2025.01.23 Meeting Notes

Agenda

• Individual/group updates
• Review non-WIP PRs

Updates

BP

• opened several issues to clarify interfaces
• Updating downstream interfaces wrt using md interface to get non-base MeshBlockData, see https://github.com/parthenon-hpc-lab/parthenon/issues/1220
• discovered that there's a difference between integrator.dt and tm.dt
  • will look into cleaning this up (potentially making Integrator const)
  • potentially also looking into reducing/reporting different dts
  • would also be useful in other downstream codes like FLASH

AR

• got a new position (with continued Flash/Parthenon development)

FG

• working on getting new tools on venado inplace

PG

• AthenaPK is JUPITER lighthouse application with MHD turbulence
• Still experiencing issues on Frontier at large scale wrt performance variablity even with the latest stack
  • need to isolate potential environment variables
• Similarly, for unknown reason a binary that was working fine last year (compiled in Dec24) is now causing issue writing large output files. Root cause again unclear.
• Are dynamically created/destroyed particle uid unique throughout a simulation?
  • Maybe. Need to check. Required downstream.

Open PR

• https://github.com/parthenon-hpc-lab/parthenon/pull/1214 PG will check status
• https://github.com/parthenon-hpc-lab/parthenon/pull/1196
  • could in principle go in, but the overall approach could be improved wrt index splits
  • BP will give it a shot
• https://github.com/parthenon-hpc-lab/parthenon/pull/1192
  • ping people on what to do
• https://github.com/parthenon-hpc-lab/parthenon/pull/1182
  • is good to go
  • PG will trigger pipelines and then merge
• https://github.com/parthenon-hpc-lab/parthenon/pull/1142
  • FG found that slowdown is related to increased register usage and will do some further analysis
  • might be related to indexer
  • check it's related to the updated inner bounds or more general

Tentative next meeting 6 Feb