2024.06.20 Meeting Notes

Agenda

• Individual/group updates
• Additional automated tooling
• Hackathon
• Tutorial
• Review non-WIP PRs

Individual/group updates

JM

• downstream develop
• will have student to work on implicit hydro solver
  • will either use parthenon-hydro or parthenon-mhd (eventually making the code public)

LR

• got bunch or PRs merged
• working on MG support >1 partitions
  • also includes block partitions living in Mesh
  • discovered inconsistencies between iterative tasks and partitions
    • should probably extend test coverage for pack_size != -1
• suggest a cleanup if interfaces
  • everyone supports this!

PM

• Q: add breaking change to history output to include meshblock count
  • general agreement and even more general interest in additional information
  • try to reuse hist output to add performance counters (in the future)
• for protoplanetary disks application
  • multiple coord systems
    • got refinement for free due to defining flux corr and prolong/restrict
  • have field that are WithFluxes but no Independent or ForceRemesh (so Refined is not set by default) -> will fix this

BP

• looking into partitioning to do (different) work on different (pieces) of the domain
  • e.g., to use different timescales in different parts of the domain
  • just skipping work in different parts (but keep communicating)

AR

• in the process of porting more code

PG

• caught up on reviews
• fixed bug for restarting
• AthenaPK now working with parthenon dev again
• more debugging on Frontier and Lumi (issues are open)
• in the process of setting up CI on AMD GPUs
• adding athenapk to exacb (continuous benchmarking environment for JUPITER)

Review non-WIP PRs

• https://github.com/parthenon-hpc-lab/parthenon/pull/1119 should get downstream reviews
• https://github.com/parthenon-hpc-lab/parthenon/pull/1047 AD will do second review

defer next meeting until after hackathon