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_2016_01_13 – NEMO
wiki:WorkingGroups/HPC/Mins_2016_01_13

Version 2 (modified by mikebell, 8 years ago) (diff)

--

Minutes Draft notes from NEMO HPC group meeting 14:00 – 15:30 (CET) on Wed 13 January 2016 =

Attending: Mike Bell (chair), Jeremy Appleyard, Tom Bradley, Miguel Castrillo, Mondher Chekki, Maff Glover, Tim Graham, Claire Levy, Silvia Mocaverro

Apologies; Andy Porter, Marie-Alice Foujols, Jason Holt, Martin Schreiber, Julien le Sommer,

  1. New members: Jeremy Appleyard (Invidia) introduced himself.
  1. Actions from previous meetings

Action 1: Tim and Sylvia to discuss what specification of this [roofline diagnostics] would be useful and then iterate with Martin, Oriol and Miguel. Done.

Action 2: Miguel, Silvia, Tim and Claire to coordinate the set up and review of these two configurations. Good progress but left open.

Action 3: Mike to put information from group meeting on the NEMO HPC WG wiki page. Not done yet.

Action 4: Mike to set up an email circulation list (to which all members can send emails). Done.

Action 5: Mike to set up a doodle poll for the next meeting between 11 and 22 Jan. Done.

  1. Specification of roofline diagnostics

We walked through the draft specification of roofline diagnostics.

The environment variable settings need to be monitored. There is a new tool within NEMO for listing this information.

It may be necessary to compare performance using the best choice of setting for each machine used. IO performance could also be examined.

Action: Mondher will email Silvia, Miguel, Tim and Mike information on computers accessible to Mercator.

  1. Implementation of benchmark configurations

Three benchmarks are being set up: ORCA025-LIM3 and two GYRE configurations

It would be easy to have additional GYRE configurations (it just needs small NAMELIST changes); this would broaden the range of roofline tests

The best version of NEMO to use for each benchmark needs to be decided.

Action: Claire, Tim, Silvia, Miguel to agree what version of NEMO to use for these benchmarks

  1. H2020 Copernicus call

A proposal to develop NEMO for CMEMS (the Copernicus Marine Environment Monitoring Service) in response to a Horizon 2020 call is being developed. Mike is leading the proposal. The money available is limited to about 1.5 M€ EU contribution. A small task will be included to explore the roofline diagnostics outlined above and build the evidence base for an HPC strategy for NEMO. The groups involved in this task will be CMCC, BSC and Met Office. The scope of the proposal is still being discussed. The present WP structure is not the final version. Letters of support from other group members would be valuable.

Action: Mike to circulate the present draft of the HPC task description and the WP structure. Members to respect the confidentiality of this information.

  1. NEMO HPC working document and role of this group – what to do next with them ? (20 mins)

It was agreed to re-visit the HPC working document in a couple of months time.

It is good that the group has started to work on some concrete and useful actions.

  1. AoB (10 mins)

Action: Mike will set up a doodle poll for a meeting in the first or second week of February (in time to influence the H2020 proposal)

Open Actions

Action 1: Miguel, Silvia, Tim and Claire to coordinate the set up and review of these two configurations. Good progress but left open.

Action 2: Mike to put information from group meeting on the NEMO HPC WG wiki page. Not done yet.

Action 3: Mondher will email Silvia, Miguel, Tim and Mike information on computers accessible to Mercator.

Action 4: Claire, Tim, Silvia, Miguel to agree what version of NEMO to use for these benchmarks

Action 5:Mike to circulate the present draft of the HPC task description and the WP structure. Members to respect the confidentiality of this information.

Action 6: Mike will set up a doodle poll for a meeting in the first or second week of February (in time to influence the H2020 proposal)