Name	Name	Last commit message	Last commit date
parent directory ..
GEM3Dsmall_1x1x1_100	GEM3Dsmall_1x1x1_100
GEM3Dsmall_2x1x1_100	GEM3Dsmall_2x1x1_100
GEM3Dsmall_2x2x1_100	GEM3Dsmall_2x2x1_100
GEM3Dsmall_2x2x2_100	GEM3Dsmall_2x2x2_100
GEM3Dsmall_4x2x2_100	GEM3Dsmall_4x2x2_100
GEM3Dsmall_4x4x2_100	GEM3Dsmall_4x4x2_100
GEM3Dsmall_4x4x4_100	GEM3Dsmall_4x4x4_100
.gitignore	.gitignore
GH200_release_baseline.csv	GH200_release_baseline.csv
benchmark.sh	benchmark.sh
benchmarkBarChart.py	benchmarkBarChart.py
build.sh	build.sh
clean.sh	clean.sh
outputCSV.sh	outputCSV.sh
readme.md	readme.md

Name

Last commit message

Last commit date

Benchmark

Usage

Create a folder for each inputfile you'd like to benchmark, and copy the inputfile into the folders.
Run the script benchmark.sh, 2 OpenMP threads per process by default, can be modified in benchmark.sh.
Wait with a cup of coffee, check the output from time to time, they are executed in serial.
Done.

NOTE: The name of the folder must be in the format name_XxYxZ_cycle, as the script relies on the second segment to launch MPI processes.

There're few baseline files in this folder, which can be used as a reference for performance evaluation.