_edge_out,
- i_startblk, i_endblk,
- i_startidx_in, i_endidx_in, slev, elev,
- nproma, nlev, nblks_v, nblks_e, lacc);
-
+// This is the binding for mo_interpolation_scalar::verts2edges_scalar_lib
+// (wp=dp)
+void verts2edges_scalar_lib_dp(
+ const double *p_vertex_in, const int *edge_vertex_idx,
+ const int *edge_vertex_blk, const double *coeff_int, double *p_edge_out,
+ const int i_startblk, const int i_endblk, const int i_startidx_in,
+ const int i_endidx_in, const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_v, const int nblks_e, const bool lacc) {
+
+ verts2edges_scalar_lib<double>(p_vertex_in, edge_vertex_idx, edge_vertex_blk,
+ coeff_int, p_edge_out, i_startblk, i_endblk,
+ i_startidx_in, i_endidx_in, slev, elev, nproma,
+ nlev, nblks_v, nblks_e, lacc);
}
-// This is the binding for mo_interpolation_scalar::verts2edges_scalar_lib (wp=sp)
-void verts2edges_scalar_lib_sp(const float* p_vertex_in,
- const int* edge_vertex_idx, const int* edge_vertex_blk,
- const float* coeff_int,
- float* p_edge_out,
- const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
- const int nproma, const int nlev, const int nblks_v, const int nblks_e, const bool lacc){
-
- verts2edges_scalar_lib<float>(p_vertex_in,
- edge_vertex_idx, edge_vertex_blk, coeff_int,
- p_edge_out,
- i_startblk, i_endblk,
- i_startidx_in, i_endidx_in, slev, elev,
- nproma, nlev, nblks_v, nblks_e, lacc);
-
+// This is the binding for mo_interpolation_scalar::verts2edges_scalar_lib
+// (wp=sp)
+void verts2edges_scalar_lib_sp(
+ const float *p_vertex_in, const int *edge_vertex_idx,
+ const int *edge_vertex_blk, const float *coeff_int, float *p_edge_out,
+ const int i_startblk, const int i_endblk, const int i_startidx_in,
+ const int i_endidx_in, const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_v, const int nblks_e, const bool lacc) {
+
+ verts2edges_scalar_lib<float>(p_vertex_in, edge_vertex_idx, edge_vertex_blk,
+ coeff_int, p_edge_out, i_startblk, i_endblk,
+ i_startidx_in, i_endidx_in, slev, elev, nproma,
+ nlev, nblks_v, nblks_e, lacc);
}
// This is the binding for mo_interpolation_scalar::cells2edges_scalar_dp_lib
-void cells2edges_scalar_lib_dp(const double* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
- const double* coeff_int, double* p_edge_out, const int* i_startblk_in,
- const int* i_endblk_in, const int* i_startidx_in, const int* i_endidx_in,
- const int slev, const int elev, const int nproma, const int nlev,
- const int nblk_c, const int nblks_e, const int patch_id,
- const bool l_limited_area, const bool lfill_latbc, const bool lacc){
-
- cells2edges_scalar_lib<double, double>(p_cell_in,edge_cell_idx,edge_cell_blk,coeff_int,p_edge_out,i_startblk_in,
- i_endblk_in,i_startidx_in,i_endidx_in,slev,elev,nproma,nlev,
- nblk_c,nblks_e,patch_id,l_limited_area,lfill_latbc,lacc);
-
+void cells2edges_scalar_lib_dp(
+ const double *p_cell_in, const int *edge_cell_idx, const int *edge_cell_blk,
+ const double *coeff_int, double *p_edge_out, const int *i_startblk_in,
+ const int *i_endblk_in, const int *i_startidx_in, const int *i_endidx_in,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblk_c, const int nblks_e, const int patch_id,
+ const bool l_limited_area, const bool lfill_latbc, const bool lacc) {
+
+ cells2edges_scalar_lib<double, double>(
+ p_cell_in, edge_cell_idx, edge_cell_blk, coeff_int, p_edge_out,
+ i_startblk_in, i_endblk_in, i_startidx_in, i_endidx_in, slev, elev,
+ nproma, nlev, nblk_c, nblks_e, patch_id, l_limited_area, lfill_latbc,
+ lacc);
}
// This is the binding for mo_interpolation_scalar::cells2edges_scalar_sp_lib
-void cells2edges_scalar_lib_sp(const float* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
- const float* coeff_int, float* p_edge_out, const int* i_startblk_in,
- const int* i_endblk_in, const int* i_startidx_in, const int* i_endidx_in,
- const int slev, const int elev, const int nproma, const int nlev,
- const int nblk_c, const int nblks_e, const int patch_id,
- const bool l_limited_area, const bool lfill_latbc, const bool lacc){
-
- cells2edges_scalar_lib<float, float>(p_cell_in,edge_cell_idx,edge_cell_blk,coeff_int,p_edge_out,i_startblk_in,
- i_endblk_in,i_startidx_in,i_endidx_in,slev,elev,nproma,nlev,
- nblk_c,nblks_e,patch_id,l_limited_area,lfill_latbc,lacc);
-
+void cells2edges_scalar_lib_sp(const float *p_cell_in, const int *edge_cell_idx,
+ const int *edge_cell_blk, const float *coeff_int,
+ float *p_edge_out, const int *i_startblk_in,
+ const int *i_endblk_in, const int *i_startidx_in,
+ const int *i_endidx_in, const int slev,
+ const int elev, const int nproma, const int nlev,
+ const int nblk_c, const int nblks_e,
+ const int patch_id, const bool l_limited_area,
+ const bool lfill_latbc, const bool lacc) {
+
+ cells2edges_scalar_lib<float, float>(
+ p_cell_in, edge_cell_idx, edge_cell_blk, coeff_int, p_edge_out,
+ i_startblk_in, i_endblk_in, i_startidx_in, i_endidx_in, slev, elev,
+ nproma, nlev, nblk_c, nblks_e, patch_id, l_limited_area, lfill_latbc,
+ lacc);
}
// This is the binding for mo_interpolation_scalar::cells2edges_scalar_sp2dp_lib
-void cells2edges_scalar_lib_sp2dp(const float* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
- const double* coeff_int, double* p_edge_out, const int* i_startblk_in,
- const int* i_endblk_in, const int* i_startidx_in, const int* i_endidx_in,
- const int slev, const int elev, const int nproma, const int nlev,
- const int nblk_c, const int nblks_e, const int patch_id,
- const bool l_limited_area, const bool lfill_latbc, const bool lacc){
-
- cells2edges_scalar_lib<float, double>(p_cell_in,edge_cell_idx,edge_cell_blk,coeff_int,p_edge_out,i_startblk_in,
- i_endblk_in,i_startidx_in,i_endidx_in,slev,elev,nproma,nlev,
- nblk_c,nblks_e,patch_id,l_limited_area,lfill_latbc,lacc);
-
+void cells2edges_scalar_lib_sp2dp(
+ const float *p_cell_in, const int *edge_cell_idx, const int *edge_cell_blk,
+ const double *coeff_int, double *p_edge_out, const int *i_startblk_in,
+ const int *i_endblk_in, const int *i_startidx_in, const int *i_endidx_in,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblk_c, const int nblks_e, const int patch_id,
+ const bool l_limited_area, const bool lfill_latbc, const bool lacc) {
+
+ cells2edges_scalar_lib<float, double>(
+ p_cell_in, edge_cell_idx, edge_cell_blk, coeff_int, p_edge_out,
+ i_startblk_in, i_endblk_in, i_startidx_in, i_endidx_in, slev, elev,
+ nproma, nlev, nblk_c, nblks_e, patch_id, l_limited_area, lfill_latbc,
+ lacc);
}
-// This is the binding for mo_interpolation_scalar::edges2verts_scalar_lib (wp=dp)
-void edges2verts_scalar_lib_dp (const double* p_edge_in, const int* vert_edge_idx, const int* vert_edge_blk,
- const double* v_int, double* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in, const int slev,
- const int elev, const int nproma, const int nlev, const int nblks_e,
- const int nblks_v, const bool lacc){
+// This is the binding for mo_interpolation_scalar::edges2verts_scalar_lib
+// (wp=dp)
+void edges2verts_scalar_lib_dp(
+ const double *p_edge_in, const int *vert_edge_idx, const int *vert_edge_blk,
+ const double *v_int, double *p_vert_out, const int i_startblk,
+ const int i_endblk, const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblks_e, const int nblks_v, const bool lacc) {
- edges2verts_scalar_lib<double>(p_edge_in, vert_edge_idx, vert_edge_blk, v_int,p_vert_out,i_startblk,
- i_endblk,i_startidx_in,i_endidx_in, slev, elev, nproma, nlev,nblks_e, nblks_v, lacc);
+ edges2verts_scalar_lib<double>(p_edge_in, vert_edge_idx, vert_edge_blk, v_int,
+ p_vert_out, i_startblk, i_endblk,
+ i_startidx_in, i_endidx_in, slev, elev, nproma,
+ nlev, nblks_e, nblks_v, lacc);
}
-// This is the binding for mo_interpolation_scalar::edges2verts_scalar_lib (wp=sp)
-void edges2verts_scalar_lib_sp(const float* p_edge_in, const int* vert_edge_idx, const int* vert_edge_blk,
- const float* v_int, float* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in, const int slev,
- const int elev, const int nproma, const int nlev, const int nblks_e,
- const int nblks_v, const bool lacc){
-
- edges2verts_scalar_lib<float>(p_edge_in, vert_edge_idx, vert_edge_blk, v_int,p_vert_out,i_startblk,
- i_endblk,i_startidx_in,i_endidx_in, slev, elev, nproma, nlev,nblks_e, nblks_v, lacc);
-
+// This is the binding for mo_interpolation_scalar::edges2verts_scalar_lib
+// (wp=sp)
+void edges2verts_scalar_lib_sp(const float *p_edge_in, const int *vert_edge_idx,
+ const int *vert_edge_blk, const float *v_int,
+ float *p_vert_out, const int i_startblk,
+ const int i_endblk, const int i_startidx_in,
+ const int i_endidx_in, const int slev,
+ const int elev, const int nproma, const int nlev,
+ const int nblks_e, const int nblks_v,
+ const bool lacc) {
+
+ edges2verts_scalar_lib<float>(p_edge_in, vert_edge_idx, vert_edge_blk, v_int,
+ p_vert_out, i_startblk, i_endblk, i_startidx_in,
+ i_endidx_in, slev, elev, nproma, nlev, nblks_e,
+ nblks_v, lacc);
}
// This is the binding for mo_interpolation_scalar::edges2cells_scalar_dp_lib
-void edges2cells_scalar_lib_dp(
- const double* p_edge_in, const int* edge_idx, const int* edge_blk,
- const double* coeff_int, double* p_cell_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_e, const int nblks_c,
- const bool lacc)
-{
- edges2cells_scalar_lib<double>(
- p_edge_in, edge_idx, edge_blk,
- coeff_int, p_cell_out, i_startblk, i_endblk,
- i_startidx_in, i_endidx_in,
- slev, elev, nproma,
- nlev, nblks_e, nblks_c,
- lacc);
+void edges2cells_scalar_lib_dp(const double *p_edge_in, const int *edge_idx,
+ const int *edge_blk, const double *coeff_int,
+ double *p_cell_out, const int i_startblk,
+ const int i_endblk, const int i_startidx_in,
+ const int i_endidx_in, const int slev,
+ const int elev, const int nproma, const int nlev,
+ const int nblks_e, const int nblks_c,
+ const bool lacc) {
+ edges2cells_scalar_lib<double>(p_edge_in, edge_idx, edge_blk, coeff_int,
+ p_cell_out, i_startblk, i_endblk,
+ i_startidx_in, i_endidx_in, slev, elev, nproma,
+ nlev, nblks_e, nblks_c, lacc);
}
// This is the binding for mo_interpolation_scalar::edges2cells_scalar_sp_lib
-void edges2cells_scalar_lib_sp(
- const float* p_edge_in, const int* edge_idx, const int* edge_blk,
- const float* coeff_int, float* p_cell_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_e, const int nblks_c,
- const bool lacc)
-{
-
- edges2cells_scalar_lib<float>(
- p_edge_in, edge_idx, edge_blk,
- coeff_int, p_cell_out, i_startblk, i_endblk,
- i_startidx_in, i_endidx_in,
- slev, elev, nproma,
- nlev, nblks_e, nblks_c,
- lacc);
+void edges2cells_scalar_lib_sp(const float *p_edge_in, const int *edge_idx,
+ const int *edge_blk, const float *coeff_int,
+ float *p_cell_out, const int i_startblk,
+ const int i_endblk, const int i_startidx_in,
+ const int i_endidx_in, const int slev,
+ const int elev, const int nproma, const int nlev,
+ const int nblks_e, const int nblks_c,
+ const bool lacc) {
+
+ edges2cells_scalar_lib<float>(p_edge_in, edge_idx, edge_blk, coeff_int,
+ p_cell_out, i_startblk, i_endblk, i_startidx_in,
+ i_endidx_in, slev, elev, nproma, nlev, nblks_e,
+ nblks_c, lacc);
}
// This is the binding for mo_interpolation_scalar::cells2verts_scalar_dp_lib
void cells2verts_scalar_lib_dp(
- const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const double* coeff_int, double* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async)
-{
- cells2verts_scalar_lib<double,double>(
- p_cell_in, vert_cell_idx, vert_cell_blk,
- coeff_int, p_vert_out, i_startblk, i_endblk,
- i_startidx_in, i_endidx_in,
- slev, elev, nproma,
- nlev, nblks_c, nblks_v,
- lacc, acc_async);
+ const double *p_cell_in, const int *vert_cell_idx, const int *vert_cell_blk,
+ const double *coeff_int, double *p_vert_out, const int i_startblk,
+ const int i_endblk, const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblks_c, const int nblks_v, const bool lacc,
+ const bool acc_async) {
+ cells2verts_scalar_lib<double, double>(
+ p_cell_in, vert_cell_idx, vert_cell_blk, coeff_int, p_vert_out,
+ i_startblk, i_endblk, i_startidx_in, i_endidx_in, slev, elev, nproma,
+ nlev, nblks_c, nblks_v, lacc, acc_async);
}
// This is the binding for mo_interpolation_scalar::cells2verts_scalar_dp2sp_lib
void cells2verts_scalar_lib_dp2sp(
- const float* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const double* coeff_int, double* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async)
-{
- cells2verts_scalar_lib<float,double>(
- p_cell_in, vert_cell_idx, vert_cell_blk,
- coeff_int, p_vert_out, i_startblk, i_endblk,
- i_startidx_in, i_endidx_in,
- slev, elev, nproma,
- nlev, nblks_c, nblks_v,
- lacc, acc_async);
+ const float *p_cell_in, const int *vert_cell_idx, const int *vert_cell_blk,
+ const double *coeff_int, double *p_vert_out, const int i_startblk,
+ const int i_endblk, const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblks_c, const int nblks_v, const bool lacc,
+ const bool acc_async) {
+ cells2verts_scalar_lib<float, double>(
+ p_cell_in, vert_cell_idx, vert_cell_blk, coeff_int, p_vert_out,
+ i_startblk, i_endblk, i_startidx_in, i_endidx_in, slev, elev, nproma,
+ nlev, nblks_c, nblks_v, lacc, acc_async);
}
// This is the binding for mo_interpolation_scalar::cells2verts_scalar_sp_lib
-void cells2verts_scalar_lib_sp(
- const float* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async)
-{
- cells2verts_scalar_lib<float,float>(
- p_cell_in, vert_cell_idx, vert_cell_blk,
- coeff_int, p_vert_out, i_startblk, i_endblk,
- i_startidx_in, i_endidx_in,
- slev, elev, nproma,
- nlev, nblks_c, nblks_v,
- lacc, acc_async);
+void cells2verts_scalar_lib_sp(const float *p_cell_in, const int *vert_cell_idx,
+ const int *vert_cell_blk, const float *coeff_int,
+ float *p_vert_out, const int i_startblk,
+ const int i_endblk, const int i_startidx_in,
+ const int i_endidx_in, const int slev,
+ const int elev, const int nproma, const int nlev,
+ const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async) {
+ cells2verts_scalar_lib<float, float>(
+ p_cell_in, vert_cell_idx, vert_cell_blk, coeff_int, p_vert_out,
+ i_startblk, i_endblk, i_startidx_in, i_endidx_in, slev, elev, nproma,
+ nlev, nblks_c, nblks_v, lacc, acc_async);
}
-// This is the binding for mo_interpolation_scalar::cells2verts_scalar_ri_lib (wp=dp, vp=dp)
+// This is the binding for mo_interpolation_scalar::cells2verts_scalar_ri_lib
+// (wp=dp, vp=dp)
void cells2verts_scalar_ri_lib_dp(
- const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const double* coeff_int, double* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async)
-{
- cells2verts_scalar_ri_lib<double,double>(
- p_cell_in, vert_cell_idx, vert_cell_blk,
- coeff_int, p_vert_out, i_startblk, i_endblk,
- i_startidx_in, i_endidx_in,
- slev, elev, nproma,
- nlev, nblks_c, nblks_v,
- lacc, acc_async);
+ const double *p_cell_in, const int *vert_cell_idx, const int *vert_cell_blk,
+ const double *coeff_int, double *p_vert_out, const int i_startblk,
+ const int i_endblk, const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblks_c, const int nblks_v, const bool lacc,
+ const bool acc_async) {
+ cells2verts_scalar_ri_lib<double, double>(
+ p_cell_in, vert_cell_idx, vert_cell_blk, coeff_int, p_vert_out,
+ i_startblk, i_endblk, i_startidx_in, i_endidx_in, slev, elev, nproma,
+ nlev, nblks_c, nblks_v, lacc, acc_async);
}
-// This is the binding for mo_interpolation_scalar::cells2verts_scalar_ri_lib (wp=dp, vp=sp)
+// This is the binding for mo_interpolation_scalar::cells2verts_scalar_ri_lib
+// (wp=dp, vp=sp)
void cells2verts_scalar_ri_lib_dp2sp(
- const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const double* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async)
-{
- cells2verts_scalar_ri_lib<double,float>(
- p_cell_in, vert_cell_idx, vert_cell_blk,
- coeff_int, p_vert_out, i_startblk, i_endblk,
- i_startidx_in, i_endidx_in,
- slev, elev, nproma,
- nlev, nblks_c, nblks_v,
- lacc, acc_async);
+ const double *p_cell_in, const int *vert_cell_idx, const int *vert_cell_blk,
+ const double *coeff_int, float *p_vert_out, const int i_startblk,
+ const int i_endblk, const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblks_c, const int nblks_v, const bool lacc,
+ const bool acc_async) {
+ cells2verts_scalar_ri_lib<double, float>(
+ p_cell_in, vert_cell_idx, vert_cell_blk, coeff_int, p_vert_out,
+ i_startblk, i_endblk, i_startidx_in, i_endidx_in, slev, elev, nproma,
+ nlev, nblks_c, nblks_v, lacc, acc_async);
}
-// This is the binding for mo_interpolation_scalar::cells2verts_scalar_ri_lib (wp=sp, vp=sp)
+// This is the binding for mo_interpolation_scalar::cells2verts_scalar_ri_lib
+// (wp=sp, vp=sp)
void cells2verts_scalar_ri_lib_sp(
- const float* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async)
-{
- cells2verts_scalar_ri_lib<float,float>(
- p_cell_in, vert_cell_idx, vert_cell_blk,
- coeff_int, p_vert_out, i_startblk, i_endblk,
- i_startidx_in, i_endidx_in,
- slev, elev, nproma,
- nlev, nblks_c, nblks_v,
- lacc, acc_async);
+ const float *p_cell_in, const int *vert_cell_idx, const int *vert_cell_blk,
+ const float *coeff_int, float *p_vert_out, const int i_startblk,
+ const int i_endblk, const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblks_c, const int nblks_v, const bool lacc,
+ const bool acc_async) {
+ cells2verts_scalar_ri_lib<float, float>(
+ p_cell_in, vert_cell_idx, vert_cell_blk, coeff_int, p_vert_out,
+ i_startblk, i_endblk, i_startidx_in, i_endidx_in, slev, elev, nproma,
+ nlev, nblks_c, nblks_v, lacc, acc_async);
}
-// This is the binding for mo_interpolation_scalar::verts2cells_scalar_lib (wp=dp)
+// This is the binding for mo_interpolation_scalar::verts2cells_scalar_lib
+// (wp=dp)
void verts2cells_scalar_lib_dp(
- const double* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
- const double* coeff_int, double* p_cell_out, const int nblks_c, const int npromz_c,
- const int slev, const int elev, const int nproma, const int nlev,
- const int nblks_v, const bool lacc)
-{
- verts2cells_scalar_lib<double>(
- p_vert_in, cell_index_idx, cell_vertex_blk,
- coeff_int, p_cell_out, nblks_c, npromz_c,
- slev, elev, nproma, nlev,
- nblks_v, lacc);
+ const double *p_vert_in, const int *cell_index_idx,
+ const int *cell_vertex_blk, const double *coeff_int, double *p_cell_out,
+ const int nblks_c, const int npromz_c, const int slev, const int elev,
+ const int nproma, const int nlev, const int nblks_v, const bool lacc) {
+ verts2cells_scalar_lib<double>(p_vert_in, cell_index_idx, cell_vertex_blk,
+ coeff_int, p_cell_out, nblks_c, npromz_c, slev,
+ elev, nproma, nlev, nblks_v, lacc);
}
-// This is the binding for mo_interpolation_scalar::verts2cells_scalar_lib (wp=sp)
+// This is the binding for mo_interpolation_scalar::verts2cells_scalar_lib
+// (wp=sp)
void verts2cells_scalar_lib_sp(
- const float* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
- const float* coeff_int, float* p_cell_out, const int nblks_c, const int npromz_c,
- const int slev, const int elev, const int nproma, const int nlev,
- const int nblks_v, const bool lacc)
-{
- verts2cells_scalar_lib<float>(
- p_vert_in, cell_index_idx, cell_vertex_blk,
- coeff_int, p_cell_out, nblks_c, npromz_c,
- slev, elev, nproma, nlev,
- nblks_v, lacc);
+ const float *p_vert_in, const int *cell_index_idx,
+ const int *cell_vertex_blk, const float *coeff_int, float *p_cell_out,
+ const int nblks_c, const int npromz_c, const int slev, const int elev,
+ const int nproma, const int nlev, const int nblks_v, const bool lacc) {
+ verts2cells_scalar_lib<float>(p_vert_in, cell_index_idx, cell_vertex_blk,
+ coeff_int, p_cell_out, nblks_c, npromz_c, slev,
+ elev, nproma, nlev, nblks_v, lacc);
}
// This is the binding for mo_interpolation_scalar::cell_avg_lib (wp=dp)
-void cell_avg_lib_dp(
- const double* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
- const double* avg_coeff, double* avg_psi_c, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c,
- const bool lacc)
-{
- cell_avg_lib<double>(
- psi_c, cell_neighbor_idx, cell_neighbor_blk,
- avg_coeff, avg_psi_c, i_startblk, i_endblk,
- i_startidx_in, i_endidx_in,
- slev, elev, nproma,
- nlev, nblks_c,
- lacc);
+void cell_avg_lib_dp(const double *psi_c, const int *cell_neighbor_idx,
+ const int *cell_neighbor_blk, const double *avg_coeff,
+ double *avg_psi_c, const int i_startblk,
+ const int i_endblk, const int i_startidx_in,
+ const int i_endidx_in, const int slev, const int elev,
+ const int nproma, const int nlev, const int nblks_c,
+ const bool lacc) {
+ cell_avg_lib<double>(psi_c, cell_neighbor_idx, cell_neighbor_blk, avg_coeff,
+ avg_psi_c, i_startblk, i_endblk, i_startidx_in,
+ i_endidx_in, slev, elev, nproma, nlev, nblks_c, lacc);
}
// This is the binding for mo_interpolation_scalar::cell_avg_lib (wp=sp)
-void cell_avg_lib_sp(
- const float* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
- const float* avg_coeff, float* avg_psi_c, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c,
- const bool lacc)
-{
- cell_avg_lib<float>(
- psi_c, cell_neighbor_idx, cell_neighbor_blk,
- avg_coeff, avg_psi_c, i_startblk, i_endblk,
- i_startidx_in, i_endidx_in,
- slev, elev, nproma,
- nlev, nblks_c,
- lacc);
+void cell_avg_lib_sp(const float *psi_c, const int *cell_neighbor_idx,
+ const int *cell_neighbor_blk, const float *avg_coeff,
+ float *avg_psi_c, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const bool lacc) {
+ cell_avg_lib<float>(psi_c, cell_neighbor_idx, cell_neighbor_blk, avg_coeff,
+ avg_psi_c, i_startblk, i_endblk, i_startidx_in,
+ i_endidx_in, slev, elev, nproma, nlev, nblks_c, lacc);
}
-
diff --git a/src/interpolation/interpolation_bindings.h b/src/interpolation/interpolation_bindings.h
index 7c6b125..7cb873d 100644
--- a/src/interpolation/interpolation_bindings.h
+++ b/src/interpolation/interpolation_bindings.h
@@ -10,177 +10,179 @@
// ---------------------------------------------------------------
#pragma once
-extern "C"{
-
- //mo_lib_interpolation_vector.F90
- void edges2cells_vector_lib_dp(
- const double* p_vn_in, const double* p_vt_in,
- const int* cell_edge_idx, const int* cell_edge_blk,
- const double* e_bln_c_u, const double* e_bln_c_v,
- double* p_u_out, double* p_v_out,
- int i_startblk, int i_endblk,
- int i_startidx_in, int i_endidx_in,
- int slev, int elev,
- int nproma,
- int nlev, int nblks_e, int nblks_c);
-
- void edges2cells_vector_lib_sp(
- const float* p_vn_in, const float* p_vt_in,
- const int* cell_edge_idx, const int* cell_edge_blk,
- const float* e_bln_c_u, const float* e_bln_c_v,
- float* p_u_out, float* p_v_out,
- int i_startblk, int i_endblk,
- int i_startidx_in, int i_endidx_in,
- int slev, int elev,
- int nproma,
- int nlev, int nblks_e, int nblks_c);
-
- //mo_lib_interpolation_scalar.F90
- void verts2edges_scalar_lib_dp(const double* p_vertex_in,
- const int* edge_vertex_idx, const int* edge_vertex_blk,
- const double* coeff_int,
- double* p_edge_out,
- const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
- const int nproma, const int nlev, const int nblks_v, const int nblks_e, const bool lacc);
-
- void verts2edges_scalar_lib_sp(const float* p_vertex_in,
- const int* edge_vertex_idx, const int* edge_vertex_blk,
- const float* coeff_int,
- float* p_edge_out,
- const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
- const int nproma, const int nlev, const int nblks_v, const int nblks_e, const bool lacc);
-
- void cells2edges_scalar_lib_dp(const double* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
- const double* coeff_int, double* p_edge_out, const int* i_startblk_in,
- const int* i_endblk_in, const int* i_startidx_in, const int* i_endidx_in,
- const int slev, const int elev, const int nproma, const int nlev,
- const int nblk_c, const int nblks_e, const int patch_id,
- const bool l_limited_area, const bool lfill_latbc, const bool lacc);
-
-
- void cells2edges_scalar_lib_sp(const float* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
- const float* coeff_int, float* p_edge_out, const int* i_startblk_in,
- const int* i_endblk_in, const int* i_startidx_in, const int* i_endidx_in,
- const int slev, const int elev, const int nproma, const int nlev,
- const int nblk_c, const int nblks_e, const int patch_id,
- const bool l_limited_area, const bool lfill_latbc, const bool lacc);
-
- void cells2edges_scalar_lib_sp2dp(const float* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
- const double* coeff_int, double* p_edge_out, const int* i_startblk_in,
- const int* i_endblk_in, const int* i_startidx_in, const int* i_endidx_in,
- const int slev, const int elev, const int nproma, const int nlev,
- const int nblk_c, const int nblks_e, const int patch_id,
- const bool l_limited_area, const bool lfill_latbc, const bool lacc);
-
- void edges2verts_scalar_lib_dp (const double* p_edge_in, const int* vert_edge_idx, const int* vert_edge_blk,
- const double* v_int, double* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in, const int slev,
- const int elev, const int nproma, const int nlev, const int nblks_e,
- const int nblks_v, const bool lacc);
-
-
- void edges2verts_scalar_lib_sp (const float* p_edge_in, const int* vert_edge_idx, const int* vert_edge_blk,
- const float* v_int, float* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in, const int slev,
- const int elev, const int nproma, const int nlev, const int nblks_e,
- const int nblks_v, const bool lacc);
-
- void edges2cells_scalar_lib_dp(
- const double* p_edge_in, const int* edge_idx, const int* edge_blk,
- const double* coeff_int, double* p_cell_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_e, const int nblks_c,
- const bool lacc);
- void edges2cells_scalar_lib_sp(
- const float* p_edge_in, const int* edge_idx, const int* edge_blk,
- const float* coeff_int, float* p_cell_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_e, const int nblks_c,
- const bool lacc);
-
- /////////////////////////////////////////////
-
- void cells2verts_scalar_lib_dp(
- const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const double* coeff_int, double* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async);
- void cells2verts_scalar_lib_dp2sp(
- const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async);
- void cells2verts_scalar_lib_sp(
- const float* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async);
-
- /////////////////////////////////////////////
-
- void cells2verts_scalar_ri_lib_dp(
- const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const double* coeff_int, double* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async);
-
- void cells2verts_scalar_ri_lib_dp2sp(
- const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const double* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async);
-
- void cells2verts_scalar_ri_lib_sp(
- const float* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async);
-
- /////////////////////////////////////////////
-
- void verts2cells_scalar_lib_dp(
- const double* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
- const double* coeff_int, double* p_cell_out, const int nblks_c, const int npromz_c,
- const int slev, const int elev, const int nproma, const int nlev,
- const int nblks_v, const bool lacc);
-
- void verts2cells_scalar_lib_sp(
- const float* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
- const float* coeff_int, float* p_cell_out, const int nblks_c, const int npromz_c,
- const int slev, const int elev, const int nproma, const int nlev,
- const int nblks_v, const bool lacc);
-
- /////////////////////////////////////////////
-
- void cell_avg_lib_dp(
- const double* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
- const double* avg_coeff, double* avg_psi_c, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c,
- const bool lacc);
- void cell_avg_lib_sp(
- const float* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
- const float* avg_coeff, float* avg_psi_c, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c,
- const bool lacc);
-
+extern "C" {
+
+// mo_lib_interpolation_vector.F90
+void edges2cells_vector_lib_dp(const double *p_vn_in, const double *p_vt_in,
+ const int *cell_edge_idx,
+ const int *cell_edge_blk,
+ const double *e_bln_c_u, const double *e_bln_c_v,
+ double *p_u_out, double *p_v_out, int i_startblk,
+ int i_endblk, int i_startidx_in, int i_endidx_in,
+ int slev, int elev, int nproma, int nlev,
+ int nblks_e, int nblks_c);
+
+void edges2cells_vector_lib_sp(const float *p_vn_in, const float *p_vt_in,
+ const int *cell_edge_idx,
+ const int *cell_edge_blk, const float *e_bln_c_u,
+ const float *e_bln_c_v, float *p_u_out,
+ float *p_v_out, int i_startblk, int i_endblk,
+ int i_startidx_in, int i_endidx_in, int slev,
+ int elev, int nproma, int nlev, int nblks_e,
+ int nblks_c);
+
+// mo_lib_interpolation_scalar.F90
+void verts2edges_scalar_lib_dp(
+ const double *p_vertex_in, const int *edge_vertex_idx,
+ const int *edge_vertex_blk, const double *coeff_int, double *p_edge_out,
+ const int i_startblk, const int i_endblk, const int i_startidx_in,
+ const int i_endidx_in, const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_v, const int nblks_e, const bool lacc);
+
+void verts2edges_scalar_lib_sp(
+ const float *p_vertex_in, const int *edge_vertex_idx,
+ const int *edge_vertex_blk, const float *coeff_int, float *p_edge_out,
+ const int i_startblk, const int i_endblk, const int i_startidx_in,
+ const int i_endidx_in, const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_v, const int nblks_e, const bool lacc);
+
+void cells2edges_scalar_lib_dp(
+ const double *p_cell_in, const int *edge_cell_idx, const int *edge_cell_blk,
+ const double *coeff_int, double *p_edge_out, const int *i_startblk_in,
+ const int *i_endblk_in, const int *i_startidx_in, const int *i_endidx_in,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblk_c, const int nblks_e, const int patch_id,
+ const bool l_limited_area, const bool lfill_latbc, const bool lacc);
+
+void cells2edges_scalar_lib_sp(const float *p_cell_in, const int *edge_cell_idx,
+ const int *edge_cell_blk, const float *coeff_int,
+ float *p_edge_out, const int *i_startblk_in,
+ const int *i_endblk_in, const int *i_startidx_in,
+ const int *i_endidx_in, const int slev,
+ const int elev, const int nproma, const int nlev,
+ const int nblk_c, const int nblks_e,
+ const int patch_id, const bool l_limited_area,
+ const bool lfill_latbc, const bool lacc);
+
+void cells2edges_scalar_lib_sp2dp(
+ const float *p_cell_in, const int *edge_cell_idx, const int *edge_cell_blk,
+ const double *coeff_int, double *p_edge_out, const int *i_startblk_in,
+ const int *i_endblk_in, const int *i_startidx_in, const int *i_endidx_in,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblk_c, const int nblks_e, const int patch_id,
+ const bool l_limited_area, const bool lfill_latbc, const bool lacc);
+
+void edges2verts_scalar_lib_dp(
+ const double *p_edge_in, const int *vert_edge_idx, const int *vert_edge_blk,
+ const double *v_int, double *p_vert_out, const int i_startblk,
+ const int i_endblk, const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblks_e, const int nblks_v, const bool lacc);
+
+void edges2verts_scalar_lib_sp(const float *p_edge_in, const int *vert_edge_idx,
+ const int *vert_edge_blk, const float *v_int,
+ float *p_vert_out, const int i_startblk,
+ const int i_endblk, const int i_startidx_in,
+ const int i_endidx_in, const int slev,
+ const int elev, const int nproma, const int nlev,
+ const int nblks_e, const int nblks_v,
+ const bool lacc);
+
+void edges2cells_scalar_lib_dp(const double *p_edge_in, const int *edge_idx,
+ const int *edge_blk, const double *coeff_int,
+ double *p_cell_out, const int i_startblk,
+ const int i_endblk, const int i_startidx_in,
+ const int i_endidx_in, const int slev,
+ const int elev, const int nproma, const int nlev,
+ const int nblks_e, const int nblks_c,
+ const bool lacc);
+void edges2cells_scalar_lib_sp(const float *p_edge_in, const int *edge_idx,
+ const int *edge_blk, const float *coeff_int,
+ float *p_cell_out, const int i_startblk,
+ const int i_endblk, const int i_startidx_in,
+ const int i_endidx_in, const int slev,
+ const int elev, const int nproma, const int nlev,
+ const int nblks_e, const int nblks_c,
+ const bool lacc);
+
+/////////////////////////////////////////////
+
+void cells2verts_scalar_lib_dp(
+ const double *p_cell_in, const int *vert_cell_idx, const int *vert_cell_blk,
+ const double *coeff_int, double *p_vert_out, const int i_startblk,
+ const int i_endblk, const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblks_c, const int nblks_v, const bool lacc,
+ const bool acc_async);
+void cells2verts_scalar_lib_dp2sp(
+ const double *p_cell_in, const int *vert_cell_idx, const int *vert_cell_blk,
+ const float *coeff_int, float *p_vert_out, const int i_startblk,
+ const int i_endblk, const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblks_c, const int nblks_v, const bool lacc,
+ const bool acc_async);
+void cells2verts_scalar_lib_sp(const float *p_cell_in, const int *vert_cell_idx,
+ const int *vert_cell_blk, const float *coeff_int,
+ float *p_vert_out, const int i_startblk,
+ const int i_endblk, const int i_startidx_in,
+ const int i_endidx_in, const int slev,
+ const int elev, const int nproma, const int nlev,
+ const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async);
+
+/////////////////////////////////////////////
+
+void cells2verts_scalar_ri_lib_dp(
+ const double *p_cell_in, const int *vert_cell_idx, const int *vert_cell_blk,
+ const double *coeff_int, double *p_vert_out, const int i_startblk,
+ const int i_endblk, const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblks_c, const int nblks_v, const bool lacc,
+ const bool acc_async);
+
+void cells2verts_scalar_ri_lib_dp2sp(
+ const double *p_cell_in, const int *vert_cell_idx, const int *vert_cell_blk,
+ const double *coeff_int, float *p_vert_out, const int i_startblk,
+ const int i_endblk, const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblks_c, const int nblks_v, const bool lacc,
+ const bool acc_async);
+
+void cells2verts_scalar_ri_lib_sp(
+ const float *p_cell_in, const int *vert_cell_idx, const int *vert_cell_blk,
+ const float *coeff_int, float *p_vert_out, const int i_startblk,
+ const int i_endblk, const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblks_c, const int nblks_v, const bool lacc,
+ const bool acc_async);
+
+/////////////////////////////////////////////
+
+void verts2cells_scalar_lib_dp(
+ const double *p_vert_in, const int *cell_index_idx,
+ const int *cell_vertex_blk, const double *coeff_int, double *p_cell_out,
+ const int nblks_c, const int npromz_c, const int slev, const int elev,
+ const int nproma, const int nlev, const int nblks_v, const bool lacc);
+
+void verts2cells_scalar_lib_sp(
+ const float *p_vert_in, const int *cell_index_idx,
+ const int *cell_vertex_blk, const float *coeff_int, float *p_cell_out,
+ const int nblks_c, const int npromz_c, const int slev, const int elev,
+ const int nproma, const int nlev, const int nblks_v, const bool lacc);
+
+/////////////////////////////////////////////
+
+void cell_avg_lib_dp(const double *psi_c, const int *cell_neighbor_idx,
+ const int *cell_neighbor_blk, const double *avg_coeff,
+ double *avg_psi_c, const int i_startblk,
+ const int i_endblk, const int i_startidx_in,
+ const int i_endidx_in, const int slev, const int elev,
+ const int nproma, const int nlev, const int nblks_c,
+ const bool lacc);
+void cell_avg_lib_sp(const float *psi_c, const int *cell_neighbor_idx,
+ const int *cell_neighbor_blk, const float *avg_coeff,
+ float *avg_psi_c, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const bool lacc);
}
diff --git a/src/interpolation/mo_lib_interpolation_scalar.cpp b/src/interpolation/mo_lib_interpolation_scalar.cpp
index 386ce0c..9e4e6c5 100644
--- a/src/interpolation/mo_lib_interpolation_scalar.cpp
+++ b/src/interpolation/mo_lib_interpolation_scalar.cpp
@@ -9,10 +9,10 @@
// SPDX-License-Identifier: BSD-3-Clause
// ---------------------------------------------------------------
+#include "mo_lib_interpolation_scalar.hpp"
+#include "mo_lib_loopindices.hpp"
#include <Kokkos_Core.hpp>
#include <iostream>
-#include "mo_lib_loopindices.hpp"
-#include "mo_lib_interpolation_scalar.hpp"
//-----------------------------------------------------------------------
//
@@ -28,45 +28,52 @@
/// The coefficients are given by coeff_int.
///
template <typename T>
-void verts2edges_scalar_lib(const T* p_vertex_in,
- const int* edge_vertex_idx, const int* edge_vertex_blk,
- const T* coeff_int,
- T* p_edge_out,
- const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
- const int nproma, const int nlev, const int nblks_v, const int nblks_e, const bool lacc){
-
- // Wrap raw pointers in unmanaged Kokkos Views.
- typedef Kokkos::View<const T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstT3D;
- typedef Kokkos::View<T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedT3D;
- typedef Kokkos::View<const int***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstInt3D;
-
- UnmanagedConstT3D p_vertex_in_view (p_vertex_in, nproma,nlev,nblks_v);
- UnmanagedConstInt3D iidx_view (edge_vertex_idx,nproma,nblks_e, 4);
- UnmanagedConstInt3D iblk_view (edge_vertex_blk, nproma,nblks_e, 4);
- UnmanagedConstT3D coeff_int_view (coeff_int, nproma,2,nblks_e);
- UnmanagedT3D p_edge_out_view (p_edge_out, nproma,nlev,nblks_e);
-
- for (int jb = i_startblk; jb < i_endblk + 1; ++jb) {
-
- int i_startidx, i_endidx;
- get_indices_c_lib(i_startidx_in, i_endidx_in, nproma,
- jb, i_startblk, i_endblk,
- i_startidx, i_endidx);
+void verts2edges_scalar_lib(const T *p_vertex_in, const int *edge_vertex_idx,
+ const int *edge_vertex_blk, const T *coeff_int,
+ T *p_edge_out, const int i_startblk,
+ const int i_endblk, const int i_startidx_in,
+ const int i_endidx_in, const int slev,
+ const int elev, const int nproma, const int nlev,
+ const int nblks_v, const int nblks_e,
+ const bool lacc) {
+
+ // Wrap raw pointers in unmanaged Kokkos Views.
+ typedef Kokkos::View<const T ***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged>
+ UnmanagedConstT3D;
+ typedef Kokkos::View<T ***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged>
+ UnmanagedT3D;
+ typedef Kokkos::View<const int ***, Kokkos::LayoutLeft,
+ Kokkos::MemoryUnmanaged>
+ UnmanagedConstInt3D;
+
+ UnmanagedConstT3D p_vertex_in_view(p_vertex_in, nproma, nlev, nblks_v);
+ UnmanagedConstInt3D iidx_view(edge_vertex_idx, nproma, nblks_e, 4);
+ UnmanagedConstInt3D iblk_view(edge_vertex_blk, nproma, nblks_e, 4);
+ UnmanagedConstT3D coeff_int_view(coeff_int, nproma, 2, nblks_e);
+ UnmanagedT3D p_edge_out_view(p_edge_out, nproma, nlev, nblks_e);
+
+ for (int jb = i_startblk; jb < i_endblk + 1; ++jb) {
- Kokkos::MDRangePolicy<Kokkos::Rank<2>> innerPolicy(
- {slev, i_startidx}, {elev + 1, i_endidx + 1});
-
- Kokkos::parallel_for("verts2edges_scalar", innerPolicy,
- KOKKOS_LAMBDA(const int jk, const int je) {
-
- p_edge_out_view(je, jk, jb) = coeff_int_view(je, 0, jb) *
- p_vertex_in_view(iidx_view(je, jb, 0), jk, iblk_view(je, jb, 0)) +
- coeff_int_view(je, 1, jb)*p_vertex_in_view(iidx_view(je, jb, 1), jk, iblk_view(je, jb, 1));
- }
- );
- Kokkos::fence();
- }
+ int i_startidx, i_endidx;
+ get_indices_c_lib(i_startidx_in, i_endidx_in, nproma, jb, i_startblk,
+ i_endblk, i_startidx, i_endidx);
+
+ Kokkos::MDRangePolicy<Kokkos::Rank<2>> innerPolicy(
+ {slev, i_startidx}, {elev + 1, i_endidx + 1});
+
+ Kokkos::parallel_for(
+ "verts2edges_scalar", innerPolicy,
+ KOKKOS_LAMBDA(const int jk, const int je) {
+ p_edge_out_view(je, jk, jb) =
+ coeff_int_view(je, 0, jb) *
+ p_vertex_in_view(iidx_view(je, jb, 0), jk,
+ iblk_view(je, jb, 0)) +
+ coeff_int_view(je, 1, jb) *
+ p_vertex_in_view(iidx_view(je, jb, 1), jk,
+ iblk_view(je, jb, 1));
+ });
+ Kokkos::fence();
+ }
}
//------------------------------------------------------------------------
@@ -77,83 +84,95 @@ void verts2edges_scalar_lib(const T* p_vertex_in,
/// velocity points.
///
template <typename T, typename S>
-void cells2edges_scalar_lib(const T* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
- const S* coeff_int, S* p_edge_out, const int* i_startblk_in,
- const int* i_endblk_in, const int* i_startidx_in, const int* i_endidx_in,
- const int slev, const int elev, const int nproma, const int nlev,
- const int nblk_c, const int nblks_e, const int patch_id,
- const bool l_limited_area, const bool lfill_latbc, const bool lacc){
-
- // Wrap raw pointers in unmanaged Kokkos Views.
- typedef Kokkos::View<const T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstT3D;
- typedef Kokkos::View<const S***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstS3D;
- typedef Kokkos::View<S***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedS3D;
- typedef Kokkos::View<const int***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstInt3D;
-
- UnmanagedConstT3D p_cell_in_view (p_cell_in, nproma, nlev, nblk_c);
- UnmanagedConstInt3D iidx_view (edge_cell_idx, nproma, nblks_e, 2);
- UnmanagedConstInt3D iblk_view (edge_cell_blk, nproma, nblks_e, 2);
- UnmanagedConstS3D coeff_int_view (coeff_int, nproma, 2, nblks_e);
- UnmanagedS3D p_edge_out_view (p_edge_out, nproma, nlev, nblks_e);
-
- //Fill outermost nest boundary
- int i_startblk, i_endblk;
- if ((l_limited_area || patch_id > 0) && (lfill_latbc)){
- i_startblk = i_startblk_in[0];
- i_endblk = i_endblk_in[0];
-
- for (int jb = i_startblk; jb < i_endblk + 1; ++jb) {
-
- int i_startidx, i_endidx;
- get_indices_e_lib(i_startidx_in[0], i_endidx_in[0], nproma, jb,
- i_startblk, i_endblk, i_startidx, i_endidx);
-
- Kokkos::MDRangePolicy<Kokkos::Rank<2>> innerPolicy(
- {slev, i_startidx}, {elev + 1, i_endidx + 1});
-
- Kokkos::parallel_for("cells2edges_scalar", innerPolicy,
- KOKKOS_LAMBDA(const int jk, const int je) {
-
- if (iidx_view(je, jb, 0) >= 0 && iblk_view(je, jb, 0) >= 0){
- p_edge_out_view(je, jk, jb) = p_cell_in_view(iidx_view(je, jb, 0), jk, iblk_view(je, jb, 0));
- }
- else if (iidx_view(je, jb, 1) >= 0 && iblk_view(je, jb, 1) >= 0){
- p_edge_out_view(je, jk, jb) = p_cell_in_view(iidx_view(je, jb, 1), jk, iblk_view(je, jb, 1));
- }
- else{
- std::cerr << "mo_interpolation:cells2edges_scalar_lib: error in lateral boundary filling" << std::endl;
- std::exit(EXIT_FAILURE);
- }
- }
- );
- Kokkos::fence();
- }
- }
- else{
- //Process the remaining grid points for which a real interpolation is possible
- i_startblk = i_startblk_in[1];
- i_endblk = i_endblk_in[1];
-
- for (int jb = i_startblk; jb < i_endblk + 1; ++jb) {
-
- int i_startidx, i_endidx;
- get_indices_e_lib(i_startidx_in[1], i_endidx_in[1], nproma, jb,
- i_startblk, i_endblk, i_startidx, i_endidx);
-
- Kokkos::MDRangePolicy<Kokkos::Rank<2>> innerPolicy(
- {slev, i_startidx}, {elev + 1, i_endidx + 1});
-
- Kokkos::parallel_for("cells2edges_scalar", innerPolicy,
- KOKKOS_LAMBDA(const int jk, const int je) {
- p_edge_out_view(je, jk, jb) = coeff_int_view(je, 0, jb) *
- p_cell_in_view(iidx_view(je, jb, 0), jk, iblk_view(je, jb, 0)) +
- coeff_int_view(je, 1, jb) * p_cell_in_view(iidx_view(je, jb, 1), jk, iblk_view(je, jb, 1));
- }
- );
- Kokkos::fence();
- }
+void cells2edges_scalar_lib(const T *p_cell_in, const int *edge_cell_idx,
+ const int *edge_cell_blk, const S *coeff_int,
+ S *p_edge_out, const int *i_startblk_in,
+ const int *i_endblk_in, const int *i_startidx_in,
+ const int *i_endidx_in, const int slev,
+ const int elev, const int nproma, const int nlev,
+ const int nblk_c, const int nblks_e,
+ const int patch_id, const bool l_limited_area,
+ const bool lfill_latbc, const bool lacc) {
+
+ // Wrap raw pointers in unmanaged Kokkos Views.
+ typedef Kokkos::View<const T ***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged>
+ UnmanagedConstT3D;
+ typedef Kokkos::View<const S ***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged>
+ UnmanagedConstS3D;
+ typedef Kokkos::View<S ***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged>
+ UnmanagedS3D;
+ typedef Kokkos::View<const int ***, Kokkos::LayoutLeft,
+ Kokkos::MemoryUnmanaged>
+ UnmanagedConstInt3D;
+
+ UnmanagedConstT3D p_cell_in_view(p_cell_in, nproma, nlev, nblk_c);
+ UnmanagedConstInt3D iidx_view(edge_cell_idx, nproma, nblks_e, 2);
+ UnmanagedConstInt3D iblk_view(edge_cell_blk, nproma, nblks_e, 2);
+ UnmanagedConstS3D coeff_int_view(coeff_int, nproma, 2, nblks_e);
+ UnmanagedS3D p_edge_out_view(p_edge_out, nproma, nlev, nblks_e);
+
+ // Fill outermost nest boundary
+ int i_startblk, i_endblk;
+ if ((l_limited_area || patch_id > 0) && (lfill_latbc)) {
+ i_startblk = i_startblk_in[0];
+ i_endblk = i_endblk_in[0];
+
+ for (int jb = i_startblk; jb < i_endblk + 1; ++jb) {
+
+ int i_startidx, i_endidx;
+ get_indices_e_lib(i_startidx_in[0], i_endidx_in[0], nproma, jb,
+ i_startblk, i_endblk, i_startidx, i_endidx);
+
+ Kokkos::MDRangePolicy<Kokkos::Rank<2>> innerPolicy(
+ {slev, i_startidx}, {elev + 1, i_endidx + 1});
+
+ Kokkos::parallel_for(
+ "cells2edges_scalar", innerPolicy,
+ KOKKOS_LAMBDA(const int jk, const int je) {
+ if (iidx_view(je, jb, 0) >= 0 && iblk_view(je, jb, 0) >= 0) {
+ p_edge_out_view(je, jk, jb) = p_cell_in_view(
+ iidx_view(je, jb, 0), jk, iblk_view(je, jb, 0));
+ } else if (iidx_view(je, jb, 1) >= 0 && iblk_view(je, jb, 1) >= 0) {
+ p_edge_out_view(je, jk, jb) = p_cell_in_view(
+ iidx_view(je, jb, 1), jk, iblk_view(je, jb, 1));
+ } else {
+ std::cerr << "mo_interpolation:cells2edges_scalar_lib: error in "
+ "lateral boundary filling"
+ << std::endl;
+ std::exit(EXIT_FAILURE);
+ }
+ });
+ Kokkos::fence();
}
+ } else {
+ // Process the remaining grid points for which a real interpolation is
+ // possible
+ i_startblk = i_startblk_in[1];
+ i_endblk = i_endblk_in[1];
+
+ for (int jb = i_startblk; jb < i_endblk + 1; ++jb) {
+
+ int i_startidx, i_endidx;
+ get_indices_e_lib(i_startidx_in[1], i_endidx_in[1], nproma, jb,
+ i_startblk, i_endblk, i_startidx, i_endidx);
+
+ Kokkos::MDRangePolicy<Kokkos::Rank<2>> innerPolicy(
+ {slev, i_startidx}, {elev + 1, i_endidx + 1});
+ Kokkos::parallel_for(
+ "cells2edges_scalar", innerPolicy,
+ KOKKOS_LAMBDA(const int jk, const int je) {
+ p_edge_out_view(je, jk, jb) =
+ coeff_int_view(je, 0, jb) *
+ p_cell_in_view(iidx_view(je, jb, 0), jk,
+ iblk_view(je, jb, 0)) +
+ coeff_int_view(je, 1, jb) *
+ p_cell_in_view(iidx_view(je, jb, 1), jk,
+ iblk_view(je, jb, 1));
+ });
+ Kokkos::fence();
+ }
+ }
}
//------------------------------------------------------------------------
@@ -164,46 +183,58 @@ void cells2edges_scalar_lib(const T* p_cell_in, const int* edge_cell_idx, const
/// centers of dual faces.
///
template <typename T>
-void edges2verts_scalar_lib (const T* p_edge_in, const int* vert_edge_idx, const int* vert_edge_blk,
- const T* v_int, T* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in, const int slev,
- const int elev, const int nproma, const int nlev, const int nblks_e,
- const int nblks_v, const bool lacc){
-
- // Wrap raw pointers in unmanaged Kokkos Views.
- typedef Kokkos::View<const T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstT3D;
- typedef Kokkos::View<T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedT3D;
- typedef Kokkos::View<const int***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstInt3D;
-
- UnmanagedConstT3D p_edge_in_view (p_edge_in, nproma, nlev, nblks_e);
- UnmanagedConstInt3D iidx_view (vert_edge_idx, nproma, nblks_v, 6);
- UnmanagedConstInt3D iblk_view (vert_edge_blk, nproma,nblks_v, 6);
- UnmanagedConstT3D v_int_view (v_int, nproma, 6, nblks_v);
- UnmanagedT3D p_vert_out_view (p_vert_out, nproma, nlev, nblks_v);
+void edges2verts_scalar_lib(const T *p_edge_in, const int *vert_edge_idx,
+ const int *vert_edge_blk, const T *v_int,
+ T *p_vert_out, const int i_startblk,
+ const int i_endblk, const int i_startidx_in,
+ const int i_endidx_in, const int slev,
+ const int elev, const int nproma, const int nlev,
+ const int nblks_e, const int nblks_v,
+ const bool lacc) {
+
+ // Wrap raw pointers in unmanaged Kokkos Views.
+ typedef Kokkos::View<const T ***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged>
+ UnmanagedConstT3D;
+ typedef Kokkos::View<T ***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged>
+ UnmanagedT3D;
+ typedef Kokkos::View<const int ***, Kokkos::LayoutLeft,
+ Kokkos::MemoryUnmanaged>
+ UnmanagedConstInt3D;
+
+ UnmanagedConstT3D p_edge_in_view(p_edge_in, nproma, nlev, nblks_e);
+ UnmanagedConstInt3D iidx_view(vert_edge_idx, nproma, nblks_v, 6);
+ UnmanagedConstInt3D iblk_view(vert_edge_blk, nproma, nblks_v, 6);
+ UnmanagedConstT3D v_int_view(v_int, nproma, 6, nblks_v);
+ UnmanagedT3D p_vert_out_view(p_vert_out, nproma, nlev, nblks_v);
+
+ for (int jb = i_startblk; jb < i_endblk + 1; ++jb) {
- for (int jb = i_startblk; jb < i_endblk + 1; ++jb) {
+ int i_startidx, i_endidx;
+ get_indices_v_lib(i_startidx_in, i_endidx_in, nproma, jb, i_startblk,
+ i_endblk, i_startidx, i_endidx);
- int i_startidx, i_endidx;
- get_indices_v_lib(i_startidx_in, i_endidx_in, nproma,
- jb, i_startblk, i_endblk,
- i_startidx, i_endidx);
-
- Kokkos::MDRangePolicy<Kokkos::Rank<2>> innerPolicy(
- {slev, i_startidx}, {elev + 1, i_endidx + 1});
-
- Kokkos::parallel_for("edges2verts_scalar", innerPolicy,
- KOKKOS_LAMBDA(const int jk, const int jv) {
- p_vert_out_view(jv, jk, jb) = v_int_view(jv, 0, jb)*p_edge_in_view(iidx_view(jv, jb, 0), jk, iblk_view(jv, jb, 0))
- + v_int_view(jv, 1, jb)*p_edge_in_view(iidx_view(jv, jb, 1), jk, iblk_view(jv, jb, 1))
- + v_int_view(jv, 2, jb)*p_edge_in_view(iidx_view(jv, jb, 2), jk, iblk_view(jv, jb, 2))
- + v_int_view(jv, 3, jb)*p_edge_in_view(iidx_view(jv, jb, 3), jk, iblk_view(jv, jb, 3))
- + v_int_view(jv, 4, jb)*p_edge_in_view(iidx_view(jv, jb, 4), jk, iblk_view(jv, jb, 4))
- + v_int_view(jv, 5, jb)*p_edge_in_view(iidx_view(jv, jb, 5), jk, iblk_view(jv, jb, 5));
- }
- );
- Kokkos::fence();
- }
+ Kokkos::MDRangePolicy<Kokkos::Rank<2>> innerPolicy(
+ {slev, i_startidx}, {elev + 1, i_endidx + 1});
+ Kokkos::parallel_for(
+ "edges2verts_scalar", innerPolicy,
+ KOKKOS_LAMBDA(const int jk, const int jv) {
+ p_vert_out_view(jv, jk, jb) =
+ v_int_view(jv, 0, jb) * p_edge_in_view(iidx_view(jv, jb, 0), jk,
+ iblk_view(jv, jb, 0)) +
+ v_int_view(jv, 1, jb) * p_edge_in_view(iidx_view(jv, jb, 1), jk,
+ iblk_view(jv, jb, 1)) +
+ v_int_view(jv, 2, jb) * p_edge_in_view(iidx_view(jv, jb, 2), jk,
+ iblk_view(jv, jb, 2)) +
+ v_int_view(jv, 3, jb) * p_edge_in_view(iidx_view(jv, jb, 3), jk,
+ iblk_view(jv, jb, 3)) +
+ v_int_view(jv, 4, jb) * p_edge_in_view(iidx_view(jv, jb, 4), jk,
+ iblk_view(jv, jb, 4)) +
+ v_int_view(jv, 5, jb) * p_edge_in_view(iidx_view(jv, jb, 5), jk,
+ iblk_view(jv, jb, 5));
+ });
+ Kokkos::fence();
+ }
}
//------------------------------------------------------------------------
@@ -214,54 +245,64 @@ void edges2verts_scalar_lib (const T* p_edge_in, const int* vert_edge_idx, const
/// cell centers via given interpolation weights
///
template <typename T>
-void edges2cells_scalar_lib(
- const T* p_edge_in, const int* edge_idx, const int* edge_blk,
- const T* coeff_int, T* p_cell_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_e, const int nblks_c,
- const bool lacc)
-{
- // Wrap raw pointers in unmanaged Kokkos Views.
- typedef Kokkos::View<const T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstT3D;
- typedef Kokkos::View<T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedT3D;
- typedef Kokkos::View<const int***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstInt3D;
-
- // edge based scalar input field, dim: (nproma,nlev,nblks_e)
- UnmanagedConstT3D p_edge_in_view(p_edge_in, nproma, nlev, nblks_e);
-
- // line indices of edges of triangles, dim: (nproma,nblks_c, 3)
- UnmanagedConstInt3D iidx_view(edge_idx, nproma, nblks_c, 3); // edge_idx_view
-
- // block indices of edges of triangles, dim: (nproma,nblks_c, 3)
- UnmanagedConstInt3D iblk_view(edge_blk, nproma, nblks_c, 3); // edge_blk_view
-
- // coefficients for (area weighted) interpolation, dim: (nproma,3-cell_type,nblks_c)
- UnmanagedConstT3D coeff_int_view(coeff_int, nproma, 3, nblks_c);
-
- // cell based scalar output field, dim: (nproma,nlev,nblks_c)
- UnmanagedT3D p_cell_out_view(p_cell_out, nproma, nlev, nblks_c);
-
- int i_startidx, i_endidx;
-
- for (int jb = i_startblk; jb < i_endblk + 1; ++jb){
- get_indices_c_lib(i_startidx_in, i_endidx_in, nproma, jb, i_startblk, i_endblk,
- i_startidx, i_endidx);
-
- Kokkos::MDRangePolicy<Kokkos::Rank<2>> innerPolicy(
- {slev, i_startidx}, {elev + 1, i_endidx + 1});
-
- Kokkos::parallel_for("edges2cells_scalar_lib_inner", innerPolicy,
+void edges2cells_scalar_lib(const T *p_edge_in, const int *edge_idx,
+ const int *edge_blk, const T *coeff_int,
+ T *p_cell_out, const int i_startblk,
+ const int i_endblk, const int i_startidx_in,
+ const int i_endidx_in, const int slev,
+ const int elev, const int nproma, const int nlev,
+ const int nblks_e, const int nblks_c,
+ const bool lacc) {
+ // Wrap raw pointers in unmanaged Kokkos Views.
+ typedef Kokkos::View<const T ***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged>
+ UnmanagedConstT3D;
+ typedef Kokkos::View<T ***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged>
+ UnmanagedT3D;
+ typedef Kokkos::View<const int ***, Kokkos::LayoutLeft,
+ Kokkos::MemoryUnmanaged>
+ UnmanagedConstInt3D;
+
+ // edge based scalar input field, dim: (nproma,nlev,nblks_e)
+ UnmanagedConstT3D p_edge_in_view(p_edge_in, nproma, nlev, nblks_e);
+
+ // line indices of edges of triangles, dim: (nproma,nblks_c, 3)
+ UnmanagedConstInt3D iidx_view(edge_idx, nproma, nblks_c, 3); // edge_idx_view
+
+ // block indices of edges of triangles, dim: (nproma,nblks_c, 3)
+ UnmanagedConstInt3D iblk_view(edge_blk, nproma, nblks_c, 3); // edge_blk_view
+
+ // coefficients for (area weighted) interpolation, dim:
+ // (nproma,3-cell_type,nblks_c)
+ UnmanagedConstT3D coeff_int_view(coeff_int, nproma, 3, nblks_c);
+
+ // cell based scalar output field, dim: (nproma,nlev,nblks_c)
+ UnmanagedT3D p_cell_out_view(p_cell_out, nproma, nlev, nblks_c);
+
+ int i_startidx, i_endidx;
+
+ for (int jb = i_startblk; jb < i_endblk + 1; ++jb) {
+ get_indices_c_lib(i_startidx_in, i_endidx_in, nproma, jb, i_startblk,
+ i_endblk, i_startidx, i_endidx);
+
+ Kokkos::MDRangePolicy<Kokkos::Rank<2>> innerPolicy(
+ {slev, i_startidx}, {elev + 1, i_endidx + 1});
+
+ Kokkos::parallel_for(
+ "edges2cells_scalar_lib_inner", innerPolicy,
KOKKOS_LAMBDA(const int jk, const int jc) {
- p_cell_out_view(jc, jk, jb) =
- coeff_int_view(jc, 0, jb)*p_edge_in_view(iidx_view(jc, jb, 0), jk, iblk_view(jc, jb, 0)) +
- coeff_int_view(jc, 1, jb)*p_edge_in_view(iidx_view(jc, jb, 1), jk, iblk_view(jc, jb, 1)) +
- coeff_int_view(jc, 2, jb)*p_edge_in_view(iidx_view(jc, jb, 2), jk, iblk_view(jc, jb, 2));
- }
- );
- Kokkos::fence();
- }
-
+ p_cell_out_view(jc, jk, jb) =
+ coeff_int_view(jc, 0, jb) * p_edge_in_view(iidx_view(jc, jb, 0),
+ jk,
+ iblk_view(jc, jb, 0)) +
+ coeff_int_view(jc, 1, jb) * p_edge_in_view(iidx_view(jc, jb, 1),
+ jk,
+ iblk_view(jc, jb, 1)) +
+ coeff_int_view(jc, 2, jb) * p_edge_in_view(iidx_view(jc, jb, 2),
+ jk,
+ iblk_view(jc, jb, 2));
+ });
+ Kokkos::fence();
+ }
}
//------------------------------------------------------------------------
@@ -269,60 +310,77 @@ void edges2cells_scalar_lib(
/// Computes average of scalar fields from centers of cells to vertices.
///
template <typename T, typename S>
-void cells2verts_scalar_lib(
- const T* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const S* coeff_int, S* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async)
-{
- // Wrap raw pointers in unmanaged Kokkos Views.
- typedef Kokkos::View<const T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstT3D;
- typedef Kokkos::View<const S***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstS3D;
- typedef Kokkos::View<S***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedS3D;
- typedef Kokkos::View<const int***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstInt3D;
-
- // cell based scalar input field, dim: (nproma,nlev,nblks_c)
- UnmanagedConstT3D p_cell_in_view(p_cell_in, nproma, nlev, nblks_c);
-
- // line indices of cells around each vertex, dim: (nproma,nblks_v, 6)
- UnmanagedConstInt3D iidx_view(vert_cell_idx, nproma, nblks_v, 6); // vert_cell_idx_view
-
- // block indices of cells around each vertex, dim: (nproma,nblks_v, 6)
- UnmanagedConstInt3D iblk_view(vert_cell_blk, nproma, nblks_v, 6); // vert_cell_blk_view
-
- // coefficients for interpolation, dim: (nproma,9-cell_type,nblks_v)
- UnmanagedConstS3D coeff_int_view(coeff_int, nproma, 6, nblks_v);
-
- // vertex based scalar output field, dim: (nproma,nlev,nblks_c)
- UnmanagedS3D p_vert_out_view(p_vert_out, nproma, nlev, nblks_c);
-
- int i_startidx, i_endidx;
-
- for (int jb = i_startblk; jb < i_endblk + 1; ++jb){
-
- get_indices_v_lib(i_startidx_in, i_endidx_in, nproma, jb, i_startblk, i_endblk,
- i_startidx, i_endidx);
-
- Kokkos::MDRangePolicy<Kokkos::Rank<2>> innerPolicy(
- {slev, i_startidx}, {elev + 1, i_endidx + 1});
-
- Kokkos::parallel_for("cells2verts_scalar_lib", innerPolicy,
+void cells2verts_scalar_lib(const T *p_cell_in, const int *vert_cell_idx,
+ const int *vert_cell_blk, const S *coeff_int,
+ S *p_vert_out, const int i_startblk,
+ const int i_endblk, const int i_startidx_in,
+ const int i_endidx_in, const int slev,
+ const int elev, const int nproma, const int nlev,
+ const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async) {
+ // Wrap raw pointers in unmanaged Kokkos Views.
+ typedef Kokkos::View<const T ***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged>
+ UnmanagedConstT3D;
+ typedef Kokkos::View<const S ***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged>
+ UnmanagedConstS3D;
+ typedef Kokkos::View<S ***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged>
+ UnmanagedS3D;
+ typedef Kokkos::View<const int ***, Kokkos::LayoutLeft,
+ Kokkos::MemoryUnmanaged>
+ UnmanagedConstInt3D;
+
+ // cell based scalar input field, dim: (nproma,nlev,nblks_c)
+ UnmanagedConstT3D p_cell_in_view(p_cell_in, nproma, nlev, nblks_c);
+
+ // line indices of cells around each vertex, dim: (nproma,nblks_v, 6)
+ UnmanagedConstInt3D iidx_view(vert_cell_idx, nproma, nblks_v,
+ 6); // vert_cell_idx_view
+
+ // block indices of cells around each vertex, dim: (nproma,nblks_v, 6)
+ UnmanagedConstInt3D iblk_view(vert_cell_blk, nproma, nblks_v,
+ 6); // vert_cell_blk_view
+
+ // coefficients for interpolation, dim: (nproma,9-cell_type,nblks_v)
+ UnmanagedConstS3D coeff_int_view(coeff_int, nproma, 6, nblks_v);
+
+ // vertex based scalar output field, dim: (nproma,nlev,nblks_c)
+ UnmanagedS3D p_vert_out_view(p_vert_out, nproma, nlev, nblks_c);
+
+ int i_startidx, i_endidx;
+
+ for (int jb = i_startblk; jb < i_endblk + 1; ++jb) {
+
+ get_indices_v_lib(i_startidx_in, i_endidx_in, nproma, jb, i_startblk,
+ i_endblk, i_startidx, i_endidx);
+
+ Kokkos::MDRangePolicy<Kokkos::Rank<2>> innerPolicy(
+ {slev, i_startidx}, {elev + 1, i_endidx + 1});
+
+ Kokkos::parallel_for(
+ "cells2verts_scalar_lib", innerPolicy,
KOKKOS_LAMBDA(const int jk, const int jv) {
-
p_vert_out_view(jv, jk, jb) =
- coeff_int_view(jv, 0, jb)*p_cell_in_view(iidx_view(jv, jb, 0), jk, iblk_view(jv, jb, 0)) +
- coeff_int_view(jv, 1, jb)*p_cell_in_view(iidx_view(jv, jb, 1), jk, iblk_view(jv, jb, 1)) +
- coeff_int_view(jv, 2, jb)*p_cell_in_view(iidx_view(jv, jb, 2), jk, iblk_view(jv, jb, 2)) +
- coeff_int_view(jv, 3, jb)*p_cell_in_view(iidx_view(jv, jb, 3), jk, iblk_view(jv, jb, 3)) +
- coeff_int_view(jv, 4, jb)*p_cell_in_view(iidx_view(jv, jb, 4), jk, iblk_view(jv, jb, 4)) +
- coeff_int_view(jv, 5, jb)*p_cell_in_view(iidx_view(jv, jb, 5), jk, iblk_view(jv, jb, 5));
-
- }
- );
- Kokkos::fence();
- }
+ coeff_int_view(jv, 0, jb) * p_cell_in_view(iidx_view(jv, jb, 0),
+ jk,
+ iblk_view(jv, jb, 0)) +
+ coeff_int_view(jv, 1, jb) * p_cell_in_view(iidx_view(jv, jb, 1),
+ jk,
+ iblk_view(jv, jb, 1)) +
+ coeff_int_view(jv, 2, jb) * p_cell_in_view(iidx_view(jv, jb, 2),
+ jk,
+ iblk_view(jv, jb, 2)) +
+ coeff_int_view(jv, 3, jb) * p_cell_in_view(iidx_view(jv, jb, 3),
+ jk,
+ iblk_view(jv, jb, 3)) +
+ coeff_int_view(jv, 4, jb) * p_cell_in_view(iidx_view(jv, jb, 4),
+ jk,
+ iblk_view(jv, jb, 4)) +
+ coeff_int_view(jv, 5, jb) * p_cell_in_view(iidx_view(jv, jb, 5),
+ jk,
+ iblk_view(jv, jb, 5));
+ });
+ Kokkos::fence();
+ }
}
//-------------------------------------------------------------------------
@@ -331,49 +389,56 @@ void cells2verts_scalar_lib(
/// assumes reversed index order of the output field in loop exchange mode
///
template <typename T, typename S>
-void cells2verts_scalar_ri_lib(
- const T* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const T* coeff_int, S* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async)
-{
- // Wrap raw pointers in unmanaged Kokkos Views.
- typedef Kokkos::View<const T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstT3D;
- typedef Kokkos::View<S***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedS3D;
- typedef Kokkos::View<const int***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstInt3D;
-
- // cell based scalar input field, dim: (nproma,nlev,nblks_c)
- UnmanagedConstT3D p_cell_in_view(p_cell_in, nproma, nlev, nblks_c);
-
- // line indices of cells around each vertex, dim: (nproma,nblks_v, 6)
- UnmanagedConstInt3D iidx_view(vert_cell_idx, nproma, nblks_v, 6); // vert_cell_idx_view
-
- // block indices of cells around each vertex, dim: (nproma,nblks_v, 6)
- UnmanagedConstInt3D iblk_view(vert_cell_blk, nproma, nblks_v, 6); // vert_cell_blk_view
-
- // coefficients for interpolation, dim: (nproma,9-cell_type,nblks_v)
- UnmanagedConstT3D coeff_int_view(coeff_int, nproma, 6, nblks_v);
-
- // vertex based scalar output field, dim: (nproma,nlev,nblks_c)
+void cells2verts_scalar_ri_lib(const T *p_cell_in, const int *vert_cell_idx,
+ const int *vert_cell_blk, const T *coeff_int,
+ S *p_vert_out, const int i_startblk,
+ const int i_endblk, const int i_startidx_in,
+ const int i_endidx_in, const int slev,
+ const int elev, const int nproma, const int nlev,
+ const int nblks_c, const int nblks_v,
+ const bool lacc, const bool acc_async) {
+ // Wrap raw pointers in unmanaged Kokkos Views.
+ typedef Kokkos::View<const T ***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged>
+ UnmanagedConstT3D;
+ typedef Kokkos::View<S ***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged>
+ UnmanagedS3D;
+ typedef Kokkos::View<const int ***, Kokkos::LayoutLeft,
+ Kokkos::MemoryUnmanaged>
+ UnmanagedConstInt3D;
+
+ // cell based scalar input field, dim: (nproma,nlev,nblks_c)
+ UnmanagedConstT3D p_cell_in_view(p_cell_in, nproma, nlev, nblks_c);
+
+ // line indices of cells around each vertex, dim: (nproma,nblks_v, 6)
+ UnmanagedConstInt3D iidx_view(vert_cell_idx, nproma, nblks_v,
+ 6); // vert_cell_idx_view
+
+ // block indices of cells around each vertex, dim: (nproma,nblks_v, 6)
+ UnmanagedConstInt3D iblk_view(vert_cell_blk, nproma, nblks_v,
+ 6); // vert_cell_blk_view
+
+ // coefficients for interpolation, dim: (nproma,9-cell_type,nblks_v)
+ UnmanagedConstT3D coeff_int_view(coeff_int, nproma, 6, nblks_v);
+
+ // vertex based scalar output field, dim: (nproma,nlev,nblks_c)
#ifdef __LOOP_EXCHANGE
- UnmanagedS3D p_vert_out_view(p_vert_out, nproma, nlev, nblks_c);
+ UnmanagedS3D p_vert_out_view(p_vert_out, nproma, nlev, nblks_c);
#else
- UnmanagedS3D p_vert_out_view(p_vert_out, nlev, nproma, nblks_c);
+ UnmanagedS3D p_vert_out_view(p_vert_out, nlev, nproma, nblks_c);
#endif
- int i_startidx, i_endidx;
+ int i_startidx, i_endidx;
- for (int jb = i_startblk; jb < i_endblk + 1; ++jb){
+ for (int jb = i_startblk; jb < i_endblk + 1; ++jb) {
- get_indices_v_lib(i_startidx_in, i_endidx_in, nproma, jb, i_startblk, i_endblk,
- i_startidx, i_endidx);
+ get_indices_v_lib(i_startidx_in, i_endidx_in, nproma, jb, i_startblk,
+ i_endblk, i_startidx, i_endidx);
- Kokkos::MDRangePolicy<Kokkos::Rank<2>> innerPolicy(
- {slev, i_startidx}, {elev + 1, i_endidx + 1});
+ Kokkos::MDRangePolicy<Kokkos::Rank<2>> innerPolicy(
+ {slev, i_startidx}, {elev + 1, i_endidx + 1});
- Kokkos::parallel_for("cells2verts_scalar_ri_lib", innerPolicy,
+ Kokkos::parallel_for(
+ "cells2verts_scalar_ri_lib", innerPolicy,
KOKKOS_LAMBDA(const int jk, const int jv) {
#ifdef __LOOP_EXCHANGE
@@ -381,18 +446,27 @@ void cells2verts_scalar_ri_lib(
#else
p_vert_out_view(jk, jv, jb) =
#endif
- coeff_int_view(jv, 0, jb)*p_cell_in_view(iidx_view(jv, jb, 0), jk, iblk_view(jv, jb, 0)) +
- coeff_int_view(jv, 1, jb)*p_cell_in_view(iidx_view(jv, jb, 1), jk, iblk_view(jv, jb, 1)) +
- coeff_int_view(jv, 2, jb)*p_cell_in_view(iidx_view(jv, jb, 2), jk, iblk_view(jv, jb, 2)) +
- coeff_int_view(jv, 3, jb)*p_cell_in_view(iidx_view(jv, jb, 3), jk, iblk_view(jv, jb, 3)) +
- coeff_int_view(jv, 4, jb)*p_cell_in_view(iidx_view(jv, jb, 4), jk, iblk_view(jv, jb, 4)) +
- coeff_int_view(jv, 5, jb)*p_cell_in_view(iidx_view(jv, jb, 5), jk, iblk_view(jv, jb, 5));
-
- }
- );
- Kokkos::fence();
- }
-
+ coeff_int_view(jv, 0, jb) * p_cell_in_view(iidx_view(jv, jb, 0),
+ jk,
+ iblk_view(jv, jb, 0)) +
+ coeff_int_view(jv, 1, jb) * p_cell_in_view(iidx_view(jv, jb, 1),
+ jk,
+ iblk_view(jv, jb, 1)) +
+ coeff_int_view(jv, 2, jb) * p_cell_in_view(iidx_view(jv, jb, 2),
+ jk,
+ iblk_view(jv, jb, 2)) +
+ coeff_int_view(jv, 3, jb) * p_cell_in_view(iidx_view(jv, jb, 3),
+ jk,
+ iblk_view(jv, jb, 3)) +
+ coeff_int_view(jv, 4, jb) * p_cell_in_view(iidx_view(jv, jb, 4),
+ jk,
+ iblk_view(jv, jb, 4)) +
+ coeff_int_view(jv, 5, jb) * p_cell_in_view(iidx_view(jv, jb, 5),
+ jk,
+ iblk_view(jv, jb, 5));
+ });
+ Kokkos::fence();
+ }
}
//-------------------------------------------------------------------------
@@ -400,56 +474,68 @@ void cells2verts_scalar_ri_lib(
/// Computes average of scalar fields from vertices to centers of cells.
///
template <typename T>
-void verts2cells_scalar_lib(
- const T* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
- const T* coeff_int, T* p_cell_out, const int nblks_c, const int npromz_c,
- const int slev, const int elev, const int nproma, const int nlev,
- const int nblks_v, const bool lacc)
-{
- // Wrap raw pointers in unmanaged Kokkos Views.
- typedef Kokkos::View<const T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstT3D;
- typedef Kokkos::View<T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedT3D;
- typedef Kokkos::View<const int***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstInt3D;
-
- // cell based scalar input field, dim: (nproma,nlev,nblks_v)
- UnmanagedConstT3D p_vert_in_view(p_vert_in, nproma, nlev, nblks_v);
-
- // line indices of vertices of triangles, dim: (nproma,nblks_c, 3)
- UnmanagedConstInt3D iidx_view(cell_index_idx, nproma, nblks_c, 3); // cell_vertex_idx
-
- // block indices of vertices of triangles, dim: (nproma,nblks_c, 3)
- UnmanagedConstInt3D iblk_view(cell_vertex_blk, nproma, nblks_c, 3); // cell_vertex_blk
-
- // coefficients for interpolation, dim: (nproma, 3, nblks_c)
- UnmanagedConstT3D coeff_int_view(coeff_int, nproma, 3, nblks_c);
-
- // vertex based scalar output field, dim: (nproma,nlev,nblks_c)
- UnmanagedT3D p_cell_out_view(p_cell_out, nproma, nlev, nblks_c);
-
- for (int jb = 0; jb<nblks_c; ++jb){
+void verts2cells_scalar_lib(const T *p_vert_in, const int *cell_index_idx,
+ const int *cell_vertex_blk, const T *coeff_int,
+ T *p_cell_out, const int nblks_c,
+ const int npromz_c, const int slev, const int elev,
+ const int nproma, const int nlev, const int nblks_v,
+ const bool lacc) {
+ // Wrap raw pointers in unmanaged Kokkos Views.
+ typedef Kokkos::View<const T ***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged>
+ UnmanagedConstT3D;
+ typedef Kokkos::View<T ***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged>
+ UnmanagedT3D;
+ typedef Kokkos::View<const int ***, Kokkos::LayoutLeft,
+ Kokkos::MemoryUnmanaged>
+ UnmanagedConstInt3D;
+
+ // cell based scalar input field, dim: (nproma,nlev,nblks_v)
+ UnmanagedConstT3D p_vert_in_view(p_vert_in, nproma, nlev, nblks_v);
+
+ // line indices of vertices of triangles, dim: (nproma,nblks_c, 3)
+ UnmanagedConstInt3D iidx_view(cell_index_idx, nproma, nblks_c,
+ 3); // cell_vertex_idx
+
+ // block indices of vertices of triangles, dim: (nproma,nblks_c, 3)
+ UnmanagedConstInt3D iblk_view(cell_vertex_blk, nproma, nblks_c,
+ 3); // cell_vertex_blk
+
+ // coefficients for interpolation, dim: (nproma, 3, nblks_c)
+ UnmanagedConstT3D coeff_int_view(coeff_int, nproma, 3, nblks_c);
+
+ // vertex based scalar output field, dim: (nproma,nlev,nblks_c)
+ UnmanagedT3D p_cell_out_view(p_cell_out, nproma, nlev, nblks_c);
+
+ for (int jb = 0; jb < nblks_c; ++jb) {
+
+ int nlen;
+ if (jb != nblks_c) {
+ nlen = nproma;
+ } else {
+ nlen = npromz_c;
+ }
- int nlen;
- if (jb != nblks_c){ nlen = nproma; }
- else { nlen = npromz_c; }
+ Kokkos::MDRangePolicy<Kokkos::Rank<2>> innerPolicy({slev, 0},
+ {elev + 1, nlen});
- Kokkos::MDRangePolicy<Kokkos::Rank<2>> innerPolicy(
- {slev, 0}, {elev + 1, nlen});
-
- Kokkos::parallel_for("cell_avg_lib_inner", innerPolicy,
+ Kokkos::parallel_for(
+ "cell_avg_lib_inner", innerPolicy,
KOKKOS_LAMBDA(const int jk, const int jc) {
-
p_cell_out_view(jc, jk, jb) =
- coeff_int_view(jc, 0, jb)*p_vert_in_view(iidx_view(jc, jb, 0), jk, iblk_view(jc, jb, 0)) +
- coeff_int_view(jc, 1, jb)*p_vert_in_view(iidx_view(jc, jb, 1), jk, iblk_view(jc, jb, 1)) +
- coeff_int_view(jc, 2, jb)*p_vert_in_view(iidx_view(jc, jb, 2), jk, iblk_view(jc, jb, 2));
-
- }
- );
- Kokkos::fence();
- }
+ coeff_int_view(jc, 0, jb) * p_vert_in_view(iidx_view(jc, jb, 0),
+ jk,
+ iblk_view(jc, jb, 0)) +
+ coeff_int_view(jc, 1, jb) * p_vert_in_view(iidx_view(jc, jb, 1),
+ jk,
+ iblk_view(jc, jb, 1)) +
+ coeff_int_view(jc, 2, jb) * p_vert_in_view(iidx_view(jc, jb, 2),
+ jk,
+ iblk_view(jc, jb, 2));
+ });
+ Kokkos::fence();
+ }
}
-
//-------------------------------------------------------------------------
//>
/// Computes the average of a cell-based variable.
@@ -462,52 +548,60 @@ void verts2cells_scalar_lib(
/// output: lives on centers of triangles
///
template <typename T>
-void cell_avg_lib(
- const T* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
- const T* avg_coeff, T* avg_psi_c, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const bool lacc)
-{
- // Wrap raw pointers in unmanaged Kokkos Views.
- typedef Kokkos::View<const T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstT3D;
- typedef Kokkos::View<T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedT3D;
- typedef Kokkos::View<const int***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstInt3D;
-
- // cell based variable before averaging, dim: (nproma,nlev,nblks_c)
- UnmanagedConstT3D psi_c_view(psi_c, nproma, nlev, nblks_c);
- // line indices of triangles next to each cell, dim: (nproma,nblks_c, 3)
- UnmanagedConstInt3D iidx_view(cell_neighbor_idx, nproma, nblks_c, 3); // cell_neighbour_idx
- // block indices of triangles next to each cell, dim: (nproma,nblks_c, 3)
- UnmanagedConstInt3D iblk_view(cell_neighbor_blk, nproma, nblks_c, 3); // cell_neighbour_blk
- // averaging coefficients, dim: (nproma,nlev,nblks_c)
- UnmanagedConstT3D avg_coeff_view(avg_coeff, nproma, nlev, nblks_c);
-
- // cell based variable after averaging, dim: (nproma,nlev,nblks_c)
- UnmanagedT3D avg_psi_c_view(avg_psi_c, nproma, nlev, nblks_c);
-
- int i_startidx, i_endidx;
-
- for (int jb = i_startblk; jb<i_endblk + 1; ++jb){
- get_indices_c_lib(i_startidx_in, i_endidx_in, nproma, jb, i_startblk, i_endblk,
- i_startidx, i_endidx);
-
- Kokkos::MDRangePolicy<Kokkos::Rank<2>> innerPolicy(
- {slev, i_startidx}, {elev + 1, i_endidx + 1});
-
- Kokkos::parallel_for("cell_avg_lib_inner", innerPolicy,
+void cell_avg_lib(const T *psi_c, const int *cell_neighbor_idx,
+ const int *cell_neighbor_blk, const T *avg_coeff,
+ T *avg_psi_c, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const bool lacc) {
+ // Wrap raw pointers in unmanaged Kokkos Views.
+ typedef Kokkos::View<const T ***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged>
+ UnmanagedConstT3D;
+ typedef Kokkos::View<T ***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged>
+ UnmanagedT3D;
+ typedef Kokkos::View<const int ***, Kokkos::LayoutLeft,
+ Kokkos::MemoryUnmanaged>
+ UnmanagedConstInt3D;
+
+ // cell based variable before averaging, dim: (nproma,nlev,nblks_c)
+ UnmanagedConstT3D psi_c_view(psi_c, nproma, nlev, nblks_c);
+ // line indices of triangles next to each cell, dim: (nproma,nblks_c, 3)
+ UnmanagedConstInt3D iidx_view(cell_neighbor_idx, nproma, nblks_c,
+ 3); // cell_neighbour_idx
+ // block indices of triangles next to each cell, dim: (nproma,nblks_c, 3)
+ UnmanagedConstInt3D iblk_view(cell_neighbor_blk, nproma, nblks_c,
+ 3); // cell_neighbour_blk
+ // averaging coefficients, dim: (nproma,nlev,nblks_c)
+ UnmanagedConstT3D avg_coeff_view(avg_coeff, nproma, nlev, nblks_c);
+
+ // cell based variable after averaging, dim: (nproma,nlev,nblks_c)
+ UnmanagedT3D avg_psi_c_view(avg_psi_c, nproma, nlev, nblks_c);
+
+ int i_startidx, i_endidx;
+
+ for (int jb = i_startblk; jb < i_endblk + 1; ++jb) {
+ get_indices_c_lib(i_startidx_in, i_endidx_in, nproma, jb, i_startblk,
+ i_endblk, i_startidx, i_endidx);
+
+ Kokkos::MDRangePolicy<Kokkos::Rank<2>> innerPolicy(
+ {slev, i_startidx}, {elev + 1, i_endidx + 1});
+
+ Kokkos::parallel_for(
+ "cell_avg_lib_inner", innerPolicy,
KOKKOS_LAMBDA(const int jk, const int jc) {
- // calculate the weighted average
-
- avg_psi_c_view(jc, jk, jb) =
- psi_c_view(jc, jk, jb)*avg_coeff_view(jc, 0, jb) +
- psi_c_view(iidx_view(jc, jb, 0), jk, iblk_view(jc, jb, 0))*avg_coeff_view(jc, 1, jb) +
- psi_c_view(iidx_view(jc, jb, 1), jk, iblk_view(jc, jb, 1))*avg_coeff_view(jc, 2, jb) +
- psi_c_view(iidx_view(jc, jb, 2), jk, iblk_view(jc, jb, 2))*avg_coeff_view(jc, 3, jb);
- }
- );
- Kokkos::fence();
- }
+ // calculate the weighted average
+
+ avg_psi_c_view(jc, jk, jb) =
+ psi_c_view(jc, jk, jb) * avg_coeff_view(jc, 0, jb) +
+ psi_c_view(iidx_view(jc, jb, 0), jk, iblk_view(jc, jb, 0)) *
+ avg_coeff_view(jc, 1, jb) +
+ psi_c_view(iidx_view(jc, jb, 1), jk, iblk_view(jc, jb, 1)) *
+ avg_coeff_view(jc, 2, jb) +
+ psi_c_view(iidx_view(jc, jb, 2), jk, iblk_view(jc, jb, 2)) *
+ avg_coeff_view(jc, 3, jb);
+ });
+ Kokkos::fence();
+ }
}
//-----------------------------------------------------------------------
@@ -516,169 +610,144 @@ void cell_avg_lib(
//
//-----------------------------------------------------------------------
-template
-void verts2edges_scalar_lib<double>(const double* p_vertex_in,
- const int* edge_vertex_idx, const int* edge_vertex_blk,
- const double* coeff_int,
- double* p_edge_out,
- const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
- const int nproma, const int nlev, const int nblks_v, const int nblks_e, const bool lacc);
-
-template
-void verts2edges_scalar_lib<float>(const float* p_vertex_in,
- const int* edge_vertex_idx, const int* edge_vertex_blk,
- const float* coeff_int,
- float* p_edge_out,
- const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
- const int nproma, const int nlev, const int nblks_v, const int nblks_e, const bool lacc);
-
-template
-void cells2edges_scalar_lib<double, double>(const double* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
- const double* coeff_int, double* p_edge_out, const int* i_startblk_in,
- const int* i_endblk_in, const int* i_startidx_in, const int* i_endidx_in,
- const int slev, const int elev, const int nproma, const int nlev,
- const int nblk_c, const int nblks_e, const int patch_id,
- const bool l_limited_area, const bool lfill_latbc, const bool lacc);
-
-template
-void cells2edges_scalar_lib<float, float>(const float* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
- const float* coeff_int, float* p_edge_out, const int* i_startblk_in,
- const int* i_endblk_in, const int* i_startidx_in, const int* i_endidx_in,
- const int slev, const int elev, const int nproma, const int nlev,
- const int nblk_c, const int nblks_e, const int patch_id,
- const bool l_limited_area, const bool lfill_latbc, const bool lacc);
-
-template // sp2dp
-void cells2edges_scalar_lib<float, double>(const float* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
- const double* coeff_int, double* p_edge_out, const int* i_startblk_in,
- const int* i_endblk_in, const int* i_startidx_in, const int* i_endidx_in,
- const int slev, const int elev, const int nproma, const int nlev,
- const int nblk_c, const int nblks_e, const int patch_id,
- const bool l_limited_area, const bool lfill_latbc, const bool lacc);
-
-
-template
-void edges2verts_scalar_lib<double> (const double* p_edge_in, const int* vert_edge_idx, const int* vert_edge_blk,
- const double* v_int, double* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in, const int slev,
- const int elev, const int nproma, const int nlev, const int nblks_e,
- const int nblks_v, const bool lacc);
-
-template
-void edges2verts_scalar_lib <float> (const float* p_edge_in, const int* vert_edge_idx, const int* vert_edge_blk,
- const float* v_int, float* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in, const int slev,
- const int elev, const int nproma, const int nlev, const int nblks_e,
- const int nblks_v, const bool lacc);
-
-
-template
-void edges2cells_scalar_lib<double>(
- const double* p_edge_in, const int* edge_idx, const int* edge_blk,
- const double* coeff_int, double* p_cell_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_e, const int nblks_c,
- const bool lacc);
-
-template
-void edges2cells_scalar_lib<float>(
- const float* p_edge_in, const int* edge_idx, const int* edge_blk,
- const float* coeff_int, float* p_cell_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_e, const int nblks_c,
- const bool lacc);
-
-
-template
-void cells2verts_scalar_lib<double,double>(
- const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const double* coeff_int, double* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async);
-
-template
-void cells2verts_scalar_lib<float,double>(
- const float* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const double* coeff_int, double* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async);
-
-template
-void cells2verts_scalar_lib<float,float>(
- const float* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async);
-
-
-template
-void cells2verts_scalar_ri_lib<double,double>(
- const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const double* coeff_int, double* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async);
-
-template
-void cells2verts_scalar_ri_lib<double,float>(
- const double* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const double* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async);
-
-template
-void cells2verts_scalar_ri_lib<float,float>(
- const float* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const float* coeff_int, float* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
- const bool lacc, const bool acc_async);
-
-
-
-template
-void verts2cells_scalar_lib<double>(
- const double* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
- const double* coeff_int, double* p_cell_out, const int nblks_c, const int npromz_c,
+template void verts2edges_scalar_lib<double>(
+ const double *p_vertex_in, const int *edge_vertex_idx,
+ const int *edge_vertex_blk, const double *coeff_int, double *p_edge_out,
+ const int i_startblk, const int i_endblk, const int i_startidx_in,
+ const int i_endidx_in, const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_v, const int nblks_e, const bool lacc);
+
+template void verts2edges_scalar_lib<float>(
+ const float *p_vertex_in, const int *edge_vertex_idx,
+ const int *edge_vertex_blk, const float *coeff_int, float *p_edge_out,
+ const int i_startblk, const int i_endblk, const int i_startidx_in,
+ const int i_endidx_in, const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_v, const int nblks_e, const bool lacc);
+
+template void cells2edges_scalar_lib<double, double>(
+ const double *p_cell_in, const int *edge_cell_idx, const int *edge_cell_blk,
+ const double *coeff_int, double *p_edge_out, const int *i_startblk_in,
+ const int *i_endblk_in, const int *i_startidx_in, const int *i_endidx_in,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblk_c, const int nblks_e, const int patch_id,
+ const bool l_limited_area, const bool lfill_latbc, const bool lacc);
+
+template void cells2edges_scalar_lib<float, float>(
+ const float *p_cell_in, const int *edge_cell_idx, const int *edge_cell_blk,
+ const float *coeff_int, float *p_edge_out, const int *i_startblk_in,
+ const int *i_endblk_in, const int *i_startidx_in, const int *i_endidx_in,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblk_c, const int nblks_e, const int patch_id,
+ const bool l_limited_area, const bool lfill_latbc, const bool lacc);
+
+// sp2dp
+template void cells2edges_scalar_lib<float, double>(
+ const float *p_cell_in, const int *edge_cell_idx, const int *edge_cell_blk,
+ const double *coeff_int, double *p_edge_out, const int *i_startblk_in,
+ const int *i_endblk_in, const int *i_startidx_in, const int *i_endidx_in,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblk_c, const int nblks_e, const int patch_id,
+ const bool l_limited_area, const bool lfill_latbc, const bool lacc);
+
+template void edges2verts_scalar_lib<double>(
+ const double *p_edge_in, const int *vert_edge_idx, const int *vert_edge_blk,
+ const double *v_int, double *p_vert_out, const int i_startblk,
+ const int i_endblk, const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblks_e, const int nblks_v, const bool lacc);
+
+template void edges2verts_scalar_lib<float>(
+ const float *p_edge_in, const int *vert_edge_idx, const int *vert_edge_blk,
+ const float *v_int, float *p_vert_out, const int i_startblk,
+ const int i_endblk, const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblks_e, const int nblks_v, const bool lacc);
+
+template void edges2cells_scalar_lib<double>(
+ const double *p_edge_in, const int *edge_idx, const int *edge_blk,
+ const double *coeff_int, double *p_cell_out, const int i_startblk,
+ const int i_endblk, const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblks_e, const int nblks_c, const bool lacc);
+
+template void edges2cells_scalar_lib<float>(
+ const float *p_edge_in, const int *edge_idx, const int *edge_blk,
+ const float *coeff_int, float *p_cell_out, const int i_startblk,
+ const int i_endblk, const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblks_e, const int nblks_c, const bool lacc);
+
+template void cells2verts_scalar_lib<double, double>(
+ const double *p_cell_in, const int *vert_cell_idx, const int *vert_cell_blk,
+ const double *coeff_int, double *p_vert_out, const int i_startblk,
+ const int i_endblk, const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblks_c, const int nblks_v, const bool lacc,
+ const bool acc_async);
+
+template void cells2verts_scalar_lib<float, double>(
+ const float *p_cell_in, const int *vert_cell_idx, const int *vert_cell_blk,
+ const double *coeff_int, double *p_vert_out, const int i_startblk,
+ const int i_endblk, const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblks_c, const int nblks_v, const bool lacc,
+ const bool acc_async);
+
+template void cells2verts_scalar_lib<float, float>(
+ const float *p_cell_in, const int *vert_cell_idx, const int *vert_cell_blk,
+ const float *coeff_int, float *p_vert_out, const int i_startblk,
+ const int i_endblk, const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblks_c, const int nblks_v, const bool lacc,
+ const bool acc_async);
+
+template void cells2verts_scalar_ri_lib<double, double>(
+ const double *p_cell_in, const int *vert_cell_idx, const int *vert_cell_blk,
+ const double *coeff_int, double *p_vert_out, const int i_startblk,
+ const int i_endblk, const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma, const int nlev,
+ const int nblks_c, const int nblks_v, const bool lacc,
+ const bool acc_async);
+
+template void cells2verts_scalar_ri_lib<double, float>(
+ const double *p_cell_in, const int *vert_cell_idx, const int *vert_cell_blk,
+ const double *coeff_int, float *p_vert_out, const int i_startblk,
+ const int i_endblk, const int i_startidx_in, const int i_endidx_in,
const int slev, const int elev, const int nproma, const int nlev,
- const int nblks_v, const bool lacc);
+ const int nblks_c, const int nblks_v, const bool lacc,
+ const bool acc_async);
-template
-void verts2cells_scalar_lib<float>(
- const float* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
- const float* coeff_int, float* p_cell_out, const int nblks_c, const int npromz_c,
+template void cells2verts_scalar_ri_lib<float, float>(
+ const float *p_cell_in, const int *vert_cell_idx, const int *vert_cell_blk,
+ const float *coeff_int, float *p_vert_out, const int i_startblk,
+ const int i_endblk, const int i_startidx_in, const int i_endidx_in,
const int slev, const int elev, const int nproma, const int nlev,
- const int nblks_v, const bool lacc);
-
-
-template
-void cell_avg_lib<double>(
- const double* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
- const double* avg_coeff, double* avg_psi_c, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c,
- const bool lacc);
-
-template
-void cell_avg_lib<float>(
- const float* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
- const float* avg_coeff, float* avg_psi_c, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c,
- const bool lacc);
+ const int nblks_c, const int nblks_v, const bool lacc,
+ const bool acc_async);
+
+template void verts2cells_scalar_lib<double>(
+ const double *p_vert_in, const int *cell_index_idx,
+ const int *cell_vertex_blk, const double *coeff_int, double *p_cell_out,
+ const int nblks_c, const int npromz_c, const int slev, const int elev,
+ const int nproma, const int nlev, const int nblks_v, const bool lacc);
+
+template void verts2cells_scalar_lib<float>(
+ const float *p_vert_in, const int *cell_index_idx,
+ const int *cell_vertex_blk, const float *coeff_int, float *p_cell_out,
+ const int nblks_c, const int npromz_c, const int slev, const int elev,
+ const int nproma, const int nlev, const int nblks_v, const bool lacc);
+
+template void cell_avg_lib<double>(
+ const double *psi_c, const int *cell_neighbor_idx,
+ const int *cell_neighbor_blk, const double *avg_coeff, double *avg_psi_c,
+ const int i_startblk, const int i_endblk, const int i_startidx_in,
+ const int i_endidx_in, const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const bool lacc);
+
+template void
+cell_avg_lib<float>(const float *psi_c, const int *cell_neighbor_idx,
+ const int *cell_neighbor_blk, const float *avg_coeff,
+ float *avg_psi_c, const int i_startblk, const int i_endblk,
+ const int i_startidx_in, const int i_endidx_in,
+ const int slev, const int elev, const int nproma,
+ const int nlev, const int nblks_c, const bool lacc);
diff --git a/src/interpolation/mo_lib_interpolation_scalar.hpp b/src/interpolation/mo_lib_interpolation_scalar.hpp
index 85c8fc5..8c8d2de 100644
--- a/src/interpolation/mo_lib_interpolation_scalar.hpp
+++ b/src/interpolation/mo_lib_interpolation_scalar.hpp
@@ -12,63 +12,79 @@
#pragma once
template <typename T>
-void verts2edges_scalar_lib(const T* p_vertex_in,
- const int* edge_vertex_idx, const int* edge_vertex_blk,
- const T* coeff_int,
- T* p_edge_out,
- const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in, const int slev, const int elev,
- const int nproma, const int nlev, const int nblks_v, const int nblks_e, const bool lacc);
+void verts2edges_scalar_lib(const T *p_vertex_in, const int *edge_vertex_idx,
+ const int *edge_vertex_blk, const T *coeff_int,
+ T *p_edge_out, const int i_startblk,
+ const int i_endblk, const int i_startidx_in,
+ const int i_endidx_in, const int slev,
+ const int elev, const int nproma, const int nlev,
+ const int nblks_v, const int nblks_e,
+ const bool lacc);
;
template <typename T, typename S>
-void cells2edges_scalar_lib(const T* p_cell_in, const int* edge_cell_idx, const int* edge_cell_blk,
- const S* coeff_int, S* p_edge_out, const int* i_startblk_in,
- const int* i_endblk_in, const int* i_startidx_in, const int* i_endidx_in,
- const int slev, const int elev, const int nproma, const int nlev,
- const int nblk_c, const int nblks_e, const int patch_id,
- const bool l_limited_area, const bool lfill_latbc, const bool lacc);
+void cells2edges_scalar_lib(const T *p_cell_in, const int *edge_cell_idx,
+ const int *edge_cell_blk, const S *coeff_int,
+ S *p_edge_out, const int *i_startblk_in,
+ const int *i_endblk_in, const int *i_startidx_in,
+ const int *i_endidx_in, const int slev,
+ const int elev, const int nproma, const int nlev,
+ const int nblk_c, const int nblks_e,
+ const int patch_id, const bool l_limited_area,
+ const bool lfill_latbc, const bool lacc);
template <typename T>
-void edges2verts_scalar_lib (const T* p_edge_in, const int* vert_edge_idx, const int* vert_edge_blk,
- const T* v_int, T* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in, const int slev,
- const int elev, const int nproma, const int nlev, const int nblks_e,
- const int nblks_v, const bool lacc);
+void edges2verts_scalar_lib(const T *p_edge_in, const int *vert_edge_idx,
+ const int *vert_edge_blk, const T *v_int,
+ T *p_vert_out, const int i_startblk,
+ const int i_endblk, const int i_startidx_in,
+ const int i_endidx_in, const int slev,
+ const int elev, const int nproma, const int nlev,
+ const int nblks_e, const int nblks_v,
+ const bool lacc);
template <typename T>
-void edges2cells_scalar_lib(const T* p_edge_in, const int* edge_idx, const int* edge_blk,
- const T* coeff_int, T* p_cell_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_e, const int nblks_c,
+void edges2cells_scalar_lib(const T *p_edge_in, const int *edge_idx,
+ const int *edge_blk, const T *coeff_int,
+ T *p_cell_out, const int i_startblk,
+ const int i_endblk, const int i_startidx_in,
+ const int i_endidx_in, const int slev,
+ const int elev, const int nproma, const int nlev,
+ const int nblks_e, const int nblks_c,
const bool lacc);
template <typename T, typename S>
-void cells2verts_scalar_lib(const T* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const S* coeff_int, S* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
+void cells2verts_scalar_lib(const T *p_cell_in, const int *vert_cell_idx,
+ const int *vert_cell_blk, const S *coeff_int,
+ S *p_vert_out, const int i_startblk,
+ const int i_endblk, const int i_startidx_in,
+ const int i_endidx_in, const int slev,
+ const int elev, const int nproma, const int nlev,
+ const int nblks_c, const int nblks_v,
const bool lacc, const bool acc_async);
template <typename T, typename S>
-void cells2verts_scalar_ri_lib(const T* p_cell_in, const int* vert_cell_idx, const int* vert_cell_blk,
- const T* coeff_int, S* p_vert_out, const int i_startblk, const int i_endblk,
- const int i_startidx_in, const int i_endidx_in,
- const int slev, const int elev, const int nproma,
- const int nlev, const int nblks_c, const int nblks_v,
+void cells2verts_scalar_ri_lib(const T *p_cell_in, const int *vert_cell_idx,
+ const int *vert_cell_blk, const T *coeff_int,
+ S *p_vert_out, const int i_startblk,
+ const int i_endblk, const int i_startidx_in,
+ const int i_endidx_in, const int slev,
+ const int elev, const int nproma, const int nlev,
+ const int nblks_c, const int nblks_v,
const bool lacc, const bool acc_async);
template <typename T>
-void verts2cells_scalar_lib(const T* p_vert_in, const int* cell_index_idx, const int* cell_vertex_blk,
- const T* coeff_int, T* p_cell_out, const int nblks_c, const int npromz_c,
- const int slev, const int elev, const int nproma, const int nlev,
- const int nblks_v, const bool lacc);
+void verts2cells_scalar_lib(const T *p_vert_in, const int *cell_index_idx,
+ const int *cell_vertex_blk, const T *coeff_int,
+ T *p_cell_out, const int nblks_c,
+ const int npromz_c, const int slev, const int elev,
+ const int nproma, const int nlev, const int nblks_v,
+ const bool lacc);
template <typename T>
-void cell_avg_lib(const T* psi_c, const int* cell_neighbor_idx, const int* cell_neighbor_blk,
- const T* avg_coeff, T* avg_psi_c, const int i_startblk, const int i_endblk,
+void cell_avg_lib(const T *psi_c, const int *cell_neighbor_idx,
+ const int *cell_neighbor_blk, const T *avg_coeff,
+ T *avg_psi_c, const int i_startblk, const int i_endblk,
const int i_startidx_in, const int i_endidx_in,
const int slev, const int elev, const int nproma,
const int nlev, const int nblks_c, const bool lacc);
diff --git a/src/interpolation/mo_lib_interpolation_vector.cpp b/src/interpolation/mo_lib_interpolation_vector.cpp
index fa1ed32..8e6a28e 100644
--- a/src/interpolation/mo_lib_interpolation_vector.cpp
+++ b/src/interpolation/mo_lib_interpolation_vector.cpp
@@ -12,108 +12,111 @@
#include "mo_lib_interpolation_vector.hpp"
template <typename T>
-void edges2cells_vector_lib(
- const T* p_vn_in, const T* p_vt_in,
- const int* cell_edge_idx, const int* cell_edge_blk,
- const T* e_bln_c_u, const T* e_bln_c_v,
- T* p_u_out, T* p_v_out,
- // Additional integer parameters.
- int i_startblk, int i_endblk,
- int i_startidx_in, int i_endidx_in,
- int slev, int elev,
- int nproma,
- // Dimensions for the arrays.
- int nlev, int nblks_e, int nblks_c)
-{
-
- // Wrap raw pointers in unmanaged Kokkos Views.
- typedef Kokkos::View<const T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstT3D;
- typedef Kokkos::View<T***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedT3D;
- typedef Kokkos::View<const int***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged> UnmanagedConstInt3D;
+void edges2cells_vector_lib(const T *p_vn_in, const T *p_vt_in,
+ const int *cell_edge_idx, const int *cell_edge_blk,
+ const T *e_bln_c_u, const T *e_bln_c_v, T *p_u_out,
+ T *p_v_out,
+ // Additional integer parameters.
+ int i_startblk, int i_endblk, int i_startidx_in,
+ int i_endidx_in, int slev, int elev, int nproma,
+ // Dimensions for the arrays.
+ int nlev, int nblks_e, int nblks_c) {
+ // Wrap raw pointers in unmanaged Kokkos Views.
+ typedef Kokkos::View<const T ***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged>
+ UnmanagedConstT3D;
+ typedef Kokkos::View<T ***, Kokkos::LayoutLeft, Kokkos::MemoryUnmanaged>
+ UnmanagedT3D;
+ typedef Kokkos::View<const int ***, Kokkos::LayoutLeft,
+ Kokkos::MemoryUnmanaged>
+ UnmanagedConstInt3D;
- UnmanagedConstT3D p_vn_in_view(p_vn_in, nproma, nlev, nblks_e);
- UnmanagedConstT3D p_vt_in_view(p_vt_in, nproma, nlev, nblks_e);
+ UnmanagedConstT3D p_vn_in_view(p_vn_in, nproma, nlev, nblks_e);
+ UnmanagedConstT3D p_vt_in_view(p_vt_in, nproma, nlev, nblks_e);
- UnmanagedConstInt3D cell_edge_idx_view(cell_edge_idx, nproma, nblks_c, 3);
- UnmanagedConstInt3D cell_edge_blk_view(cell_edge_blk, nproma, nblks_c, 3);
+ UnmanagedConstInt3D cell_edge_idx_view(cell_edge_idx, nproma, nblks_c, 3);
+ UnmanagedConstInt3D cell_edge_blk_view(cell_edge_blk, nproma, nblks_c, 3);
- UnmanagedConstT3D e_bln_c_u_view(e_bln_c_u, nproma, 6, nblks_c);
- UnmanagedConstT3D e_bln_c_v_view(e_bln_c_v, nproma, 6, nblks_c);
+ UnmanagedConstT3D e_bln_c_u_view(e_bln_c_u, nproma, 6, nblks_c);
+ UnmanagedConstT3D e_bln_c_v_view(e_bln_c_v, nproma, 6, nblks_c);
- UnmanagedT3D p_u_out_view(p_u_out, nproma, nlev, nblks_c);
- UnmanagedT3D p_v_out_view(p_v_out, nproma, nlev, nblks_c);
+ UnmanagedT3D p_u_out_view(p_u_out, nproma, nlev, nblks_c);
+ UnmanagedT3D p_v_out_view(p_v_out, nproma, nlev, nblks_c);
- // Loop over cell blocks as in the original Fortran code.
- for (int jb = i_startblk; jb <= i_endblk; ++jb) {
- // Call get_indices_c_lib to get inner loop indices for block jb.
- int i_startidx, i_endidx;
- get_indices_c_lib(i_startidx_in, i_endidx_in, nproma,
- jb, i_startblk, i_endblk,
- i_startidx, i_endidx);
+ // Loop over cell blocks as in the original Fortran code.
+ for (int jb = i_startblk; jb <= i_endblk; ++jb) {
+ // Call get_indices_c_lib to get inner loop indices for block jb.
+ int i_startidx, i_endidx;
+ get_indices_c_lib(i_startidx_in, i_endidx_in, nproma, jb, i_startblk,
+ i_endblk, i_startidx, i_endidx);
- Kokkos::MDRangePolicy<Kokkos::Rank<2>> innerPolicy(
- {slev, i_startidx}, {elev + 1, i_endidx + 1});
- Kokkos::parallel_for("edges2cells_inner", innerPolicy,
+ Kokkos::MDRangePolicy<Kokkos::Rank<2>> innerPolicy(
+ {slev, i_startidx}, {elev + 1, i_endidx + 1});
+ Kokkos::parallel_for(
+ "edges2cells_inner", innerPolicy,
KOKKOS_LAMBDA(const int jk, const int jc) {
- // Compute the bilinear interpolation for cell (jc, jk, jb).
- p_u_out_view(jc, jk, jb) =
- e_bln_c_u_view(jc, 0, jb) *
- p_vn_in_view(cell_edge_idx_view(jc, jb, 0) - 1, jk, cell_edge_blk_view(jc, jb, 0) - 1) +
- e_bln_c_u_view(jc, 1, jb) *
- p_vt_in_view(cell_edge_idx_view(jc, jb, 0) - 1, jk, cell_edge_blk_view(jc, jb, 0) - 1) +
- e_bln_c_u_view(jc, 2, jb) *
- p_vn_in_view(cell_edge_idx_view(jc, jb, 1) - 1, jk, cell_edge_blk_view(jc, jb, 1) - 1) +
- e_bln_c_u_view(jc, 3, jb) *
- p_vt_in_view(cell_edge_idx_view(jc, jb, 1) - 1, jk, cell_edge_blk_view(jc, jb, 1) - 1) +
- e_bln_c_u_view(jc, 4, jb) *
- p_vn_in_view(cell_edge_idx_view(jc, jb, 2) - 1, jk, cell_edge_blk_view(jc, jb, 2) - 1) +
- e_bln_c_u_view(jc, 5, jb) *
- p_vt_in_view(cell_edge_idx_view(jc, jb, 2) - 1, jk, cell_edge_blk_view(jc, jb, 2) - 1);
+ // Compute the bilinear interpolation for cell (jc, jk, jb).
+ p_u_out_view(jc, jk, jb) =
+ e_bln_c_u_view(jc, 0, jb) *
+ p_vn_in_view(cell_edge_idx_view(jc, jb, 0) - 1, jk,
+ cell_edge_blk_view(jc, jb, 0) - 1) +
+ e_bln_c_u_view(jc, 1, jb) *
+ p_vt_in_view(cell_edge_idx_view(jc, jb, 0) - 1, jk,
+ cell_edge_blk_view(jc, jb, 0) - 1) +
+ e_bln_c_u_view(jc, 2, jb) *
+ p_vn_in_view(cell_edge_idx_view(jc, jb, 1) - 1, jk,
+ cell_edge_blk_view(jc, jb, 1) - 1) +
+ e_bln_c_u_view(jc, 3, jb) *
+ p_vt_in_view(cell_edge_idx_view(jc, jb, 1) - 1, jk,
+ cell_edge_blk_view(jc, jb, 1) - 1) +
+ e_bln_c_u_view(jc, 4, jb) *
+ p_vn_in_view(cell_edge_idx_view(jc, jb, 2) - 1, jk,
+ cell_edge_blk_view(jc, jb, 2) - 1) +
+ e_bln_c_u_view(jc, 5, jb) *
+ p_vt_in_view(cell_edge_idx_view(jc, jb, 2) - 1, jk,
+ cell_edge_blk_view(jc, jb, 2) - 1);
- p_v_out_view(jc, jk, jb) =
- e_bln_c_v_view(jc, 0, jb) *
- p_vn_in_view(cell_edge_idx_view(jc, jb, 0) - 1, jk, cell_edge_blk_view(jc, jb, 0) - 1) +
- e_bln_c_v_view(jc, 1, jb) *
- p_vt_in_view(cell_edge_idx_view(jc, jb, 0) - 1, jk, cell_edge_blk_view(jc, jb, 0) - 1) +
- e_bln_c_v_view(jc, 2, jb) *
- p_vn_in_view(cell_edge_idx_view(jc, jb, 1) - 1, jk, cell_edge_blk_view(jc, jb, 1) - 1) +
- e_bln_c_v_view(jc, 3, jb) *
- p_vt_in_view(cell_edge_idx_view(jc, jb, 1) - 1, jk, cell_edge_blk_view(jc, jb, 1) - 1) +
- e_bln_c_v_view(jc, 4, jb) *
- p_vn_in_view(cell_edge_idx_view(jc, jb, 2) - 1, jk, cell_edge_blk_view(jc, jb, 2) - 1) +
- e_bln_c_v_view(jc, 5, jb) *
- p_vt_in_view(cell_edge_idx_view(jc, jb, 2) - 1, jk, cell_edge_blk_view(jc, jb, 2) - 1);
+ p_v_out_view(jc, jk, jb) =
+ e_bln_c_v_view(jc, 0, jb) *
+ p_vn_in_view(cell_edge_idx_view(jc, jb, 0) - 1, jk,
+ cell_edge_blk_view(jc, jb, 0) - 1) +
+ e_bln_c_v_view(jc, 1, jb) *
+ p_vt_in_view(cell_edge_idx_view(jc, jb, 0) - 1, jk,
+ cell_edge_blk_view(jc, jb, 0) - 1) +
+ e_bln_c_v_view(jc, 2, jb) *
+ p_vn_in_view(cell_edge_idx_view(jc, jb, 1) - 1, jk,
+ cell_edge_blk_view(jc, jb, 1) - 1) +
+ e_bln_c_v_view(jc, 3, jb) *
+ p_vt_in_view(cell_edge_idx_view(jc, jb, 1) - 1, jk,
+ cell_edge_blk_view(jc, jb, 1) - 1) +
+ e_bln_c_v_view(jc, 4, jb) *
+ p_vn_in_view(cell_edge_idx_view(jc, jb, 2) - 1, jk,
+ cell_edge_blk_view(jc, jb, 2) - 1) +
+ e_bln_c_v_view(jc, 5, jb) *
+ p_vt_in_view(cell_edge_idx_view(jc, jb, 2) - 1, jk,
+ cell_edge_blk_view(jc, jb, 2) - 1);
});
- // Optionally fence after each block if required.
- Kokkos::fence();
- }
+ // Optionally fence after each block if required.
+ Kokkos::fence();
+ }
}
-template
-void edges2cells_vector_lib<double>(
- const double* p_vn_in, const double* p_vt_in,
- const int* cell_edge_idx, const int* cell_edge_blk,
- const double* e_bln_c_u, const double* e_bln_c_v,
- double* p_u_out, double* p_v_out,
+template void edges2cells_vector_lib<double>(
+ const double *p_vn_in, const double *p_vt_in, const int *cell_edge_idx,
+ const int *cell_edge_blk, const double *e_bln_c_u, const double *e_bln_c_v,
+ double *p_u_out, double *p_v_out,
// Additional integer parameters.
- int i_startblk, int i_endblk,
- int i_startidx_in, int i_endidx_in,
- int slev, int elev,
- int nproma,
+ int i_startblk, int i_endblk, int i_startidx_in, int i_endidx_in, int slev,
+ int elev, int nproma,
// Dimensions for the arrays.
int nlev, int nblks_e, int nblks_c);
-template
-void edges2cells_vector_lib<float>(
- const float* p_vn_in, const float* p_vt_in,
- const int* cell_edge_idx, const int* cell_edge_blk,
- const float* e_bln_c_u, const float* e_bln_c_v,
- float* p_u_out, float* p_v_out,
+template void edges2cells_vector_lib<float>(
+ const float *p_vn_in, const float *p_vt_in, const int *cell_edge_idx,
+ const int *cell_edge_blk, const float *e_bln_c_u, const float *e_bln_c_v,
+ float *p_u_out, float *p_v_out,
// Additional integer parameters.
- int i_startblk, int i_endblk,
- int i_startidx_in, int i_endidx_in,
- int slev, int elev,
- int nproma,
+ int i_startblk, int i_endblk, int i_startidx_in, int i_endidx_in, int slev,
+ int elev, int nproma,
// Dimensions for the arrays.
int nlev, int nblks_e, int nblks_c);
\ No newline at end of file
diff --git a/src/interpolation/mo_lib_interpolation_vector.hpp b/src/interpolation/mo_lib_interpolation_vector.hpp
index 1ee6c5b..9186997 100644
--- a/src/interpolation/mo_lib_interpolation_vector.hpp
+++ b/src/interpolation/mo_lib_interpolation_vector.hpp
@@ -22,15 +22,12 @@
// - e_bln_c_u and e_bln_c_v: dimensions [nproma, 6, nblks_c]
// - p_u_out and p_v_out: dimensions [nproma, nlev, nblks_c]
template <typename T>
-void edges2cells_vector_lib(
- const T* p_vn_in, const T* p_vt_in,
- const int* cell_edge_idx, const int* cell_edge_blk,
- const T* e_bln_c_u, const T* e_bln_c_v,
- T* p_u_out, T* p_v_out,
- // Additional integer parameters.
- int i_startblk, int i_endblk,
- int i_startidx_in, int i_endidx_in,
- int slev, int elev,
- int nproma,
- // Dimensions for the arrays.
- int nlev, int nblks_e, int nblks_c);
\ No newline at end of file
+void edges2cells_vector_lib(const T *p_vn_in, const T *p_vt_in,
+ const int *cell_edge_idx, const int *cell_edge_blk,
+ const T *e_bln_c_u, const T *e_bln_c_v, T *p_u_out,
+ T *p_v_out,
+ // Additional integer parameters.
+ int i_startblk, int i_endblk, int i_startidx_in,
+ int i_endidx_in, int slev, int elev, int nproma,
+ // Dimensions for the arrays.
+ int nlev, int nblks_e, int nblks_c);
\ No newline at end of file
diff --git a/test/c/test_interpolation_scalar.cpp b/test/c/test_interpolation_scalar.cpp
index 606d1e9..0ee7fa3 100644
--- a/test/c/test_interpolation_scalar.cpp
+++ b/test/c/test_interpolation_scalar.cpp
@@ -9,145 +9,146 @@
// SPDX-License-Identifier: BSD-3-Clause
// ---------------------------------------------------------------
-#include <gtest/gtest.h>
+#include "mo_lib_interpolation_scalar.hpp"
#include <Kokkos_Core.hpp>
+#include <gtest/gtest.h>
#include <vector>
-#include "mo_lib_interpolation_scalar.hpp"
// Free-function helpers for 3D and 4D array sizes (assumed column-major)
-template<typename T>
-size_t num_elements_3d(int d1, int d2, int d3) {
+template <typename T> size_t num_elements_3d(int d1, int d2, int d3) {
return static_cast<size_t>(d1) * d2 * d3;
}
-template<typename T>
-size_t num_elements_4d(int d1, int d2, int d3, int d4) {
+template <typename T> size_t num_elements_4d(int d1, int d2, int d3, int d4) {
return static_cast<size_t>(d1) * d2 * d3 * d4;
}
// Define a helper struct that holds the two types.
-template<typename InT, typename OutT>
-struct MixedPrecision {
- using in_type = InT;
+template <typename InT, typename OutT> struct MixedPrecision {
+ using in_type = InT;
using out_type = OutT;
};
// Define the list of type pairs we want to test.
-typedef ::testing::Types< MixedPrecision<double, double>,
- MixedPrecision<double, float>,
- MixedPrecision<float, float> > MixedTypes;
-
-typedef ::testing::Types< MixedPrecision<double, double>,
- MixedPrecision<float, double>,
- MixedPrecision<float, float> > MixedTypesSP2DP;
+typedef ::testing::Types<MixedPrecision<double, double>,
+ MixedPrecision<double, float>,
+ MixedPrecision<float, float>>
+ MixedTypes;
+typedef ::testing::Types<MixedPrecision<double, double>,
+ MixedPrecision<float, double>,
+ MixedPrecision<float, float>>
+ MixedTypesSP2DP;
// Shared dimensions for all routines and classes
class interp_dimensions {
public:
// Constant dimensions.
- static constexpr int nproma = 16; // inner loop length
- static constexpr int nlev = 7; // number of vertical levels
- static constexpr int nblks_c = 2; // number of cell blocks
- static constexpr int nblks_e = 2; // number of edge blocks (for p_e_in)
- static constexpr int nblks_v = 2; // number of vertex blocks (for rbf arrays and outputs)
+ static constexpr int nproma = 16; // inner loop length
+ static constexpr int nlev = 7; // number of vertical levels
+ static constexpr int nblks_c = 2; // number of cell blocks
+ static constexpr int nblks_e = 2; // number of edge blocks (for p_e_in)
+ static constexpr int nblks_v =
+ 2; // number of vertex blocks (for rbf arrays and outputs)
// Parameter values.
- const int i_startblk = 0;
- const int i_endblk = 1; // Test blocks [0, 1]
- const int i_startidx = 2;
- const int i_endidx = nproma - 3; // Partial range: 2 .. nproma-3
- const int slev = 1;
- const int elev = nlev - 1; // Partial vertical range (1 .. nlev-1)
- const bool lacc = false; // Not using ACC-specific behavior.
- const bool acc_async = false; // No asynchronous execution.
+ const int i_startblk = 0;
+ const int i_endblk = 1; // Test blocks [0, 1]
+ const int i_startidx = 2;
+ const int i_endidx = nproma - 3; // Partial range: 2 .. nproma-3
+ const int slev = 1;
+ const int elev = nlev - 1; // Partial vertical range (1 .. nlev-1)
+ const bool lacc = false; // Not using ACC-specific behavior.
+ const bool acc_async = false; // No asynchronous execution.
};
template <typename T>
-class InterpolationScalarTypedTestFixture : public ::testing::Test, public interp_dimensions {
+class InterpolationScalarTypedTestFixture : public ::testing::Test,
+ public interp_dimensions {
public:
-
// Arrays used for verts2edges
- std::vector<T> p_vertex_in; // Dimensions: (nproma, nlev, nblks_v)
- std::vector<int> edge_vertex_idx; // Dimensions: (nproma, nblks_e, 4)
- std::vector<int> edge_vertex_blk; // Dimensions: (nproma, nblks_e, 4)
- std::vector<T> coeff_int_edges; // Dimensions: (nproma, 2, nblks_e)
- std::vector<T> p_edge_out; // Dimensions: (nproma, nlev, nblks_e)
+ std::vector<T> p_vertex_in; // Dimensions: (nproma, nlev, nblks_v)
+ std::vector<int> edge_vertex_idx; // Dimensions: (nproma, nblks_e, 4)
+ std::vector<int> edge_vertex_blk; // Dimensions: (nproma, nblks_e, 4)
+ std::vector<T> coeff_int_edges; // Dimensions: (nproma, 2, nblks_e)
+ std::vector<T> p_edge_out; // Dimensions: (nproma, nlev, nblks_e)
// Arrays used for edges2verts
- std::vector<T> p_edge_in; // Dimensions: (nproma, nlev, nblks_e)
- std::vector<int> edge_vert_idx; // Dimensions: (nproma, nblks_e, 6)
- std::vector<int> edge_vert_blk; // Dimensions: (nproma, nblks_e, 6)
- std::vector<T> v_int; // Dimensions: (nproma, 6, nblks_v)
- std::vector<T> p_vert_out; // Dimensions: (nproma, nlev, nblks_v)
+ std::vector<T> p_edge_in; // Dimensions: (nproma, nlev, nblks_e)
+ std::vector<int> edge_vert_idx; // Dimensions: (nproma, nblks_e, 6)
+ std::vector<int> edge_vert_blk; // Dimensions: (nproma, nblks_e, 6)
+ std::vector<T> v_int; // Dimensions: (nproma, 6, nblks_v)
+ std::vector<T> p_vert_out; // Dimensions: (nproma, nlev, nblks_v)
// Arrays used for edges2cells
// std::vector<T> p_edge_in; // Dimensions: (nproma, nlev, nblks_e)
- std::vector<int> edge_idx; // Dimensions: (nproma, nblks_c, 3)
- std::vector<int> edge_blk; // Dimensions: (nproma, nblks_c, 3)
- std::vector<T> coeff_int_cells; // Dimensions: (nproma, 3, nblks_c)
- std::vector<T> p_cell_out; // Dimensions: (nproma, nlev, nblks_c)
+ std::vector<int> edge_idx; // Dimensions: (nproma, nblks_c, 3)
+ std::vector<int> edge_blk; // Dimensions: (nproma, nblks_c, 3)
+ std::vector<T> coeff_int_cells; // Dimensions: (nproma, 3, nblks_c)
+ std::vector<T> p_cell_out; // Dimensions: (nproma, nlev, nblks_c)
// Arrays used for verts2cells
- std::vector<T> p_vert_in; // Dimensions: (nproma, nlev, nblks_v)
- std::vector<int> cell_index_idx; // Dimensions: (nproma, nblks_c, 3)
- std::vector<int> cell_index_blk; // Dimensions: (nproma, nblks_c, 3)
- // std::vector<InType> coeff_int; // Dimensions: (nproma, 3, nblks_c)
- // std::vector<T> p_cell_out; // Dimensions: (nproma, nlev, nblks_c)
+ std::vector<T> p_vert_in; // Dimensions: (nproma, nlev, nblks_v)
+ std::vector<int> cell_index_idx; // Dimensions: (nproma, nblks_c, 3)
+ std::vector<int> cell_index_blk; // Dimensions: (nproma, nblks_c, 3)
// Arrays used for avg_lib
- std::vector<T> psi_c; // Dimensions: (nproma, nlev, nblks_c)
- std::vector<int> cell_neighbor_idx; // Dimensions: (nproma, nblks_c, 3)
- std::vector<int> cell_neighbor_blk; // Dimensions: (nproma, nblks_c, 3)
+ std::vector<T> psi_c; // Dimensions: (nproma, nlev, nblks_c)
+ std::vector<int> cell_neighbor_idx; // Dimensions: (nproma, nblks_c, 3)
+ std::vector<int> cell_neighbor_blk; // Dimensions: (nproma, nblks_c, 3)
std::vector<T> avg_coeff; // Dimensions: (nproma, nlev, nblks_c)
- std::vector<T> avg_psi_c; // Dimensions: (nproma, nlev, nblks_c)
+ std::vector<T> avg_psi_c; // Dimensions: (nproma, nlev, nblks_c)
const int cell_type = 6;
const int npromz_c = 32;
InterpolationScalarTypedTestFixture() {
// Allocate and initialize arrays needed for verts2edges
- p_vertex_in.resize(num_elements_3d<T>(nproma, nlev, nblks_v), static_cast<T>(1));
+ p_vertex_in.resize(num_elements_3d<T>(nproma, nlev, nblks_v),
+ static_cast<T>(1));
edge_vertex_idx.resize(num_elements_3d<int>(nproma, nblks_e, 4), 1);
edge_vertex_blk.resize(num_elements_3d<int>(nproma, nblks_e, 4), 0);
- coeff_int_edges.resize(num_elements_3d<T>(nproma, 2, nblks_e), static_cast<T>(1));
+ coeff_int_edges.resize(num_elements_3d<T>(nproma, 2, nblks_e),
+ static_cast<T>(1));
- p_edge_out.resize(num_elements_3d<T>(nproma, nlev, nblks_e), static_cast<T>(0));
+ p_edge_out.resize(num_elements_3d<T>(nproma, nlev, nblks_e),
+ static_cast<T>(0));
// Allocate & Initialize arrays needed for edges2verts
- p_edge_in.resize(num_elements_3d<T>(nproma, nlev, nblks_e), static_cast<T>(1));
+ p_edge_in.resize(num_elements_3d<T>(nproma, nlev, nblks_e),
+ static_cast<T>(1));
edge_vert_idx.resize(num_elements_3d<int>(nproma, nblks_e, 6), 1);
edge_vert_blk.resize(num_elements_3d<int>(nproma, nblks_e, 6), 0);
v_int.resize(num_elements_3d<T>(nproma, 6, nblks_v), static_cast<T>(1));
- p_vert_out.resize(num_elements_3d<T>(nproma, nlev, nblks_v), static_cast<T>(0));
+ p_vert_out.resize(num_elements_3d<T>(nproma, nlev, nblks_v),
+ static_cast<T>(0));
// Allocate & Initialize arrays needed for edges2cells
- // p_edge_in.resize(num_elements_3d<T>(nproma, nlev, nblks_e), static_cast<T>(1));
edge_idx.resize(num_elements_3d<int>(nproma, nblks_c, 3), 1);
edge_blk.resize(num_elements_3d<int>(nproma, nblks_c, 3), 0);
- coeff_int_cells.resize(num_elements_3d<T>(nproma, 3, nblks_c), static_cast<T>(1));
+ coeff_int_cells.resize(num_elements_3d<T>(nproma, 3, nblks_c),
+ static_cast<T>(1));
- p_cell_out.resize(num_elements_3d<T>(nproma, nlev, nblks_c), static_cast<T>(0));
+ p_cell_out.resize(num_elements_3d<T>(nproma, nlev, nblks_c),
+ static_cast<T>(0));
// Allocate and initialize arrays needed for verts2cells
- p_vert_in.resize(num_elements_3d<T>(nproma, nlev, nblks_v), static_cast<T>(1));
+ p_vert_in.resize(num_elements_3d<T>(nproma, nlev, nblks_v),
+ static_cast<T>(1));
cell_index_idx.resize(num_elements_3d<int>(nproma, nblks_c, 3), 1);
cell_index_blk.resize(num_elements_3d<int>(nproma, nblks_c, 3), 0);
- // coeff_int.resize(num_elements_3d<T>(nproma, cell_type, nblks_c), static_cast<T>(1));
-
- // Allocate output arrays and initialize to zero.
- // p_cell_out.resize(num_elements_3d<T>(nproma, nlev, nblks_c), static_cast<T>(0));
// Allocate and initialize arrays needed for avg_lib
psi_c.resize(num_elements_3d<T>(nproma, nlev, nblks_c), static_cast<T>(1));
cell_neighbor_idx.resize(num_elements_3d<int>(nproma, nblks_c, 3), 1);
cell_neighbor_blk.resize(num_elements_3d<int>(nproma, nblks_c, 3), 0);
- avg_coeff.resize(num_elements_3d<T>(nproma, nlev, nblks_c), static_cast<T>(1));
+ avg_coeff.resize(num_elements_3d<T>(nproma, nlev, nblks_c),
+ static_cast<T>(1));
// Allocate output arrays and initialize to zero.
- avg_psi_c.resize(num_elements_3d<T>(nproma, nlev, nblks_c), static_cast<T>(0));
-
+ avg_psi_c.resize(num_elements_3d<T>(nproma, nlev, nblks_c),
+ static_cast<T>(0));
}
};
@@ -164,22 +165,11 @@ TYPED_TEST_SUITE(InterpolationScalarTypedTestFixture, SingleType);
TYPED_TEST(InterpolationScalarTypedTestFixture, Verts2Edges) {
verts2edges_scalar_lib<TypeParam>(
- this->p_vertex_in.data(),
- this->edge_vertex_idx.data(),
- this->edge_vertex_blk.data(),
- this->coeff_int_edges.data(),
- this->p_edge_out.data(),
- this->i_startblk,
- this->i_endblk,
- this->i_startidx,
- this->i_endidx,
- this->slev,
- this->elev,
- this->nproma,
- this->nlev,
- this->nblks_v,
- this->nblks_e,
- this->lacc);
+ this->p_vertex_in.data(), this->edge_vertex_idx.data(),
+ this->edge_vertex_blk.data(), this->coeff_int_edges.data(),
+ this->p_edge_out.data(), this->i_startblk, this->i_endblk,
+ this->i_startidx, this->i_endidx, this->slev, this->elev, this->nproma,
+ this->nlev, this->nblks_v, this->nblks_e, this->lacc);
// Check the outputs only for blocks in the range
// { [i_startblk, i_endblk], [slev,elev], [i_startidx, i_endidx] }
@@ -187,10 +177,14 @@ TYPED_TEST(InterpolationScalarTypedTestFixture, Verts2Edges) {
for (int level = this->slev; level < this->elev; ++level) {
for (int i = this->i_startidx; i < this->i_endidx; ++i) {
// Compute the linear index for a 3D array in column-major order:
- size_t idx = i + level * this->nproma + block * this->nproma * this->nlev;
- // Since every contribution is 1 and there are 2 stencil points, expect 2.
- EXPECT_NEAR(this->p_edge_out[idx], static_cast<TypeParam>(2), static_cast<TypeParam>(1e-5))
- << "Failure at block " << block << ", level " << level << ", index " << i;
+ size_t idx =
+ i + level * this->nproma + block * this->nproma * this->nlev;
+ // Since every contribution is 1 and there are 2 stencil points,
+ // expect 2.
+ EXPECT_NEAR(this->p_edge_out[idx], static_cast<TypeParam>(2),
+ static_cast<TypeParam>(1e-5))
+ << "Failure at block " << block << ", level " << level << ", index "
+ << i;
}
}
}
@@ -205,22 +199,11 @@ TYPED_TEST(InterpolationScalarTypedTestFixture, Verts2Edges) {
TYPED_TEST(InterpolationScalarTypedTestFixture, Edges2Verts) {
edges2verts_scalar_lib<TypeParam>(
- this->p_edge_in.data(),
- this->edge_vert_idx.data(),
- this->edge_vert_blk.data(),
- this->v_int.data(),
- this->p_vert_out.data(),
- this->i_startblk,
- this->i_endblk,
- this->i_startidx,
- this->i_endidx,
- this->slev,
- this->elev,
- this->nproma,
- this->nlev,
- this->nblks_e,
- this->nblks_v,
- this->lacc);
+ this->p_edge_in.data(), this->edge_vert_idx.data(),
+ this->edge_vert_blk.data(), this->v_int.data(), this->p_vert_out.data(),
+ this->i_startblk, this->i_endblk, this->i_startidx, this->i_endidx,
+ this->slev, this->elev, this->nproma, this->nlev, this->nblks_e,
+ this->nblks_v, this->lacc);
// Check the outputs only for blocks in the range
// { [i_startblk, i_endblk], [slev,elev], [i_startidx_in, i_endidx_in] }
@@ -228,10 +211,14 @@ TYPED_TEST(InterpolationScalarTypedTestFixture, Edges2Verts) {
for (int level = this->slev; level < this->elev; ++level) {
for (int i = this->i_startidx; i < this->i_endidx; ++i) {
// Compute the linear index for a 3D array in column-major order:
- size_t idx = i + level * this->nproma + block * this->nproma * this->nlev;
- // Since every contribution is 1 and there are 6 stencil points, expect 6.
- EXPECT_NEAR(this->p_vert_out[idx], static_cast<TypeParam>(6), static_cast<TypeParam>(1e-5))
- << "Failure at block " << block << ", level " << level << ", index " << i;
+ size_t idx =
+ i + level * this->nproma + block * this->nproma * this->nlev;
+ // Since every contribution is 1 and there are 6 stencil points,
+ // expect 6.
+ EXPECT_NEAR(this->p_vert_out[idx], static_cast<TypeParam>(6),
+ static_cast<TypeParam>(1e-5))
+ << "Failure at block " << block << ", level " << level << ", index "
+ << i;
}
}
}
@@ -246,22 +233,10 @@ TYPED_TEST(InterpolationScalarTypedTestFixture, Edges2Verts) {
TYPED_TEST(InterpolationScalarTypedTestFixture, Edges2Cells) {
edges2cells_scalar_lib<TypeParam>(
- this->p_edge_in.data(),
- this->edge_idx.data(),
- this->edge_blk.data(),
- this->coeff_int_cells.data(),
- this->p_cell_out.data(),
- this->i_startblk,
- this->i_endblk,
- this->i_startidx,
- this->i_endidx,
- this->slev,
- this->elev,
- this->nproma,
- this->nlev,
- this->nblks_e,
- this->nblks_c,
- this->lacc);
+ this->p_edge_in.data(), this->edge_idx.data(), this->edge_blk.data(),
+ this->coeff_int_cells.data(), this->p_cell_out.data(), this->i_startblk,
+ this->i_endblk, this->i_startidx, this->i_endidx, this->slev, this->elev,
+ this->nproma, this->nlev, this->nblks_e, this->nblks_c, this->lacc);
// Check the outputs only for blocks in the range
// { [i_startblk, i_endblk], [slev,elev], [i_startidx_in, i_endidx_in] }
@@ -269,10 +244,14 @@ TYPED_TEST(InterpolationScalarTypedTestFixture, Edges2Cells) {
for (int level = this->slev; level < this->elev; ++level) {
for (int i = this->i_startidx; i < this->i_endidx; ++i) {
// Compute the linear index for a 3D array in column-major order:
- size_t idx = i + level * this->nproma + block * this->nproma * this->nlev;
- // Since every contribution is 1 and there are 3 stencil points, expect 3.
- EXPECT_NEAR(this->p_cell_out[idx], static_cast<TypeParam>(3), static_cast<TypeParam>(1e-5))
- << "Failure at block " << block << ", level " << level << ", index " << i;
+ size_t idx =
+ i + level * this->nproma + block * this->nproma * this->nlev;
+ // Since every contribution is 1 and there are 3 stencil points,
+ // expect 3.
+ EXPECT_NEAR(this->p_cell_out[idx], static_cast<TypeParam>(3),
+ static_cast<TypeParam>(1e-5))
+ << "Failure at block " << block << ", level " << level << ", index "
+ << i;
}
}
}
@@ -281,20 +260,10 @@ TYPED_TEST(InterpolationScalarTypedTestFixture, Edges2Cells) {
TYPED_TEST(InterpolationScalarTypedTestFixture, Verts2Cells) {
verts2cells_scalar_lib<TypeParam>(
- this->p_vert_in.data(),
- this->cell_index_idx.data(),
- this->cell_index_blk.data(),
- this->coeff_int_cells.data(),
- this->p_cell_out.data(),
- this->nblks_c,
- this->npromz_c,
- this->slev,
- this->elev,
- this->nproma,
- this->nlev,
- this->nblks_v,
- this->lacc);
-
+ this->p_vert_in.data(), this->cell_index_idx.data(),
+ this->cell_index_blk.data(), this->coeff_int_cells.data(),
+ this->p_cell_out.data(), this->nblks_c, this->npromz_c, this->slev,
+ this->elev, this->nproma, this->nlev, this->nblks_v, this->lacc);
// Check the outputs only for blocks in the range
// { [i_startblk, i_endblk], [slev,elev], [i_startidx_in, i_endidx_in] }
@@ -302,10 +271,14 @@ TYPED_TEST(InterpolationScalarTypedTestFixture, Verts2Cells) {
for (int level = this->slev; level < this->elev; ++level) {
for (int i = this->i_startidx; i < this->i_endidx; ++i) {
// Compute the linear index for a 3D array in column-major order:
- size_t idx = i + level * this->nproma + block * this->nproma * this->nlev;
- // Since every contribution is 1 and there are 3 stencil points, expect 3.
- EXPECT_NEAR(this->p_cell_out[idx], static_cast<TypeParam>(3), static_cast<TypeParam>(1e-5))
- << "Failure at block " << block << ", level " << level << ", index " << i;
+ size_t idx =
+ i + level * this->nproma + block * this->nproma * this->nlev;
+ // Since every contribution is 1 and there are 3 stencil points,
+ // expect 3.
+ EXPECT_NEAR(this->p_cell_out[idx], static_cast<TypeParam>(3),
+ static_cast<TypeParam>(1e-5))
+ << "Failure at block " << block << ", level " << level << ", index "
+ << i;
}
}
}
@@ -320,23 +293,12 @@ TYPED_TEST(InterpolationScalarTypedTestFixture, Verts2Cells) {
TYPED_TEST(InterpolationScalarTypedTestFixture, AvgLib) {
// Call the function
- cell_avg_lib<TypeParam>(
- this->psi_c.data(),
- this->cell_neighbor_idx.data(),
- this->cell_neighbor_blk.data(),
- this->avg_coeff.data(),
- this->avg_psi_c.data(),
- this->i_startblk,
- this->i_endblk,
- this->i_startidx,
- this->i_endidx,
- this->slev,
- this->elev,
- this->nproma,
- this->nlev,
- this->nblks_c,
- this->lacc);
-
+ cell_avg_lib<TypeParam>(this->psi_c.data(), this->cell_neighbor_idx.data(),
+ this->cell_neighbor_blk.data(),
+ this->avg_coeff.data(), this->avg_psi_c.data(),
+ this->i_startblk, this->i_endblk, this->i_startidx,
+ this->i_endidx, this->slev, this->elev, this->nproma,
+ this->nlev, this->nblks_c, this->lacc);
// Check the outputs only for blocks in the range
// { [i_startblk, i_endblk], [slev,elev], [i_startidx_in, i_endidx_in] }
@@ -344,27 +306,32 @@ TYPED_TEST(InterpolationScalarTypedTestFixture, AvgLib) {
for (int level = this->slev; level < this->elev; ++level) {
for (int i = this->i_startidx; i < this->i_endidx; ++i) {
// Compute the linear index for a 3D array in column-major order:
- size_t idx = i + level * this->nproma + block * this->nproma * this->nlev;
- // Since every contribution is 1 and there are 4 stencil points, expect 4.
- EXPECT_NEAR(this->avg_psi_c[idx], static_cast<TypeParam>(4), static_cast<TypeParam>(1e-5))
- << "Failure at block " << block << ", level " << level << ", index " << i;
+ size_t idx =
+ i + level * this->nproma + block * this->nproma * this->nlev;
+ // Since every contribution is 1 and there are 4 stencil points,
+ // expect 4.
+ EXPECT_NEAR(this->avg_psi_c[idx], static_cast<TypeParam>(4),
+ static_cast<TypeParam>(1e-5))
+ << "Failure at block " << block << ", level " << level << ", index "
+ << i;
}
}
}
}
template <typename TypePair>
-class InterpolationScalarMixedTestFixture : public ::testing::Test, public interp_dimensions {
+class InterpolationScalarMixedTestFixture : public ::testing::Test,
+ public interp_dimensions {
public:
- using InType = typename TypePair::in_type;
+ using InType = typename TypePair::in_type;
using OutType = typename TypePair::out_type;
// Arrays used for cells2edges
- std::vector<InType> p_cell_in; // Dimensions: (nproma, nlev, nblks_c)
- std::vector<int> edge_cell_idx; // Dimensions: (nproma, nblks_e, 2)
- std::vector<int> edge_cell_blk; // Dimensions: (nproma, nblks_e, 2)
- std::vector<OutType> coeff_int_edges; // Dimensions: (nproma, 2, nblks_e)
- std::vector<OutType> p_edge_out; // Dimensions: (nproma, nlev, nblks_e)
+ std::vector<InType> p_cell_in; // Dimensions: (nproma, nlev, nblks_c)
+ std::vector<int> edge_cell_idx; // Dimensions: (nproma, nblks_e, 2)
+ std::vector<int> edge_cell_blk; // Dimensions: (nproma, nblks_e, 2)
+ std::vector<OutType> coeff_int_edges; // Dimensions: (nproma, 2, nblks_e)
+ std::vector<OutType> p_edge_out; // Dimensions: (nproma, nlev, nblks_e)
// Further parameters for cells2edges
const int patch_id = 0;
@@ -376,19 +343,22 @@ public:
std::vector<int> i_endidx_in; // Dimensions: (2)
// Arrays used for cells2verts
- std::vector<int> vert_cell_idx; // Dimensions: (nproma, nblks_v, 6)
- std::vector<int> vert_cell_blk; // Dimensions: (nproma, nblks_v, 6)
- std::vector<OutType> coeff_int_verts; // Dimensions: (nproma, 6, nblks_v)
- std::vector<OutType> p_vert_out; // Dimensions: (nproma, nlev, nblks_v)
+ std::vector<int> vert_cell_idx; // Dimensions: (nproma, nblks_v, 6)
+ std::vector<int> vert_cell_blk; // Dimensions: (nproma, nblks_v, 6)
+ std::vector<OutType> coeff_int_verts; // Dimensions: (nproma, 6, nblks_v)
+ std::vector<OutType> p_vert_out; // Dimensions: (nproma, nlev, nblks_v)
InterpolationScalarMixedTestFixture() {
// Allocate and initialize arrays needed for cells2edges
- p_cell_in.resize(num_elements_3d<InType>(nproma, nlev, nblks_c), static_cast<InType>(1));
+ p_cell_in.resize(num_elements_3d<InType>(nproma, nlev, nblks_c),
+ static_cast<InType>(1));
edge_cell_idx.resize(num_elements_3d<int>(nproma, nblks_e, 2), 1);
edge_cell_blk.resize(num_elements_3d<int>(nproma, nblks_e, 2), 0);
- coeff_int_edges.resize(num_elements_3d<InType>(nproma, 2, nblks_e), static_cast<OutType>(1));
+ coeff_int_edges.resize(num_elements_3d<InType>(nproma, 2, nblks_e),
+ static_cast<OutType>(1));
- p_edge_out.resize(num_elements_3d<OutType>(nproma, nlev, nblks_e), static_cast<OutType>(0));
+ p_edge_out.resize(num_elements_3d<OutType>(nproma, nlev, nblks_e),
+ static_cast<OutType>(0));
// Allocate neighbour indexes for cells2edges
i_startblk_in.resize(2, i_startblk);
@@ -399,9 +369,11 @@ public:
// Allocate & Initialize arrays needed for cells2verts
vert_cell_idx.resize(num_elements_3d<int>(nproma, nblks_v, 6), 1);
vert_cell_blk.resize(num_elements_3d<int>(nproma, nblks_v, 6), 0);
- coeff_int_verts.resize(num_elements_3d<InType>(nproma, 6, nblks_v), static_cast<OutType>(1));
+ coeff_int_verts.resize(num_elements_3d<InType>(nproma, 6, nblks_v),
+ static_cast<OutType>(1));
- p_vert_out.resize(num_elements_3d<OutType>(nproma, nlev, nblks_v), static_cast<OutType>(0));
+ p_vert_out.resize(num_elements_3d<OutType>(nproma, nlev, nblks_v),
+ static_cast<OutType>(0));
}
};
@@ -414,30 +386,18 @@ TYPED_TEST_SUITE(InterpolationScalarMixedTestFixture, MixedTypesSP2DP);
////////////////////////////////////////////////////////////////////////////////
TYPED_TEST(InterpolationScalarMixedTestFixture, cells2edges) {
- using InType = typename TestFixture::InType;
+ using InType = typename TestFixture::InType;
using OutType = typename TestFixture::OutType;
// Call the function
- cells2edges_scalar_lib<InType,OutType>(
- this->p_cell_in.data(),
- this->edge_cell_idx.data(),
- this->edge_cell_blk.data(),
- this->coeff_int_edges.data(),
- this->p_edge_out.data(),
- this->i_startblk_in.data(),
- this->i_endblk_in.data(),
- this->i_startidx_in.data(),
- this->i_endidx_in.data(),
- this->slev,
- this->elev,
- this->nproma,
- this->nlev,
- this->nblks_c,
- this->nblks_e,
- this->patch_id,
- this->l_limited_area,
- this->lfill_latbc,
- this->lacc);
+ cells2edges_scalar_lib<InType, OutType>(
+ this->p_cell_in.data(), this->edge_cell_idx.data(),
+ this->edge_cell_blk.data(), this->coeff_int_edges.data(),
+ this->p_edge_out.data(), this->i_startblk_in.data(),
+ this->i_endblk_in.data(), this->i_startidx_in.data(),
+ this->i_endidx_in.data(), this->slev, this->elev, this->nproma,
+ this->nlev, this->nblks_c, this->nblks_e, this->patch_id,
+ this->l_limited_area, this->lfill_latbc, this->lacc);
// Check the outputs only for blocks in the range
// { [i_startblk, i_endblk], [slev,elev], [i_startidx_in, i_endidx_in] }
@@ -445,10 +405,14 @@ TYPED_TEST(InterpolationScalarMixedTestFixture, cells2edges) {
for (int level = this->slev; level < this->elev; ++level) {
for (int i = this->i_startidx; i < this->i_endidx; ++i) {
// Compute the linear index for a 3D array in column-major order:
- size_t idx = i + level * this->nproma + block * this->nproma * this->nlev;
- // Since every contribution is 1 and there are 2 stencil points, expect 2.
- EXPECT_NEAR(this->p_edge_out[idx], static_cast<OutType>(2), static_cast<OutType>(1e-5))
- << "Failure at block " << block << ", level " << level << ", index " << i;
+ size_t idx =
+ i + level * this->nproma + block * this->nproma * this->nlev;
+ // Since every contribution is 1 and there are 2 stencil points,
+ // expect 2.
+ EXPECT_NEAR(this->p_edge_out[idx], static_cast<OutType>(2),
+ static_cast<OutType>(1e-5))
+ << "Failure at block " << block << ", level " << level << ", index "
+ << i;
}
}
}
@@ -461,27 +425,15 @@ TYPED_TEST(InterpolationScalarMixedTestFixture, cells2edges) {
////////////////////////////////////////////////////////////////////////////////
TYPED_TEST(InterpolationScalarMixedTestFixture, cells2verts) {
- using InType = typename TestFixture::InType;
+ using InType = typename TestFixture::InType;
using OutType = typename TestFixture::OutType;
- cells2verts_scalar_lib<InType,OutType>(
- this->p_cell_in.data(),
- this->vert_cell_idx.data(),
- this->vert_cell_blk.data(),
- this->coeff_int_verts.data(),
- this->p_vert_out.data(),
- this->i_startblk,
- this->i_endblk,
- this->i_startidx,
- this->i_endidx,
- this->slev,
- this->elev,
- this->nproma,
- this->nlev,
- this->nblks_c,
- this->nblks_v,
- this->lacc,
- this->acc_async);
+ cells2verts_scalar_lib<InType, OutType>(
+ this->p_cell_in.data(), this->vert_cell_idx.data(),
+ this->vert_cell_blk.data(), this->coeff_int_verts.data(),
+ this->p_vert_out.data(), this->i_startblk, this->i_endblk,
+ this->i_startidx, this->i_endidx, this->slev, this->elev, this->nproma,
+ this->nlev, this->nblks_c, this->nblks_v, this->lacc, this->acc_async);
// Check the outputs only for blocks in the range
// { [i_startblk, i_endblk], [slev,elev], [i_startidx_in, i_endidx_in] }
@@ -489,10 +441,14 @@ TYPED_TEST(InterpolationScalarMixedTestFixture, cells2verts) {
for (int level = this->slev; level < this->elev; ++level) {
for (int i = this->i_startidx; i < this->i_endidx; ++i) {
// Compute the linear index for a 3D array in column-major order:
- size_t idx = i + level * this->nproma + block * this->nproma * this->nlev;
- // Since every contribution is 1 and there are 6 stencil points, expect 6.
- EXPECT_NEAR(this->p_vert_out[idx], static_cast<OutType>(6), static_cast<OutType>(1e-5))
- << "Failure at block " << block << ", level " << level << ", index " << i;
+ size_t idx =
+ i + level * this->nproma + block * this->nproma * this->nlev;
+ // Since every contribution is 1 and there are 6 stencil points,
+ // expect 6.
+ EXPECT_NEAR(this->p_vert_out[idx], static_cast<OutType>(6),
+ static_cast<OutType>(1e-5))
+ << "Failure at block " << block << ", level " << level << ", index "
+ << i;
}
}
}
@@ -504,30 +460,35 @@ TYPED_TEST(InterpolationScalarMixedTestFixture, cells2verts) {
//
////////////////////////////////////////////////////////////////////////////////
-// The test for cells2verts_ri is similar to cells2verts, but is done here separtely
-// to avoid as a differebt template instantiation is needed for the function call
+// The test for cells2verts_ri is similar to cells2verts, but is done here
+// separtely to avoid as a differebt template instantiation is needed for the
+// function call
template <typename Types>
-class Cells2vertsriScalarLibTestFixture : public testing::Test, public interp_dimensions{
+class Cells2vertsriScalarLibTestFixture : public testing::Test,
+ public interp_dimensions {
public:
- using InType = typename Types::in_type;
- using OutType = typename Types::out_type;
+ using InType = typename Types::in_type;
+ using OutType = typename Types::out_type;
// Arrays stored in std::vector.
- std::vector<InType> p_cell_in; // Dimensions: (nproma, nlev, nblks_c)
- std::vector<int> vert_cell_idx; // Dimensions: (nproma, nblks_v, 6)
- std::vector<int> vert_cell_blk; // Dimensions: (nproma, nblks_v, 6)
- std::vector<InType> coeff_int; // Dimensions: (nproma, 6, nblks_v)
- std::vector<OutType> p_vert_out; // Dimensions: (nproma, nlev, nblks_v)
+ std::vector<InType> p_cell_in; // Dimensions: (nproma, nlev, nblks_c)
+ std::vector<int> vert_cell_idx; // Dimensions: (nproma, nblks_v, 6)
+ std::vector<int> vert_cell_blk; // Dimensions: (nproma, nblks_v, 6)
+ std::vector<InType> coeff_int; // Dimensions: (nproma, 6, nblks_v)
+ std::vector<OutType> p_vert_out; // Dimensions: (nproma, nlev, nblks_v)
Cells2vertsriScalarLibTestFixture() {
// Allocate and initialize inputs.
- p_cell_in.resize(num_elements_3d<InType>(nproma, nlev, nblks_c), static_cast<InType>(1));
+ p_cell_in.resize(num_elements_3d<InType>(nproma, nlev, nblks_c),
+ static_cast<InType>(1));
vert_cell_idx.resize(num_elements_3d<int>(nproma, nblks_v, 6), 1);
vert_cell_blk.resize(num_elements_3d<int>(nproma, nblks_v, 6), 0);
- coeff_int.resize(num_elements_3d<InType>(nproma, 6, nblks_v), static_cast<InType>(1));
+ coeff_int.resize(num_elements_3d<InType>(nproma, 6, nblks_v),
+ static_cast<InType>(1));
// Allocate output arrays and initialize to zero.
- p_vert_out.resize(num_elements_3d<OutType>(nproma, nlev, nblks_v), static_cast<OutType>(0));
+ p_vert_out.resize(num_elements_3d<OutType>(nproma, nlev, nblks_v),
+ static_cast<OutType>(0));
}
};
@@ -536,28 +497,16 @@ TYPED_TEST_SUITE(Cells2vertsriScalarLibTestFixture, MixedTypes);
// Add test
TYPED_TEST(Cells2vertsriScalarLibTestFixture, cells2verts_ri) {
- using InType = typename TestFixture::InType;
- using OutType = typename TestFixture::OutType;
+ using InType = typename TestFixture::InType;
+ using OutType = typename TestFixture::OutType;
// Call the function
- cells2verts_scalar_ri_lib<InType,OutType>(
- this->p_cell_in.data(),
- this->vert_cell_idx.data(),
- this->vert_cell_blk.data(),
- this->coeff_int.data(),
- this->p_vert_out.data(),
- this->i_startblk,
- this->i_endblk,
- this->i_startidx,
- this->i_endidx,
- this->slev,
- this->elev,
- this->nproma,
- this->nlev,
- this->nblks_c,
- this->nblks_v,
- this->lacc,
- this->acc_async);
+ cells2verts_scalar_ri_lib<InType, OutType>(
+ this->p_cell_in.data(), this->vert_cell_idx.data(),
+ this->vert_cell_blk.data(), this->coeff_int.data(),
+ this->p_vert_out.data(), this->i_startblk, this->i_endblk,
+ this->i_startidx, this->i_endidx, this->slev, this->elev, this->nproma,
+ this->nlev, this->nblks_c, this->nblks_v, this->lacc, this->acc_async);
// Check the outputs only for blocks in the range
// { [i_startblk, i_endblk], [slev,elev], [i_startidx_in, i_endidx_in] }
@@ -566,13 +515,17 @@ TYPED_TEST(Cells2vertsriScalarLibTestFixture, cells2verts_ri) {
for (int i = this->i_startidx; i < this->i_endidx; ++i) {
// Compute the linear index for a 3D array in column-major order:
#ifdef __LOOP_EXCHANGE
- size_t idx = i + level * this->nproma + block * this->nproma * this->nlev;
+ size_t idx =
+ i + level * this->nproma + block * this->nproma * this->nlev;
#else
size_t idx = level + i * this->nlev + block * this->nproma * this->nlev;
#endif
- // Since every contribution is 1 and there are 6 stencil points, expect 6.
- EXPECT_NEAR(this->p_vert_out[idx], static_cast<OutType>(6), static_cast<OutType>(1e-5))
- << "Failure at block " << block << ", level " << level << ", index " << i;
+ // Since every contribution is 1 and there are 6 stencil points,
+ // expect 6.
+ EXPECT_NEAR(this->p_vert_out[idx], static_cast<OutType>(6),
+ static_cast<OutType>(1e-5))
+ << "Failure at block " << block << ", level " << level << ", index "
+ << i;
}
}
}
diff --git a/test/c/test_interpolation_vector.cpp b/test/c/test_interpolation_vector.cpp
index 974efcd..680fb6e 100644
--- a/test/c/test_interpolation_vector.cpp
+++ b/test/c/test_interpolation_vector.cpp
@@ -9,30 +9,31 @@
// SPDX-License-Identifier: BSD-3-Clause
// ---------------------------------------------------------------
-#include <gtest/gtest.h>
#include <Kokkos_Core.hpp>
+#include <gtest/gtest.h>
#include <vector>
#include "mo_lib_interpolation_vector.hpp"
// Dimensions for the test (small, trivial test).
-// We assume Fortran ordering: column-major, but our C wrappers will wrap raw pointers into Kokkos::Views with LayoutLeft.
+// We assume Fortran ordering: column-major, but our C wrappers will wrap raw
+// pointers into Kokkos::Views with LayoutLeft.
constexpr int nproma = 2;
-constexpr int nlev = 3;
-constexpr int nblks_e = 2; // For the edge arrays (p_vn_in, p_vt_in)
-constexpr int nblks_c = 2; // For the cell arrays and interpolation coefficients
+constexpr int nlev = 3;
+constexpr int nblks_e = 2; // For the edge arrays (p_vn_in, p_vt_in)
+constexpr int nblks_c = 2; // For the cell arrays and interpolation coefficients
// For the get_indices_c_lib inputs.
-constexpr int i_startblk = 0;
-constexpr int i_endblk = 1; // two blocks: indices 0 and 1
+constexpr int i_startblk = 0;
+constexpr int i_endblk = 1; // two blocks: indices 0 and 1
constexpr int i_startidx_in = 0;
-constexpr int i_endidx_in = nproma - 1; // 0 and 1
-constexpr int slev = 0;
-constexpr int elev = nlev - 1; // 0 .. 2
+constexpr int i_endidx_in = nproma - 1; // 0 and 1
+constexpr int slev = 0;
+constexpr int elev = nlev - 1; // 0 .. 2
-// Helper to compute total number of elements for a 3D array stored in column-major order.
-template<typename T>
-size_t num_elements(int dim1, int dim2, int dim3) {
+// Helper to compute total number of elements for a 3D array stored in
+// column-major order.
+template <typename T> size_t num_elements(int dim1, int dim2, int dim3) {
return static_cast<size_t>(dim1) * dim2 * dim3;
}
@@ -47,12 +48,13 @@ TEST(Edges2CellsTest, DPTest) {
// Here we set cell_edge_idx to 1, 2, 1 for every triple.
for (int i = 0; i < num_elements<int>(nproma, nblks_c, 3); i += 3) {
- cell_edge_idx[i] = 1;
- cell_edge_idx[i+1] = 2;
- cell_edge_idx[i+2] = 1;
+ cell_edge_idx[i] = 1;
+ cell_edge_idx[i + 1] = 2;
+ cell_edge_idx[i + 2] = 1;
}
- // Similarly, set cell_edge_blk to all ones (valid since nblks_e=2, so index 1 means block 0 after subtracting 1).
- // e_bln_c_u and e_bln_c_v: dimensions [nproma, 6, nblks_c]
+ // Similarly, set cell_edge_blk to all ones (valid since nblks_e=2, so index 1
+ // means block 0 after subtracting 1). e_bln_c_u and e_bln_c_v: dimensions
+ // [nproma, 6, nblks_c]
std::vector<double> e_bln_c_u(num_elements<double>(nproma, 6, nblks_c), 1.0);
std::vector<double> e_bln_c_v(num_elements<double>(nproma, 6, nblks_c), 1.0);
// Output arrays: dimensions [nproma, nlev, nblks_c]
@@ -64,15 +66,10 @@ TEST(Edges2CellsTest, DPTest) {
// Call the dp (double precision) version.
edges2cells_vector_lib<double>(
- p_vn_in.data(), p_vt_in.data(),
- cell_edge_idx.data(), cell_edge_blk.data(),
- e_bln_c_u.data(), e_bln_c_v.data(),
- p_u_out.data(), p_v_out.data(),
- i_startblk, i_endblk,
- i_startidx_in, i_endidx_in,
- slev, elev,
- nproma,
- nlev, nblks_e, nblks_c);
+ p_vn_in.data(), p_vt_in.data(), cell_edge_idx.data(),
+ cell_edge_blk.data(), e_bln_c_u.data(), e_bln_c_v.data(), p_u_out.data(),
+ p_v_out.data(), i_startblk, i_endblk, i_startidx_in, i_endidx_in, slev,
+ elev, nproma, nlev, nblks_e, nblks_c);
// Check that for each computed cell in p_u_out and p_v_out, the value is 6.
// This is because for each cell, the kernel adds 6 terms of 1*1.
@@ -91,9 +88,9 @@ TEST(Edges2CellsTest, SPTest) {
std::vector<int> cell_edge_blk(num_elements<int>(nproma, nblks_c, 3), 1);
// Set cell_edge_idx values to 1, 2, 1.
for (int i = 0; i < num_elements<int>(nproma, nblks_c, 3); i += 3) {
- cell_edge_idx[i] = 1;
- cell_edge_idx[i+1] = 2;
- cell_edge_idx[i+2] = 1;
+ cell_edge_idx[i] = 1;
+ cell_edge_idx[i + 1] = 2;
+ cell_edge_idx[i + 2] = 1;
}
std::vector<float> e_bln_c_u(num_elements<float>(nproma, 6, nblks_c), 1.0f);
std::vector<float> e_bln_c_v(num_elements<float>(nproma, 6, nblks_c), 1.0f);
@@ -105,15 +102,10 @@ TEST(Edges2CellsTest, SPTest) {
// Call the sp (float precision) version.
edges2cells_vector_lib<float>(
- p_vn_in.data(), p_vt_in.data(),
- cell_edge_idx.data(), cell_edge_blk.data(),
- e_bln_c_u.data(), e_bln_c_v.data(),
- p_u_out.data(), p_v_out.data(),
- i_startblk, i_endblk,
- i_startidx_in, i_endidx_in,
- slev, elev,
- nproma,
- nlev, nblks_e, nblks_c);
+ p_vn_in.data(), p_vt_in.data(), cell_edge_idx.data(),
+ cell_edge_blk.data(), e_bln_c_u.data(), e_bln_c_v.data(), p_u_out.data(),
+ p_v_out.data(), i_startblk, i_endblk, i_startidx_in, i_endidx_in, slev,
+ elev, nproma, nlev, nblks_e, nblks_c);
// Verify that every computed output equals 6.
for (size_t idx = 0; idx < p_u_out.size(); ++idx) {
--
GitLab