Agenda

• Individual/group updates
• Discuss/share current threading/concurrency model (pgrete)
• Usage of FORCEINLINE macro (pgrete)
• Quick run-through of outstanding non-WIP pull request reviews

Goals

From two weeks ago for this meeting:

• Discuss offline the best strategy for 1) doing performance tests and 2) automating them
• Start Profiling Code w/o AMR
  • F. Glines
  • J. Miller (~a little bit)
• Start Kokkosifying AMR
  • Assignment of execution spaces to mesh blocks (necessary for performance)
    • P. Grete
  • Porting for functionality (just get things not crashing and working correctly)
    • J. Dolence
    • A. Gaspar

Next Two Week Goals

• tbd

Updates

Athena++/K-Athena

Phil

Got the Kokkosification change merged.

Forrest

Testing GPU performance. 100x worse performance for small mesh block sizes (16^3) in uniform static grid. https://github.com/lanl/parthenon/issues/189

Kokkos AMR

Joshua and Andrew started porting kokkos AMR code: https://github.com/lanl/parthenon/pull/177 Tests:

• calculate_pi + CPU
• calculate_pi + GPU + serial
• calculate_pi + GPU + parallel
• advection + GPU + serial
• advection + GPU + parallel

Joshua Brown

In progress implementation of code coverage: https://github.com/lanl/parthenon/pull/187

Discussion

Concurrency

Discussion of our concurrency strategy.

KOKKOS_FORCEINLINE_FUNCTION

We think this is maybe causing issues for IPO with Intel. Need to check. Phil will verify that it's actually FORCEINLINE causing the issue.