WeeklyTelcon_20161122

Open MPI Weekly Telcon

---

• Dialup Info: (Do not post to public mailing list or public wiki)

Attendees

• Ralph
• Howard
• Josh Hursey
• Josh Ladd
• Sylvain Jeaugey
• Tood Kordenbrock

---

Agenda

Review 1.10.x: v1.10.5

• All issues and pull requests for v1.10.x
• 1.10.5
  • Nothing of note

Review 2.0.x: v2.0.2

• All issues and pull requests for v2.0.2
• Desired / must-haves for v2.0.x series
• Known / ongoing issues to discuss
  • #2411 STAT Debugger:
    • PR #2425
    • STAT is still not connecting correctly with the application. Need more details
    • STAT relies on ORTE to launch their daemons. Possible that STAT is not catching one of the debugging symbols.
    • STAT is (very) difficult to install, so testing is a challenge.
      • Due to this challenge and since it is a problem for all branches, this is not longer a blocker for v2.0.2
  • #2442 coll/libnbc race condition
    • needs review
  • #2424 io/ompio MCA parameter
    • Discussion pending
  • Release: Probably sometime in Dec. 2016

Review 2.x: v2.1.0

• All issues and pull requests for v2.1.0
• Desired / must-haves for v2.1.x series
• Known / ongoing issues to discuss
  • PMIx 1.2.0: status?
    • Waiting on Issue #144 to be resolved. There is PR #217 pending review to fix this issue. Needs review/testing.
    • Once that fix is in then we need to measure the memory footprint again.
  • Performance Issue #1831
    • PSM2 seems to be impacted more than others. Intel to take another look.
    • Both master and v2.x branches are giving poorer performance compare to the v1.10 branch.
• Probably release early 2017 (Jan.), after the v2.0.x release

Question to user community: should we make a v2.2.x?

• This question posed to the user community in the BOF
• We were requested to make two lists:
  1. Features that we anticipate we could port to a v2.2.x release
  2. Features that we anticipate would be too difficult to port to a v2.2.x release
• Here are the lists as discussed on the con-call:
  1. Features that we anticipate we could port to a v2.2.x release
     • Improved collective performance (new “tuned” module)
     • Enable Linux CMA shared memory support by default
  2. Features that we anticipate would be too difficult to port to a v2.2.x release
     • THREAD_MULTIPLE improvements for MTLs (not sure about this being difficult)
     • Revamped CUDA support
     • PMIx 3.0 integration (actually rhc doesn't think this would be hard but questions whether we should do it if PMIx 3.0 functionality is not being used)
     • MPI_ALLOC_MEM integration with memkind
     • OpenMP affinity / placement integration
  3. Features that are not in master yet but might come in soon(ish) that would make keeping the branches in sync difficult.
     • ULFM changes
     • New CUDA features

Master review

• N/A

SPI and Software Freedom Conservancy

• We have been officially invited to SPI
• Ralph's new information about the two organizations.
• Will try to invite representatives from SPI and SFC to a teleconf so folks can ask questions.
  • Will discuss this more next week.

Review Master MTT testing (https://mtt.open-mpi.org/)

• Not seeing Morning MTT reports, or tarball generation email or coverity.

MTT Dev status:

• Not getting morning MTT result emails. Jeff looked into that last week, and went back and forth with Brian.

Open MPI Developer's Meeting

• Week of Jan 23, 2016
• Meeting-2017-01

---

Status Update Rotation

1. LANL, Houston, IBM
2. Cisco, ORNL, UTK, NVIDIA
3. Mellanox, Sandia, Intel

---

Back to 2016 WeeklyTelcon-2016