Skip to content

WeeklyTelcon_20210706

Jump to bottom
Tomislav Janjusic edited this page Jan 6, 2024 · 1 revision

Open MPI Weekly Telecon ---

Attendees (on Web-ex)

  • Austen Lauria (IBM)
  • Aurelien Bouteiller (UTK)
  • Brendan Cunningham (Cornelis Networks)
  • Brian Barrett (AWS)
  • Harumi Kuno (HPE)
  • Hessam Mirsadeghi (NVIDIA))
  • Howard Pritchard (LANL)
  • Jeff Squyres (Cisco)
  • Josh Hursey (IBM)
  • Joseph Schuchart (HLRS)
  • Naughton III, Thomas (ORNL)
  • Samuel Gutierrez (LANL)
  • Todd Kordenbrock (Sandia)
  • Tomislav Janjusic (NVIDIA)

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

  • Akshay Venkatesh (NVIDIA)
  • Artem Polyakov (NVIDIA)
  • Brandon Yates (Intel)
  • Charles Shereda (LLNL)
  • Christoph Niethammer (HLRS)
  • David Bernholdt (ORNL)
  • Edgar Gabriel (UH)
  • Erik Zeiske (HPE)
  • Geoffrey Paulsen (IBM)
  • Geoffroy Vallee (ARM)
  • George Bosilca (UTK)
  • Joshua Ladd (NVIDIA)
  • Matthew Dosanjh (Sandia)
  • Michael Heinz (Cornelis Networks)
  • Marisa Roman (Cornelius)
  • Mark Allen (IBM)
  • Matias Cabral (Intel)
  • Nathan Hjelm (Google)
  • Noah Evans (Sandia)
  • Raghu Raja
  • Ralph Castain (Intel)
  • Scott Breyer (Sandia?)
  • Shintaro iwasaki
  • William Zhang (AWS)
  • Xin Zhao (NVIDIA)

New Items

v4.0.x

  • One IOF PR against v4.0.x. #9119. Calls it an 'enhancement', Wonder if v4.0.x should take it - RM's will sync next week. It's a backport from prrte.

  • Issue #9123 'Crash in MPI_Win_lock_all when used with libfabric < 1.12' is new, effects several versions of v4.0 and v4.1, may need to look at that. OFI specific one-sided issue. Will want to see if it is fixed on master.

v4.1.x

  • Same IOF PR (#9118). RM's not sure why it is needed, will take a look.

  • New PR #9114 - 'Skip SLURM provided PMIx detection when appropriate'. RM's would like to look at it before merging, but thinks it should be fine.

  • #9093 - 'Long live MPI_LONG/UNSIGNED_LONG PR' still pending, will merge since it went into v4.0.x.

  • #8981 'Remove unnecessary dependencies to ORTE' is still in draft, Howard still looking at it.

v5.0.x

  • Went over the blockers. Two PR's currently open against v5, will meet Thursday to go over/merge them. Link to blockers posted in OMPI Slack.

Master

  • No discussion.

Other issues

NAG compiler master issues (PR #6378):

  • If using an older Libtool, get some warnings with the NAG PR on master. Appears to be calling a function that doesn't exist. Doesn't seem to error on the build, but throws warnings.

    Leaving it in 'feels better' for now, will open an issue so that someday it will hopefully get fixed. Probably should only apply this on Libtool versions v2.4.6 and higher.

Issue #9120 OSC performance regression from v4.1.1 to master (OFI)

  • Seems strange, last checked they were identical. Is it a btl problem, or an osc/rdma problem? Someone with portals could check to see if they see a similar issue. However, no available systems currently.

PMIx

  • No discussion

PRRTE v2.0

  • No update

Longer Term discussions

  • No discussion.

Reminder

Clone this wiki locally