portelli/Grid
1
0
Fork 0
mirror of https://github.com/paboyle/Grid.git synced 2025-06-14 13:57:07 +01:00
Code Releases Activity

Performance regressed and is OK in icpx 2023.2

Browse Source
This commit is contained in:
Peter Boyle
2023-10-03 15:53:14 +00:00
parent afc316f501
commit d93eac7b1c
8 changed files with 1063 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
systems/Sunspot/benchmarks/bench.pbs
View File

@ -20,7 +20,7 @@ unset OMP_PLACES
cd $PBS_O_WORKDIR
qsub jobscript.pbs
#qsub jobscript.pbs
echo Jobid: $PBS_JOBID
echo Running on host `hostname`
@ -44,3 +44,4 @@ CMD="mpiexec -np ${NTOTRANKS} -ppn ${NRANKS} -d ${NDEPTH} --cpu-bind=depth -enva
./Benchmark_dwf_fp32 --mpi 1.1.2.6 --grid 16.32.64.192 --comms-overlap \
--shm-mpi 0 --shm 2048 --device-mem 32000 --accelerator-threads 32"
$CMD

4
systems/Sunspot/benchmarks/gpu_tile_compact.sh
View File

@ -45,8 +45,8 @@ echo "rank $PALS_RANKID ; local rank $PALS_LOCAL_RANKID ; ZE_AFFINITY_MASK=$ZE_A
if [ $PALS_LOCAL_RANKID = 0 ]
then
onetrace --chrome-device-timeline "$@"
# "$@"
# onetrace --chrome-device-timeline "$@"
"$@"
else
"$@"
fi

2
systems/Sunspot/config-command
View File

@ -11,6 +11,6 @@ TOOLS=$HOME/tools
--enable-unified=no \
MPICXX=mpicxx \
CXX=icpx \
LDFLAGS="-fiopenmp -fsycl -fsycl-device-code-split=per_kernel -fsycl-device-lib=all -lze_loader -lapmidg -L$TOOLS/lib64/" \
LDFLAGS="-fiopenmp -fsycl -fsycl-device-code-split=per_kernel -fsycl-device-lib=all -lze_loader -L$TOOLS/lib64/" \
CXXFLAGS="-fiopenmp -fsycl-unnamed-lambda -fsycl -I$INSTALL/include -Wno-tautological-compare -I$HOME/ -I$TOOLS/include"