specfem / scaling-benchmarks Goto Github PK

View Code? Open in Web Editor NEW

0.0 0.0 2.0 1.61 MB

Scaling benchmark results of SPECFEM codes for different HPC clusters

Shell 3.83% Perl 9.40% C 42.49% Makefile 5.86% C++ 34.36% Gnuplot 3.77% Python 0.30%

scaling-benchmarks's People

Contributors

Watchers

Forkers

chukren wjlei1990

scaling-benchmarks's Issues

Questions regarding to Strong and Scaling on Summit

Hi Daniel,

Thanks for putting all these together. I have some questions regarding to the parameters used for benchmark Summit@ORNL.

For the strong scaling how do I pick the parameters for strong scaling, including:

NEX_XI
NPROC_XI (several values for strong scaling?)
MODEL (what model to pick?)
RECORD_LENGTH_IN_MINUTES (how long should the simulation be to ensure the stability of benchmark results)

From my previous experience, it is a bit tricky to pick the problem size and GPU numbers. Because using least GPUs may blow out the GPU memory.

SPECFEM compilation failed on Summit

Hi Daniel,

The compilation no longer works on Summit. The errors shows as:

/sw/summit/xl/16.1.1-1/xlf/16.1.1/bin/.orig/xlf2008_r: 1501-295 (W) The -qstrict option is unsupported by IPA Link. The strict settings defined in the compilation unit will be used and processing continues without this option.
1586-403 (U) Library file "cudart" not found.
make: *** [bin/xspecfem3D] Error 1
compilation failed, please check...

I am not sure what happens becuase before the New Year I can still compile the code on Summit...

Do you have any idea?

Recommend Projects