2022.05.05 Meeting Notes

Agenda

• Individual/group updates
• Call with Intel
• Parthenon developer meeting

Individual/group updates

LR

• debugging sparse bug https://github.com/lanl/parthenon/issues/668
• there are still some small differences compared to current develop, which are also in the process of being tracked down

BR

• added parallel_scan abstraction
• finalized removal of SetPrecise
• will take care of Darwin performance test failing (limit to 1k blocks)

GS

• working on putting together benchmarks
• will use advection example (including one with random work)
• eval of Parthenon can serve as a proxy/miniapp

FG

• will start looking into face centered fields
• potentially derive from AthenaK impl

JS

• Have GR Rad Hydro up and running with mesh refinement
• Radiation implemented as cell centered variable -> should easily map to other downstream codes
• Looking forward to compare against impl in Phoebus

PG

• Replaced phys_id tags with MPI comms
• Sparse debugging (on Spock and locally)
• Updated clang-format action
• Has access to Crusher now

Call with Intel

• took place last Monday
• key results:
  • Intel contact will double check/ensure everyone is under NDA so that we can get access to corresponding hardware
  • Intel contact offered to be point of contact to forward issues/requests to the engineers if we encounter problems
• Will have follow-up meeting in 2 weeks

Parthenon developer meeting

• no news, need to ping JD/JMM for update next time

Next meeting 19 May