Published October 18, 2020 | Version v1

Scalasca summary analysis of HemeLB application execution with 300000 MPI processes on SuperMUC-NG

Authors/Creators

  • 1. Jülich Supercomputing Centre

Description

The CompBioMed HPC CoE flagship application HemeLB was run with a 6.4 micron resolution "circle of Willis" geometry dataset on LRZ's SuperMUC-NG supercomputer, and its execution performance with 300000 MPI processes on 6250 dual 24-core compute nodes measured by Score-P (including PAPI hardware counters) and analysed by Scalasca.

Two compute nodes with poor memory performance can be distinguished by examining the PAPI_RES_STL metric for resource stall cycles.

Files

Files (1.2 GB)

Name Size
md5:3ab2109115b120ffee17c5212262a963
1.2 GB Download

Additional details

Related works

References
Other: 10.5281/zenodo.2605026 (DOI)
Other: 10.5281/zenodo.3885304 (DOI)
Other: 10.5281/zenodo.4103923 (DOI)

Funding

European Commission
CompBioMed - A Centre of Excellence in Computational Biomedicine 675451
European Commission
POP2 - Performance Optimisation and Productivity 2 824080