New URL for NEMO forge!   http://forge.nemo-ocean.eu

Since March 2022 along with NEMO 4.2 release, the code development moved to a self-hosted GitLab.
This present forge is now archived and remained online for history.
WorkingGroups/HPC/Mins_sub_2017_08_30 – NEMO
wiki:WorkingGroups/HPC/Mins_sub_2017_08_30

Version 2 (modified by mocavero, 6 years ago) (diff)

--

NEMO HPC subgroup: Wed 30 Aug 2017

Attending: Mike Bell (Met Office), Tim Graham (Met Office), Miroslaw Andrejczuk (Met Office), Andy Porter (STFC), Miguel Castrillo (BSC), Alicia Sanchez (BSC), Marie-Alice Foujols (CNRS), Silvia Mocavero (CMCC)

1. Single-core performance

Tim tested single-core performance on Broadwell nodes by running single-instance, one-instance per socket and multiple-instances per socket and by measuring FLOPS and L3 cache misses. Some areas of the code have a different behaviour when single or multiple instances per socket are executed and have to be deeply investigated. Other tests could be useful (e.g. changing the domain size, without wrk_allocate, with cache blocking). Mondher is performing tests on GENCI and Mercator machines.

Action: Tim to share results on Broadwell

2. Hybrid parallelization status

Silvia is working on ZDF coarse-grained version to increase the number of tiles per thread. Moreover, she is working to port the OpenMP fine-grained version on CMCC high-resolution NEMO configuration (at 1/16°). Mondher is testing the ZDF package on GENCI and Mercator systems and Matt performed some preliminary tests on OpenMP scalability on KNL.

Action: Silvia to continue to work for improving the coarse-grained version. Matt to share the preliminary results on KNL.

3. Next meeting call

Next meeting will be the in the second half of October.

Action: Silvia to send the doodle poll for the next meeting.