[Mpi-forum] Reading for Buffer Readiness on Partitioned Communication for Accelerator Optimization - Issue #302

[Grant, Ryan Eric (-EXP)]

Hi Everyone,

Apologies if you received this twice, my previous email seems to have not gone out to the list.

We would like to announce the following reading for the next MPI Forum meeting:  Synchronization on Partitioned Communication for Accelerator Optimization - Issue #302. This introduces explicit buffer readiness knowledge to allow for optimization on accelerator (e.g. GPU) type devices. This is an updated version of the ticket that was read at the last forum meeting.

This ticket addresses the issues discussed in the virtual forum meeting (non-voting) on July 29th (link to recording: https://youtu.be/__y9yLCVN98 part comm discussion begins at 49:30​).

Issue: https://github.com/mpi-forum/mpi-issues/issues/302
Pull Request: https://github.com/mpi-forum/mpi-standard/pull/264
Diff in github: https://github.com/mpiwg-persistence/mpi-standard/compare/1caba35..a3a8d36

Thanks,
Ryan
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mpi-forum.org/pipermail/mpi-forum/attachments/20200915/965f6a26/attachment.html>