Skip to content

2026 05 21 meeting minutes

Jump to bottom
Carl Pearson edited this page May 21, 2026 · 1 revision

Attendees: Matthew, Nicole, Evan, Gabriel, Hugo, Kaelyn, Carl

General Topics

  • (Nicole) Camera-ready PASC submission in

    • ACM rights review forms / etc in progress

  • 0.1 release

    • Some key additions are in
      • MPI-backend exposed in the core API
      • Refactored Requests: shared API between backends, easier to write CommSpace-generic code (doc up-to-date)
      • Refactored Communicators: shared API between backends, new capabilities: split, duplicate (doc up-to-date)
      • DeepCopy packing strategy can now handle Views of arbitrary rank
      • Collectives don't have limitations on View ranks anymore (doc specifies semantics)
      • Cleaned up Kokkos Comm traits (doc up-to-date)

Changelog (post 0.1)

  • Maybe Cedric will give this a try, follow up after his absence
  • changelog.md, modified when we merge a PR
    • When you open a PR, you get a number -> add a commit to your PR that adds the PR to the changelog
  • Items postponed to 0.2 (end of June, PASC):
    • static_assert on GPU buffers in MPICH
    • fix MPI detection in cray-mpich
      • Workaround at configure time to say that MPI is MPICH
    • Non-GPU-aware MPI host staging (Gabriel)
      • critical to have before PASC
    • Spack package recipe (Cedric)
      • better to have 0.1 released anyway before writing the recipe
    • NCCL and MPI results from the same code (Gabriel)
      • OSU-style latency benchmark (MPI + NCCL enabled at the same time)
        • Gabriel: some overhead from fence in MPI backend (hopefully not present in stream-triggered backend)
      • Nicole has the Heat3D benchmark in her fork -> open a PR
    • Blog post (Carl)
    • Fixing documentation (Carl)
      • Cedric opened #201
      • Things have been partially addressed in multiple PRs that have been merged recently
    • miniFE port (Carl)
    • Lulesh port (Gabriel + minions)
      • port is working, but not 1-1 of existing code (some refactor, etc)

Carried forward from last time

  • Updates on CI dev container? (Cedric)

    • Waiting for build system rework from Gabriel

  • MPICH (via Spack) + CudaSpace views, exclusive scan fails

    • SNL CI only tests Open MPI + CUDA
      • Carl will make a container
    • Not sure if it's an MPICH problem or something else we're doing wrong
    • Gabriel is splitting up the unit test

Round Table

  • Carl

    • Larger message size in the MPI + HIP smoke test to catch errors better

  • Evan

    • stream triggered MPI + Kokkos Comm @ SNL this summer
    • Kokkos Comm + stream-triggered MPI on Tuolumne at least somewhat working
    • Working on CMake: trying to follow along with NCCL, but not quite working yet
      • may share in channel to get help

  • Nicole

    • PASC paper and presentation

  • Gabriel

    • did the release stuff
    • want to look at RCCL PR
    • expecting to work on host staging next

Clone this wiki locally