Skip to content

Latest commit

 

History

History
40 lines (36 loc) · 2.12 KB

README.md

File metadata and controls

40 lines (36 loc) · 2.12 KB

KSTBench

This repository is a collection of python wrappers for popular benchmark tests on heterogeneous platform.
Sensible parameters are automatically set by the scripts without user inputs, although exhaustive parameters scan can be done using included templates. The development is done under direction of KISTI Infrastructure Center. We are grateful for the granted acccess to KISTI GPU cluster.

List of benchmarks

Caterogy Benchmark Version System Requirements
Memory Bandwidth STREAM_OMP 5.10 working gcc
Memory Bandwidth STREAM_CUDA 3.4 cuda >= 10.1
Disk IO IOZONE 3.419 working gcc
Disk IO IOR 3.3.0 openmpi >= 3
Linear Algebra HPL
HPL-AI
HPCG
21.4 singularity >= 3.4.1
openmpi >= 4
nvidia >= 450.36
connectx >= 4
Numerical Simulation QE 6.8 singularity >= 3.1
openmpi >= 3
nvidia >= 450.36
Numerical Simulation GROMACS 2021.3 singularity >= 3.1
nvidia >= 450.36

Python requirements:

Usage:

  • Download KSTBench:
git clone https://github.com/vitduck/KSTBench
  • Add KSTBench modules to python search path
cd KSTBench/src 
export PYTHONPATH=`pwd`
  • Start slurm iteractive session
salloc --partition=cas_v100_2 --nodes=2 --ntasks-per-node=40 --gres=gpu:2 --time=12:00:00 --comment=etc
  • Perform all test
cd test
./test_all.py