Skip to content

WeeklyTelcon_20210817

Jump to bottom
Geoffrey Paulsen edited this page Aug 17, 2021 · 1 revision

Open MPI Weekly Telecon ---

Attendees (on Web-ex)

  • Austen Lauria (IBM)
  • Brendan Cunningham (Cornelis Networks)
  • David Bernholdt (ORNL)
  • Edgar Gabriel (UH)
  • Geoffrey Paulsen (IBM)
  • Hessam Mirsadeghi (NVIDIA))
  • Marisa Roman (Cornelius)
  • Matthew Dosanjh (Sandia)
  • Raghu Raja
  • Sam Gutierrez (LANL)
  • Todd Kordenbrock (Sandia)
  • Tomislav Janjusic (NVIDIA)
  • William Zhang (AWS)

not there today (I keep this for easy cut-n-paste for future notes)

  • Akshay Venkatesh (NVIDIA)
  • Artem Polyakov (NVIDIA)
  • Aurelien Bouteiller (UTK)
  • Brandon Yates (Intel)
  • Brian Barrett (AWS)
  • Charles Shereda (LLNL)
  • Christoph Niethammer (HLRS)
  • Erik Zeiske (HPE)
  • Geoffroy Vallee (ARM)
  • George Bosilca (UTK)
  • Harumi Kuno (HPE)
  • Howard Pritchard (LANL)
  • Jeff Squyres (Cisco)
  • Joseph Schuchart (HLRS)
  • Josh Hursey (IBM)
  • Joshua Ladd (NVIDIA)
  • Mark Allen (IBM)
  • Matias Cabral (Intel)
  • Michael Heinz (Cornelis Networks)
  • Nathan Hjelm (Google)
  • Naughton III, Thomas (ORNL)
  • Noah Evans (Sandia)
  • Ralph Castain (Intel)
  • Scott Breyer (Sandia?)
  • Shintaro iwasaki
  • Xin Zhao (NVIDIA)

New Topics For Today

  • Over the weekend and even last night.
    • Coverity hasn't been running for months. (Something broken on Coverity for months and months)
    • Not sure what's changed, but when Coverity runs, our AWS instances now runs out of memory.
    • Coverity runs every time there is a new master tarball.
    • Jeff and William are discussing.
    • Can we discuss AWS instance
    • master and release branches are built.
    • Jeff just restarts jenkins
  • One of Austen's PRs are failing in AUTHORS.
    • Just run contrib/make_authors and see what happens.
  • nvidia get mtt for master / v4.1, but will begin enabling v5.0
  • Remind everyone of the Annual Open MPI committer's review.

v4.0.x

  • Schedule: milestone is set for September for 4.0.7
  • Should PR9210 - be PRed back to v4.0.x
  • Nothing special to discuss this week

v4.1.x

  • Schedule: mileston is currently (August) for 4.1.2 acculated bugfixes.
  • William did a PR 9155 for master for libfabric API, and want this in v4.1 branch.
  • Nothing special to discuss this week

v5.0.x

  • PMIx v4.1.0 released
  • PRRTE v2.0.0 released
    • Geoff will check that our submodule pointers are updated correctly.
  • Went over the Github Project of [critical v5.0.x issues|https://github.com/open-mpi/ompi/projects/3]
  • Proposed rc1 on Sept 24th, 2021
    • No discussion, so we're going forward with that.
    • Worried that TCP OSC blockers might not be fixed in time, so will put in code to issue an error at runtime to prevent getting into those paths, and document it heavily.

Master

Documentation

  • No update

MPI 4.0 API

  • Howard showed off new [Open MPI 4.0 API Compliance Github Project|https://github.com/open-mpi/ompi/projects/2]
    • This is not yet the FULL list of things for 4.0, Howard is still working through minutes.
  • Please use MPI-4.0 label
  • Error handling changes are done, and in the Done catagory of
    • Random Notes about don't need to do something, or a way our implementation is already working correctly.
    • Comm Dup, and non-propagation of info.
  • MPI_T Events - someone needs to review
  • Sessions, just about to remove the WIP label.
    • Can leave it, or squish it, please comment on PR.
    • No intention to merge back to v5.0.0
  • Do need people to review oustanding PRs in the "In Progress" column of MPI 4.0
    • MPI_T events 8057
    • mpi_minimum_alignment info 7985
  • All of the issues in the MPI 4.0 Compliance
  • The To-Dos, some are big and some are small.
    • MPI_INFO_GET string is small, and infrastructure is small
    • Isendrecv and Isendrecv_replace should be easy
  • embiggening is big but not hard.
    • A lot of files will be impacted up at the top, but not much internally.
    • Important in next few years.
  • MPI_Comm_Idup_with_info - small one.
  • MPI_Cancel - should add deprecation warning
    • Tricky because it's only deprecated for sends. So can't just do what we did with headerfiles.
      • Could make the message in the header call out only send requests. If warning at compile time is happy enough.
      • Or might need runtime warning.
  • MPI_Sizeof should be deprecated
    • This is only for Fortran, and need to find a way at compiletime Particularly at F08
    • Geoff will see if XLF has something like this.
  • Can't gaurantee didn't miss anything with MPI Forum changelog.
  • Should use this these projects going forward.
  • What's intent of MPI-4.0 Label?
    • We'll keep using that for now, but in future may just use projects.

MTT

  • Hiccup yesterday. May have some data-loss from yesterday.
    • LANL - still having issue with cherry-pi
    • If you're posting MTT data but don't see it there, let Jeff or Howard Know.
  • NVIDIA will investigate to get v5.0 and master back online

Longer Term discussions

  • No discussion.
Clone this wiki locally