Fix C_MPI library linking

Fix all warnings Less verbose default flags
UoB-HPC · Aug 11, 2024 · 3e10ff9 · 3e10ff9
1 parent 72f1295
commit 3e10ff9
Show file tree

Hide file tree

Showing 6 changed files with 35 additions and 45 deletions.
diff --git a/CMakeLists.txt b/CMakeLists.txt
@@ -4,7 +4,7 @@ if (CMAKE_VERSION VERSION_GREATER_EQUAL "3.24.0")
     cmake_policy(SET CMP0135 NEW)
 endif ()
 
-project(cloverleaf VERSION 1.0 LANGUAGES CXX)
+project(cloverleaf VERSION 1.0 LANGUAGES C CXX)
 set(CMAKE_CXX_EXTENSIONS OFF)
 set(CMAKE_CXX_STANDARD_REQUIRED ON)
 
@@ -35,8 +35,8 @@ if ((NOT BUILD_TYPE STREQUAL RELEASE) AND (NOT BUILD_TYPE STREQUAL DEBUG))
 endif ()
 
 # setup some defaults flags for everything
-set(DEFAULT_DEBUG_FLAGS -Wall -O2)
-set(DEFAULT_RELEASE_FLAGS -Wall -O3)
+set(DEFAULT_DEBUG_FLAGS -Wall -Wno-unused-parameter -Wno-unused-function -Wno-unused-variable -O2)
+set(DEFAULT_RELEASE_FLAGS -Wall -Wno-unused-parameter -Wno-unused-function -Wno-unused-variable -O3)
 
 macro(hint_flag FLAG DESCRIPTION)
     if (NOT DEFINED ${FLAG})
@@ -71,7 +71,7 @@ option(USE_ONETBB "Enable oneTBB library for *supported* models. Enabling this o
 
 option(FETCH_ONETBB "Fetch (download) the oneTBB library for *supported* models. This uses CMake's
                   FetchContent feature. Specify version by setting FETCH_ONETBB_VERSION" OFF)
-set(FETCH_ONETBB_VERSION "v2021.10.0" CACHE STRING "Specify version of oneTBB to use if FETCH_ONETBB is ON")
+set(FETCH_ONETBB_VERSION "v2021.13.0" CACHE STRING "Specify version of oneTBB to use if FETCH_ONETBB is ON")
 
 if (FETCH_ONETBB)
     FetchContent_Declare(
@@ -96,7 +96,7 @@ option(USE_ONEDPL "Enable oneDPL library for *supported* models. Enabling this o
 
 option(FETCH_ONEDPL "Fetch (download) the oneDPL library for *supported* models. This uses CMake's
                   FetchContent feature. Specify version by setting FETCH_ONEDPL_VERSION" OFF)
-set(FETCH_ONEDPL_VERSION "oneDPL-2022.2.0-rc1" CACHE STRING "Specify version of oneTBB to use if FETCH_ONEDPL is ON")
+set(FETCH_ONEDPL_VERSION "oneDPL-2022.6.0-rc1" CACHE STRING "Specify version of oneTBB to use if FETCH_ONEDPL is ON")
 
 if (FETCH_ONEDPL)
     FetchContent_Declare(

diff --git a/cmake/register_models.cmake b/cmake/register_models.cmake
@@ -8,12 +8,12 @@ endmacro()
 
 macro(register_append_cxx_flags CONFIG)
     if ("${CONFIG}" STREQUAL "ANY")
-        list(APPEND DEFAULT_RELEASE_CXX_FLAGS ${ARGN})
-        list(APPEND DEFAULT_DEBUG_CXX_FLAGS ${ARGN})
+        list(APPEND DEFAULT_RELEASE_FLAGS ${ARGN})
+        list(APPEND DEFAULT_DEBUG_FLAGS ${ARGN})
     elseif ("${CONFIG}" STREQUAL "RELEASE")
-        list(APPEND DEFAULT_RELEASE_CXX_FLAGS ${ARGN})
+        list(APPEND DEFAULT_RELEASE_FLAGS ${ARGN})
     elseif ("${CONFIG}" STREQUAL "DEBUG")
-        list(APPEND DEFAULT_DEBUG_CXX_FLAGS ${ARGN})
+        list(APPEND DEFAULT_DEBUG_FLAGS ${ARGN})
     else ()
         message(FATAL_ERROR "register_flags supports only RELEASE, DEBUG, or ANY for all configs, got `${CONFIG}`")
     endif ()

diff --git a/src/cuda/accelerate.cpp b/src/cuda/accelerate.cpp
@@ -21,15 +21,6 @@
 #include "context.h"
 #include "timer.h"
 
-//#define par_ranged2m(rg, f) \
-//{"KERNEL2D_START";                        \
-//                           \
-//    Range2d range = rg;                    \
-//    for (size_t j = range.fromY; j < range.toY; j++) {        \
-//        for (size_t i = range.fromX; i < range.toX; i++)  {"KERNEL2D_A";f"KERNEL2D_B";}                           \
-//    }                                                      \
-//"KERNEL2D_END";}
-
 // @brief Fortran acceleration kernel
 // @author Wayne Gaudin
 // @details The pressure and viscosity gradients are used to update the

diff --git a/src/cuda/calc_dt.cpp b/src/cuda/calc_dt.cpp
@@ -47,8 +47,6 @@ void calc_dt_kernel(global_variables &globals, int x_min, int x_max, int y_min,
   //   DO j=x_min,x_max
   //	Kokkos::MDRangePolicy <Kokkos::Rank<2>> policy({x_min + 1, y_min + 1}, {x_max + 2, y_max + 2});
 
-  auto policy = clover::Range2d(x_min + 1, y_min + 1, x_max + 2, y_max + 2);
-
   int xStart = x_min + 1, xEnd = x_max + 2;
   int yStart = y_min + 1, yEnd = y_max + 2;
   int sizeX = xEnd - xStart;

diff --git a/src/cuda/context.h b/src/cuda/context.h
@@ -144,6 +144,7 @@ template <typename T> struct Buffer2D {
       return sizeY;
     } else {
       static_assert(D < 2);
+      return 0; // make nvc++ happy
     }
   }
 

diff --git a/test.sh b/test.sh
@@ -2,7 +2,7 @@
 
 set -eu
 
-export NVHPC_DIR="/opt/nvidia/hpc_sdk/Linux_x86_64/23.5"
+export NVHPC_DIR="/opt/nvidia/hpc_sdk/Linux_x86_64/24.7"
 export CUDA_DIR="$NVHPC_DIR/cuda/"
 export PATH=$NVHPC_DIR/compilers/bin/:${PATH:-}
 export KOKKOS_DIR="/home/tom/Downloads/kokkos-4.0.01/"
@@ -13,7 +13,7 @@ export CPU_RANKS=$(($(nproc) / 2))
 export GPU_RANKS=3
 
 VERBOSE="ON"
-PROBLEM="InputDecks/clover_bm16_short.in"
+PROBLEM="InputDecks/clover_bm16_very_short.in"
 
 declare -A enabled_models=(
   ["serial"]=true
@@ -79,10 +79,10 @@ function test_nompi() {
   )
   (
     :
-    test_nompi "0" std-indices -DCMAKE_CXX_COMPILER=g++ -DCXX_EXTRA_FLAGS="-Ofast" -DUSE_TBB=ON
-    test_nompi "0" std-indices -DCMAKE_CXX_COMPILER=g++ -DCXX_EXTRA_FLAGS="-Ofast" -DUSE_ONEDPL=OPENMP
-    test_nompi "0" std-indices -DCMAKE_CXX_COMPILER=clang++ -DCXX_EXTRA_FLAGS="-Ofast" -DUSE_TBB=ON
-    test_nompi "0" std-indices -DCMAKE_CXX_COMPILER=clang++ -DCXX_EXTRA_FLAGS="-Ofast" -DUSE_ONEDPL=OPENMP
+    test_nompi "0" std-indices -DCMAKE_CXX_COMPILER=g++ -DCXX_EXTRA_FLAGS="-Ofast" -DUSE_ONETBB=ON -DFETCH_ONETBB=ON -DCXX_EXTRA_LIBRARIES=tbb
+    test_nompi "0" std-indices -DCMAKE_CXX_COMPILER=g++ -DCXX_EXTRA_FLAGS="-Ofast" -DUSE_ONEDPL=OPENMP -DFETCH_ONEDPL=ON
+    test_nompi "0" std-indices -DCMAKE_CXX_COMPILER=clang++ -DCXX_EXTRA_FLAGS="-Ofast" -DUSE_ONETBB=ON -DFETCH_ONETBB=ON -DCXX_EXTRA_LIBRARIES=tbb
+    test_nompi "0" std-indices -DCMAKE_CXX_COMPILER=clang++ -DCXX_EXTRA_FLAGS="-Ofast" -DUSE_ONEDPL=OPENMP -DFETCH_ONEDPL=ON
     test_nompi "0" std-indices -DCMAKE_CXX_COMPILER=nvc++ -DCXX_EXTRA_FLAGS="-Ofast;-stdpar;-target=multicore"
     test_nompi "0" std-indices -DCMAKE_CXX_COMPILER=nvc++ -DCXX_EXTRA_FLAGS="-Ofast;-stdpar;-gpu=cc61"
   )
@@ -110,9 +110,9 @@ function test_nompi() {
   )
   (
     :
-    test_nompi "0" cuda -DCXX_EXTRA_FLAGS="-Ofast" -DCMAKE_CUDA_COMPILER=nvcc -DCUDA_ARCH=sm_60
-    test_nompi "0" cuda -DMANAGED_ALLOC=ON -DCXX_EXTRA_FLAGS="-Ofast" -DCMAKE_CUDA_COMPILER=nvcc -DCUDA_ARCH=sm_60
-    test_nompi "0" cuda -DSYNC_ALL_KERNELS=ON -DCXX_EXTRA_FLAGS="-Ofast" -DCMAKE_CUDA_COMPILER=nvcc -DCUDA_ARCH=sm_60
+    test_nompi "0" cuda -DCXX_EXTRA_FLAGS="-O3" -DCMAKE_CUDA_COMPILER=nvcc -DCUDA_ARCH=sm_60
+    test_nompi "0" cuda -DMANAGED_ALLOC=ON -DCXX_EXTRA_FLAGS="-O3" -DCMAKE_CUDA_COMPILER=nvcc -DCUDA_ARCH=sm_60
+    test_nompi "0" cuda -DSYNC_ALL_KERNELS=ON -DCXX_EXTRA_FLAGS="-O3" -DCMAKE_CUDA_COMPILER=nvcc -DCUDA_ARCH=sm_60
   )
   (
     :
@@ -136,8 +136,8 @@ function test_nompi() {
   (
     :
     set +eu
-    source /opt/intel/oneapi/tbb/2021.10.0/env/vars.sh
-    source /opt/intel/oneapi/compiler/2023.2.0/env/vars.sh
+    source /opt/intel/oneapi/tbb/2021.13/env/vars.sh
+    source /opt/intel/oneapi/compiler/2024.2/env/vars.sh
     set -eu
     export DPCPP_CPU_NUM_CUS=1
     export DPCPP_CPU_SCHEDULE=static
@@ -149,8 +149,8 @@ function test_nompi() {
   (
     :
     set +eu
-    source /opt/intel/oneapi/tbb/2021.10.0/env/vars.sh
-    source /opt/intel/oneapi/compiler/2023.2.0/env/vars.sh --include-intel-llvm
+    source /opt/intel/oneapi/tbb/2021.13/env/vars.sh
+    source /opt/intel/oneapi/compiler/2024.2/env/vars.sh --include-intel-llvm
     set -eu
     cuda_sycl_flags="-fsycl-targets=nvptx64-nvidia-cuda;-Xsycl-target-backend;--cuda-gpu-arch=sm_60;--cuda-path=$CUDA_DIR"
     test_nompi "0" std-indices -DCMAKE_CXX_COMPILER=clang++ -DUSE_ONEDPL=DPCPP -DCXX_EXTRA_FLAGS="-fsycl;$cuda_sycl_flags"
@@ -180,10 +180,10 @@ function test_nompi() {
   test_mpi $CPU_RANKS "0" serial -DCMAKE_CXX_COMPILER=clang++
   (
     :
-    test_mpi $CPU_RANKS "0" std-indices -DCMAKE_CXX_COMPILER=g++ -DUSE_TBB=ON
-    test_mpi $CPU_RANKS "0" std-indices -DCMAKE_CXX_COMPILER=g++ -DUSE_ONEDPL=OPENMP
-    test_mpi $CPU_RANKS "0" std-indices -DCMAKE_CXX_COMPILER=clang++ -DUSE_TBB=ON
-    test_mpi $CPU_RANKS "0" std-indices -DCMAKE_CXX_COMPILER=clang++ -DUSE_ONEDPL=OPENMP
+    test_mpi $CPU_RANKS "0" std-indices -DCMAKE_CXX_COMPILER=g++ -DUSE_ONETBB=ON -DFETCH_ONETBB=ON -DCXX_EXTRA_LIBRARIES=tbb
+    test_mpi $CPU_RANKS "0" std-indices -DCMAKE_CXX_COMPILER=g++ -DUSE_ONEDPL=OPENMP -DFETCH_ONEDPL=ON
+    test_mpi $CPU_RANKS "0" std-indices -DCMAKE_CXX_COMPILER=clang++ -DUSE_ONETBB=ON -DFETCH_ONETBB=ON -DCXX_EXTRA_LIBRARIES=tbb
+    test_mpi $CPU_RANKS "0" std-indices -DCMAKE_CXX_COMPILER=clang++ -DUSE_ONEDPL=OPENMP -DFETCH_ONEDPL=ON
     test_mpi $CPU_RANKS "0" std-indices -DCMAKE_CXX_COMPILER=nvc++ -DCXX_EXTRA_FLAGS="-Ofast;-stdpar;-target=multicore"
     test_mpi $GPU_RANKS "0" std-indices -DCMAKE_CXX_COMPILER=nvc++ -DCXX_EXTRA_FLAGS="-Ofast;-stdpar;-gpu=cc61;--restrict"
   )
@@ -223,8 +223,8 @@ function test_nompi() {
     :
     set +eu
     module load mpi
-    source /opt/intel/oneapi/tbb/2021.10.0/env/vars.sh
-    source /opt/intel/oneapi/compiler/2023.2.0/env/vars.sh
+    source /opt/intel/oneapi/tbb/2021.13/env/vars.sh
+    source /opt/intel/oneapi/compiler/2024.2/env/vars.sh
     set -eu
     export DPCPP_CPU_NUM_CUS=1
     export DPCPP_CPU_SCHEDULE=static
@@ -250,8 +250,8 @@ function test_nompi() {
     :
     set +eu
     module load mpi
-    source /opt/intel/oneapi/tbb/2021.10.0/env/vars.sh
-    source /opt/intel/oneapi/compiler/2023.2.0/env/vars.sh --include-intel-llvm
+    source /opt/intel/oneapi/tbb/2021.13/env/vars.sh
+    source /opt/intel/oneapi/compiler/2024.2/env/vars.sh --include-intel-llvm
     set -eu
     cuda_sycl_flags="-fsycl-targets=nvptx64-nvidia-cuda;-Xsycl-target-backend;--cuda-gpu-arch=sm_60;--cuda-path=$CUDA_DIR"
     test_mpi $GPU_RANKS "0" std-indices -DCMAKE_CXX_COMPILER=clang++ -DUSE_ONEDPL=DPCPP -DCXX_EXTRA_FLAGS="-fsycl;$cuda_sycl_flags"
@@ -275,8 +275,8 @@ function test_nompi() {
 
 ### CUDA-aware MPI ###
 (
-  export MPI_HOME=/opt/nvidia/hpc_sdk/Linux_x86_64/23.5/comm_libs/openmpi/openmpi-3.1.5/
-  export PATH="/opt/nvidia/hpc_sdk/Linux_x86_64/23.5/comm_libs/openmpi/openmpi-3.1.5/bin/:${PATH:-}"
+  export MPI_HOME=/opt/nvidia/hpc_sdk/Linux_x86_64/24.7/comm_libs/openmpi/openmpi-3.1.5/
+  export PATH="/opt/nvidia/hpc_sdk/Linux_x86_64/24.7/comm_libs/openmpi/openmpi-3.1.5/bin/:${PATH:-}"
   (
     :
     test_mpi $GPU_RANKS "0" std-indices -DCMAKE_CXX_COMPILER=nvc++ -DCXX_EXTRA_FLAGS="-Ofast;-stdpar;-gpu=cc61;--restrict"
@@ -302,8 +302,8 @@ function test_nompi() {
   (
     :
     set +eu
-    source /opt/intel/oneapi/tbb/2021.10.0/env/vars.sh
-    source /opt/intel/oneapi/compiler/2023.2.0/env/vars.sh --include-intel-llvm
+    source /opt/intel/oneapi/tbb/2021.13/env/vars.sh
+    source /opt/intel/oneapi/compiler/2024.2/env/vars.sh --include-intel-llvm
     set -eu
     cuda_sycl_flags="-fsycl-targets=nvptx64-nvidia-cuda;-Xsycl-target-backend;--cuda-gpu-arch=sm_60;--cuda-path=$CUDA_DIR"
     test_mpi $GPU_RANKS "0" std-indices -DCMAKE_CXX_COMPILER=clang++ -DUSE_ONEDPL=DPCPP -DCXX_EXTRA_FLAGS="-fsycl;$cuda_sycl_flags"