[mpiwg-hybridpm] HACC WG Meeting this week: stream-triggered operations

[Patrick Bridges]

Thanks, Joseph. Here is the comment describing the state of this proposal that highlights just some of the issued to address. This PR is to the hybrid working group repository’s fork of the standard. I’ve also directly attached the relevant chapter cas a latex error I recently introduced and just fixed prevented the previous version from actually including the new queued operations chapter.


This PR is an initial draft of a new chapter on MPI queues for supporting stream-triggered operations through a "queued communication" interface. The goal of this PR is to begin discussions on this topic. The current draft is not complete. Current issues include:

     *   It includes text in the new chapter (chap-streams.tex) that will eventually need to be moved to other chapters. The idea is to keep these changes in the stream triggering chapter while we work out semantic details for the PR, and then move them to the appropriate other chapters.
        *   It introduces a new //matched// semantic term for persistent operations that will need go into the semantic terms chapter.
        *   It will include a new state for persistent operations (enqueued) which is neither active nor inactive that will need to go in to persistent operations chapter
     *   It includes an MPI_Match operation that impacts the semantics of persistent operations that will need to go in persistent operations chapter.
     *   It does not yet include discussion or a definition of a needed new state for persistent requests (//enqueued// is neither inactive not active)
     *   It does not include changes to appendices for local/non-local for the new functions.
     *   It needs to carefully define in MPI semantic terms what a successful MPI_Match on a persistent collective means. Intuitively, it means that all participants in the collective have completed initialization of the collective so that we know the final buffers into which data will be placed, but the MPI semantic terms around persistent collectives are complex and need to be dealt with carefully.




On Apr 14, 2026, at 1:34 PM, Joseph Schuchart via mpiwg-hybridpm <mpiwg-hybridpm at lists.mpi-forum.org> wrote:


  [EXTERNAL]

This Message Is From an External Sender
This message came from outside your organization.

Dear all,

We'll meet this week Wednesday 10am EST to discuss the draft for
Patrick's proposal on stream-triggered MPI operations [1]. The
connection info is available at
https://urldefense.us/v3/__https://github.com/mpiwg-hybrid/mpi-standard/wiki__;!!G_uCfscf7eWS!dS7y_Ke_rkvEkd-8T0iPYHTxpS6VBF56O_fQuxo01RyNV7ok7JoDK3J5c6lmqHdCWQ4uBv94_8PpKMPUOU-aPZUin3ekLJbuMopkYIfeXM_thAzE$. Let me know if you
want to join but cannot access the page.

Cheers
Joseph


[1] https://urldefense.us/v3/__https://github.com/mpiwg-hybrid/mpi-standard/pull/6__;!!G_uCfscf7eWS!dS7y_Ke_rkvEkd-8T0iPYHTxpS6VBF56O_fQuxo01RyNV7ok7JoDK3J5c6lmqHdCWQ4uBv94_8PpKMPUOU-aPZUin3ekLJbuMopkYIfeXI3zG0rA$



_______________________________________________
mpiwg-hybridpm mailing list
mpiwg-hybridpm at lists.mpi-forum.org<mailto:mpiwg-hybridpm at lists.mpi-forum.org>
https://urldefense.us/v3/__https://lists.mpi-forum.org/mailman/listinfo/mpiwg-hybridpm__;!!G_uCfscf7eWS!cBVs5yFP5c8E8LBOB2VDrScpuS4HkrjyZFs4QgtKjcZwRIfee3b0tytMBy20hdX39VTDLT2WXEYEj31jYH7JZmIccH_6u-tF$ 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mpi-forum.org/pipermail/mpiwg-hybridpm/attachments/20260415/ca9d2cb8/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: mpi51-report.pdf
Type: application/pdf
Size: 589526 bytes
Desc: mpi51-report.pdf
URL: <http://lists.mpi-forum.org/pipermail/mpiwg-hybridpm/attachments/20260415/ca9d2cb8/attachment-0001.pdf>