Skip to content
James Dinan edited this page Dec 2, 2021 · 1 revision

12/1/2021 -- HACC WG

Attendees: Jim Dinan, Joseph Schuchart, Andrew W., Guillaume Mercier, Ken Raffenetti, Maria Garzaran, Ryan Grant, Whit Schonbein, Scott Levy, Jaeger J.

Continuations

  • Open question on how we integrated it with sessions
    • Contacted sessions WG about world model being a session, no response yet

Multithreading and logically concurrent

  • Present an update on WG discussion at the MPI Forum meeting next week
  • We have two "camps" and a broader solution space
  • Need feedback from implementor and application developers

MPI_Pbuf_prepare

  • Working through synchronization vs nop cases, not yet ready for another plenary discussion

CUDA Bindings for partitioned communication

  • To present at the MPI Forum meeting
  • Make sure to set the context that we intend to target all platforms and work toward MPI goals of portability and interopability. We took CUDA as a well known starting point and will expand to include Sycl, OpenMP, HIP, etc.
Clone this wiki locally