-
Notifications
You must be signed in to change notification settings - Fork 861
WeeklyTelcon_20161101
Jeff Squyres edited this page Nov 18, 2016
·
1 revision
- Dialup Info: (Do not post to public mailing list or public wiki)
- All issues and pull requests for v1.10.x
- 1.10.x
- Ralph committed a few things yesterday.
- 4 PRs still here, waiting for reviews (Jeff).
- Still no drivers for a 1.10.5.
- All issues and pull requests for v2.0.2
- Desired / must-haves for v2.0.x series
- Known / ongoing issues to discuss
-
#2234 COMM_SPAWN broken: status?
- Jeff confirmed that 2215 CID rewrite stuff fixes the issue for v2.x.
- Nathan to look at seeing if we can get a smaller version for v2.0.x.
- Maybe don't include Comm_Split_Type, just
- Report from Orion P: we need to fix COMM_SPAWN for v2.0.x
-
#1831: Communication rate degradation: status?
- Arm's graph doesn't show 2.0.x results. Will have to ask.
- MTL owners need to do similar test to compare, and see if this is a blocker for v2.0.x
- C++ compile error -- missing a PR from master (remove useless / duplicate commit). It's on v2.1.
- PR'ed the fix to v2.0.x. DONE.
- v2.0.2 schedule:
- Close bugs this Friday, Oct 28
- Aim to release Fri Nov 4
- Do we require IPoIB for the openib BTL?
- No, Nathan will reply to email.
- If so, we should document it.
- https://github.com/open-mpi/ompi/issues/2326
- Should we remove
--enable-btl-openib-failover?- @swise has volunteered to remove the code.
- https://github.com/open-mpi/ompi/issues/2228
- Yes, remove this, we don't have a PML that would support this.
- New issue from yesterday: neighborhood collectives
- https://github.com/open-mpi/ompi/issues/2324
- Neighbor gatherv and Neighbor Igatherv. - Giles asked for test case this morning.
- Yoda:
- If the fix is a one-liner, can we get it in v2.0.x?
- Looks like fragmentation. Yes. Josh Ladd's team is driving.
- Might push-back MPI_Dims of 0
- Might push-back https://github.com/open-mpi/ompi/pull/2250
-
#2234 COMM_SPAWN broken: status?
- All issues and pull requests for v2.1.0
-
Desired / must-haves for v2.1.x series
- Reviewed this today (Nov 1st)
- MPI-IO - no good MT test for MPI_IO - Edger would like this.
- Possible we'll see something for new coll_tuned component.
- Known / ongoing issues to discuss
- PMIx 1.2.0: status?
-
PR 2286: will update to PMIx v1.2.0rc1. Testing is looking good. Two outstanding issues -- both should be done this week:
- Update to Get
- One thing Boris is working on.
- Estimate release PMIx v1.2.0 this time next week.
- People please try it out!
-
PR 2286: will update to PMIx v1.2.0rc1. Testing is looking good. Two outstanding issues -- both should be done this week:
- PMIx 1.2.0: status?
-
PR #2285: enabling orte to use libfabric
- Please go test it!
- Uses RDM messaging
- @hppritcha would like to test, but will not be able to test until next week
SPI - http://www.spi-inc.org/
- On November Agenda (next week).
- Guy who heads up conservancy has been emailing Ralph.
- Ralph filled out Conservancy form as well (no obligation)
- Big difference between SPI and conservancy. SPI is all volunteer, Conservancy has 4 fulltime staff.
- SPI is still a lot of confusion. switched to use Conservancy financing.
- Ralph pointed out that at most they'll get like $100 from us (10% fee).
- Jeff proposed a nomenclature. Has not been implemented yet.
- 5:30 - Asked a number of people for content, and others to talk.
Review Master MTT testing (https://mtt.open-mpi.org/)
- Not seeing Morning MTT reports, or tarball generation email or coverity.
- 2.0.x series, still having some failures. Cisco has 2041 failures (1800ish is OSHMEM)
- LANL, Houston, IBM
- Cisco, ORNL, UTK, NVIDIA
- Mellanox, Sandia, Intel