-
Notifications
You must be signed in to change notification settings - Fork 859
WeeklyTelcon_20200324
Geoffrey Paulsen edited this page Mar 24, 2020
·
1 revision
- Dialup Info: (Do not post to public mailing list or public wiki)
- Geoffrey Paulsen (IBM)
- Jeff Squyres (Cisco)
- Austen Lauria (IBM)
- Brian Barrett (AWS)
- Brendan Cunningham (Intel)
- Erik Zeiske
- Geoffroy Vallee (ARM)
- George Bosilca (UTK)
- Harumi Kuno (HPE)
- Howard Pritchard (LANL)
- Josh Hursey (IBM)
- Joshua Ladd (Mellanox)
- Matthew Dosanjh (Sandia)
- Michael Heinz (Intel)
- Noah Evans (Sandia)
- Ralph Castain (Intel)
- Scott Breyer (Sandia?)
- Shintaro iwasaki
- William Zhang (AWS)
- Todd Kordenbrock (Sandia)
- Thomas Naughton (ORNL)
- Akshay Venkatesh (NVIDIA)
- Joseph Schuchart
- David Bernhold (ORNL)
- Artem Polyakov (Mellanox)
- Edgar Gabriel (UH)
- Nathan Hjelm (Google)
- Charles Shereda (LLNL)
- Brandon Yates (Intel)
- Mark Allen (IBM)
- Matias Cabral (Intel)
- Xin Zhao (Mellanox)
- mohan (AWS)
-
MTT -
- If you change your MTT to startup PRRTE at begining of session, and just use prun.
- Can see times cut in half or more.
- This is good, but also need to test mpirun wrapper.
- Cisco is converting half of MPI installs to use prrte/prun
-
OMPI master submodule pointers setup to track PMIx and PRRTE master.
- Jeff discussed an idea to have some integration with PRRTE that putting a string in a PRRTE PR would automatically open an Open-MPI PR to update the PRRTE submodule after that PRRTE PR is merged to PRRTE master.
Blockers All Open Blockers
- v3.0.6 and v3.1.6 branches Released!
- Raising the bar on another release.
- Have pulled a few PRs to the branch
Review v4.0.x Milestones v4.0.4
- v4.0.4 in the works.
- will set arbitrary 6 monthish milestone for now.
- No Schedule yet.
- Jeff is looking at PMIx issue, some issue with dstore working with Ralph.
- Error message that turns off - GEOFF to follow up to ensure we have PR on v4.0.x
- We may need a new PMIx v3.1.x release.
- Update 3/23: Until we hear a problem, we won't backport a PR or ship another PMIx v3.1.x
- It's a bug, but may not expose itself unless you direct launch.
- There's a one-liner code fix needed for SLURM with > ppn64
- This may drive a v4.0.4
- Right in MPI_Init (not related to specific component)
- Schedule:
- Feature Freeze: April
- Release: End of June
- Austen took an initial stab at issues and is starting a google sheets of v5.0 features.
- Today we went through all of the items on the google sheets document (https://docs.google.com/spreadsheets/d/1OXxoxT9P_YLtepHg6vsW3-vp4pdzGQgyknNbkzenYvw/edit#gid=0) which were taken from the face to face wiki.
- Josh Ladd led us to gather owners and a status for each of the various tasks.
- Updated status in above google sheets.
- PMIx v4.0.0 - on track
- PRRTE v2.0 - on track
- Remove OSC pt2pt - need some work from PSM2/OFI before we can remove. - at risk
- Discussed Multithreaded Framework
- Concerns about some non-posix implementations and MPI progress in general.
- see https://github.com/open-mpi/ompi/pull/6578
- Consensus that we want the framework / reorganization (using pthread as default)
- Will address a few other PR specific issues before merging.
- Greater progress issues in various components can be discussed in the future.
- Static linking is failing on master right now.
- Issue 7560
- May be an issue in static build support in PMIx and PRTE as well as how we're pulling it in.
- Affects everything, just masked at the moment because static linking is broken.
- Jeff will investigate
- SLURM PMIx plugin has been locked on PMIx v2 for some time.
- There are some NEW PMIx calls that SHOULD be added to bring it up.
- Ralph has started a PR, but needs help.
- So for now, there's some optional info that won't be passed correctly.
- No OMPI_INFO for now.
- Ralph gets pinged occasionally.
- Not sure priority of this.
- There are some NEW PMIx calls that SHOULD be added to bring it up.
- MTT on master is looking pretty good.
- Defered.
- scale-testing, PRs have to opt-into it.
Review Master Master Pull Requests
- CI testing only tests build and did it run, but doesn't test HOW it ran.
- Environment setup can be a bit different.
- For example no-permissions in
/tmp
. Might pass on one machine, and fail on another without/tmp
permissions.