(towards #2671) Parallel array privatisation

.github/workflows/nemo_v5_tests.yml

@@ -80,26 +80,25 @@ jobs:
         export PSYCLONE_NEMO_DIR=${GITHUB_WORKSPACE}/examples/nemo/scripts
         export PSYCLONE_HOME=${PWD}/.runner_venv
         export NEMO_DIR=${HOME}/NEMOv5
+        export TEST_DIR=BENCH_PASSTHROUGH_GCC
 
         # Set up FCM: PATHs are loaded from SPACK, we only need to set the FCFLAGS
         cd $NEMO_DIR
         cp $PSYCLONE_NEMO_DIR/KGOs/arch-linux_spack.fcm arch/arch-linux_spack.fcm
         export FCFLAGS="-fdefault-real-8 -O2 -fcray-pointer -ffree-line-length-none -g"
 
         # Clean up and compile
-        ./makenemo -r BENCH -m linux_spack -n BENCH_PASSTHROUGH_GCC clean -y
-        ./makenemo -r BENCH -m linux_spack -n BENCH_PASSTHROUGH_GCC -p ${PSYCLONE_NEMO_DIR}/passthrough.py \
+        ./makenemo -r BENCH -m linux_spack -n ${TEST_DIR} clean -y
+        ./makenemo -r BENCH -m linux_spack -n ${TEST_DIR} -p ${PSYCLONE_NEMO_DIR}/passthrough.py \
             add_key "key_nosignedzero" -j 4 -v 1
 
         # Run test
-        cd $NEMO_DIR/tests/BENCH_PASSTHROUGH_GCC/EXP00
+        cd $NEMO_DIR/tests/${TEST_DIR}/EXP00
         cp $PSYCLONE_NEMO_DIR/KGOs/namelist_cfg_bench_small namelist_cfg
         mpirun -np 4 ./nemo
         tail run.stat
         # This was produced with gfortran, so we can do an exact diff
-        diff $PSYCLONE_NEMO_DIR/KGOs/run.stat.bench.small.100steps run.stat
-        export VAR_TIME=$(awk '/ step /{print $3}' timing.output | head -n 1 | sed -e 's/s//')
-        echo "Time-stepping duration = " $VAR_TIME
+        diff $PSYCLONE_NEMO_DIR/KGOs/run.stat.bench.gfortran.small.100steps run.stat
 
     - name: NEMO 5.0 nvidia passthrough
       run: |
@@ -110,24 +109,25 @@ jobs:
         export PSYCLONE_NEMO_DIR=${GITHUB_WORKSPACE}/examples/nemo/scripts
         export PSYCLONE_HOME=${PWD}/.runner_venv
         export NEMO_DIR=${HOME}/NEMOv5
+        export TEST_DIR=BENCH_PASSTHROUGH_NVHPC
 
         # Set up FCM: PATHs are loaded from SPACK, we only need to set the FCFLAGS
         cd $NEMO_DIR
         cp $PSYCLONE_NEMO_DIR/KGOs/arch-linux_spack.fcm arch/arch-linux_spack.fcm
-        export FCFLAGS="-i4 -Mr8 -O2 -Minline -Mcray=pointer -Mpre -g"
+        export FCFLAGS="-i4 -Mr8 -O2 -g"
 
         # Clean up and compile
         # Without key_mpi_off it fails to compile (even without psyclone)
-        ./makenemo -r BENCH -m linux_spack -n BENCH_PASSTHROUGH_NVHPC clean -y
-        ./makenemo -r BENCH -m linux_spack -n BENCH_PASSTHROUGH_NVHPC -p ${PSYCLONE_NEMO_DIR}/passthrough.py \
+        ./makenemo -r BENCH -m linux_spack -n ${TEST_DIR} clean -y
+        ./makenemo -r BENCH -m linux_spack -n ${TEST_DIR} -p ${PSYCLONE_NEMO_DIR}/passthrough.py \
           add_key "key_mpi_off key_nosignedzero" -j 4 -v 1
 
         # Run test
-        cd $NEMO_DIR/tests/BENCH_PASSTHROUGH_NVHPC/EXP00
+        cd $NEMO_DIR/tests/${TEST_DIR}/EXP00
         cp $PSYCLONE_NEMO_DIR/KGOs/namelist_cfg_bench_small namelist_cfg
         ./nemo
-        python $PSYCLONE_NEMO_DIR/compare_ouput.py $PSYCLONE_NEMO_DIR/KGOs/run.stat.bench.small.100steps run.stat
         tail run.stat
+        diff $PSYCLONE_NEMO_DIR/KGOs/run.stat.bench.nvhpc.small.100steps run.stat
         export VAR_TIME=$(awk '/ step /{print $3}' timing.output | head -n 1 | sed -e 's/s//')
         echo "Time-stepping duration = " $VAR_TIME
 
@@ -140,25 +140,25 @@ jobs:
         export PSYCLONE_NEMO_DIR=${GITHUB_WORKSPACE}/examples/nemo/scripts
         export PSYCLONE_HOME=${PWD}/.runner_venv
         export NEMO_DIR=${HOME}/NEMOv5
+        export TEST_DIR=BENCH_PASSTHROUGH_ONEAPI
 
         # Set up FCM: PATHs are loaded from SPACK, we only need to set the FCFLAGS
         cd $NEMO_DIR
         cp $PSYCLONE_NEMO_DIR/KGOs/arch-linux_spack.fcm arch/arch-linux_spack.fcm
         export FCFLAGS="-i4 -r8 -O2 -fp-model precise -fno-alias -g"
 
         # Clean up and compile
-        ./makenemo -r BENCH -m linux_spack -n BENCH_PASSTHROUGH_ONEAPI clean -y
-        ./makenemo -r BENCH -m linux_spack -n BENCH_PASSTHROUGH_ONEAPI -p ${PSYCLONE_NEMO_DIR}/passthrough.py \
+        ./makenemo -r BENCH -m linux_spack -n ${TEST_DIR} clean -y
+        ./makenemo -r BENCH -m linux_spack -n ${TEST_DIR} -p ${PSYCLONE_NEMO_DIR}/passthrough.py \
           add_key "key_nosignedzero" -j 4 -v 1
 
         # Run test
-        cd $NEMO_DIR/tests/BENCH_PASSTHROUGH_ONEAPI/EXP00
+        cd $NEMO_DIR/tests/${TEST_DIR}/EXP00
         cp $PSYCLONE_NEMO_DIR/KGOs/namelist_cfg_bench_small namelist_cfg
         mpirun -np 6 ./nemo
         tail run.stat
-        python $PSYCLONE_NEMO_DIR/compare_ouput.py $PSYCLONE_NEMO_DIR/KGOs/run.stat.bench.small.100steps run.stat
-        export VAR_TIME=$(awk '/ step /{print $3}' timing.output | head -n 1 | sed -e 's/s//')
-        echo "Time-stepping duration = " $VAR_TIME
+        # We don't compare all digits because they are results from a different compiler
+        python $PSYCLONE_NEMO_DIR/compare_ouput.py $PSYCLONE_NEMO_DIR/KGOs/run.stat.gfrotran.bench.small.100steps run.stat
 
     - name: NEMO 5.0 gfortran OpenMP for CPUs
       run: |
@@ -169,25 +169,34 @@ jobs:
         export PSYCLONE_NEMO_DIR=${GITHUB_WORKSPACE}/examples/nemo/scripts
         export PSYCLONE_HOME=${PWD}/.runner_venv
         export NEMO_DIR=${HOME}/NEMOv5
+        export TEST_DIR=BENCH_OMP_THREADING_GCC
 
         # Set up FCM: PATHs are loaded from SPACK, we only need to set the FCFLAGS
         cd $NEMO_DIR
         cp $PSYCLONE_NEMO_DIR/KGOs/arch-linux_spack.fcm arch/arch-linux_spack.fcm
         export FCFLAGS="-fdefault-real-8 -O2 -fcray-pointer -ffree-line-length-none -g -fopenmp"
 
         # Clean up and compile
-        ./makenemo -r BENCH -m linux_spack -n BENCH_OMP_THREADING_GCC clean -y
-        ./makenemo -r BENCH -m linux_spack -n BENCH_OMP_THREADING_GCC -p ${PSYCLONE_NEMO_DIR}/omp_cpu_trans.py \
+        ./makenemo -r BENCH -m linux_spack -n ${TEST_DIR} clean -y
+        ./makenemo -r BENCH -m linux_spack -n ${TEST_DIR} -p ${PSYCLONE_NEMO_DIR}/omp_cpu_trans.py \
           add_key "key_nosignedzero" -j 4 -v 1
 
         # Run test
-        cd $NEMO_DIR/tests/BENCH_OMP_THREADING_GCC/EXP00
+        cd $NEMO_DIR/tests/${TEST_DIR}/EXP00
         cp $PSYCLONE_NEMO_DIR/KGOs/namelist_cfg_bench_small namelist_cfg
         OMP_NUM_THREADS=4 mpirun -np 1 ./nemo
         tail run.stat
-        python $PSYCLONE_NEMO_DIR/compare_ouput.py $PSYCLONE_NEMO_DIR/KGOs/run.stat.bench.small.100steps run.stat
-        export VAR_TIME=$(awk '/ step /{print $3}' timing.output | head -n 1 | sed -e 's/s//')
-        echo "Time-stepping duration = " $VAR_TIME
+        diff $PSYCLONE_NEMO_DIR/KGOs/run.stat.bench.gfortran.small.100steps run.stat
+        export TIME_microsec=$(grep "Mean:" timing.output | head -n 1 | awk '{print $2}')
+        export TIME_sec=$(bc <<< "${TIME_microsec} / 10000")
+        ${HOME}/mongosh-2.1.1-linux-x64/bin/mongosh \
+          "mongodb+srv://cluster0.x8ncpxi.mongodb.net/PerformanceMonitoring" \
+          --quiet --apiVersion 1 --username ${{ secrets.MONGODB_USERNAME }} \
+          --password ${{ secrets.MONGODB_PASSWORD }} \
+          --eval 'db.GitHub_CI.insertOne({branch_name: "'"$GITHUB_REF_NAME"'", commit: "'"$GITHUB_SHA"'",
+          github_job: "'"$GITHUB_RUN_ID"'"-"'"$GITHUB_RUN_ATTEMPT"'",
+          ci_test: "NEMOv5 OpenMP for CPU", nemo_version: "NEMOv5", system: "GlaDos",
+          compiler:"gfortran-14" , date: new Date(), elapsed_time: '"${TIME_sec}"'})'
 
     - name: NEMO 5.0 nvidia OpenMP for GPUs (managed memory)
       run: |
@@ -198,24 +207,33 @@ jobs:
         export PSYCLONE_NEMO_DIR=${GITHUB_WORKSPACE}/examples/nemo/scripts
         export PSYCLONE_HOME=${PWD}/.runner_venv
         export NEMO_DIR=${HOME}/NEMOv5
+        export TEST_DIR=BENCH_OMP_OFFLOAD_NVHPC
 
         # Set up FCM: PATHs are loaded from SPACK, we only need to set the FCFLAGS
         # We compile at -O1 to permit comparison of the results.
         cd $NEMO_DIR
         cp $PSYCLONE_NEMO_DIR/KGOs/arch-linux_spack.fcm arch/arch-linux_spack.fcm
-        export FCFLAGS="-i4 -Mr8 -O3 -Minline -Mcray=pointer -Mpre -g -mp=gpu -gpu=managed"
+        export FCFLAGS="-i4 -Mr8 -O2 -g -mp=gpu -gpu=managed"
 
         # Clean up and compile
         # Without key_mpi_off it fails to compile (even without psyclone)
-        ./makenemo -r BENCH -m linux_spack -n BENCH_OMP_OFFLOAD_NVHPC clean -y
-        ./makenemo -r BENCH -m linux_spack -n BENCH_OMP_OFFLOAD_NVHPC -p ${PSYCLONE_NEMO_DIR}/omp_gpu_trans.py \
+        ./makenemo -r BENCH -m linux_spack -n ${TEST_DIR} clean -y
+        ./makenemo -r BENCH -m linux_spack -n ${TEST_DIR} -p ${PSYCLONE_NEMO_DIR}/omp_gpu_trans.py \
           add_key "key_mpi_off key_nosignedzero" -j 4 -v 1
 
         # Run test
-        cd $NEMO_DIR/tests/BENCH_OMP_OFFLOAD_NVHPC/EXP00
+        cd $NEMO_DIR/tests/${TEST_DIR}/EXP00
         cp $PSYCLONE_NEMO_DIR/KGOs/namelist_cfg_bench_small namelist_cfg
         ./nemo
         tail run.stat
-        python $PSYCLONE_NEMO_DIR/compare_ouput.py $PSYCLONE_NEMO_DIR/KGOs/run.stat.bench.small.100steps run.stat
-        export VAR_TIME=$(awk '/ step /{print $3}' timing.output | head -n 1 | sed -e 's/s//')
-        echo "Time-stepping duration = " $VAR_TIME
+        diff $PSYCLONE_NEMO_DIR/KGOs/run.stat.bench.nvhpc.small.100steps run.stat
+        export TIME_microsec=$(grep "Mean:" timing.output | head -n 1 | awk '{print $2}')
+        export TIME_sec=$(bc <<< "${TIME_microsec} / 10000")
+        ${HOME}/mongosh-2.1.1-linux-x64/bin/mongosh \
+          "mongodb+srv://cluster0.x8ncpxi.mongodb.net/PerformanceMonitoring" \
+          --quiet --apiVersion 1 --username ${{ secrets.MONGODB_USERNAME }} \
+          --password ${{ secrets.MONGODB_PASSWORD }} \
+          --eval 'db.GitHub_CI.insertOne({branch_name: "'"$GITHUB_REF_NAME"'", commit: "'"$GITHUB_SHA"'",
+          github_job: "'"$GITHUB_RUN_ID"'"-"'"$GITHUB_RUN_ATTEMPT"'",
+          ci_test: "NEMOv5 OpenMP for GPU", nemo_version: "NEMOv5", system: "GlaDos",
+          compiler:"nvhpc-24.5" , date: new Date(), elapsed_time: '"${TIME_sec}"'})'