forked from open-mpi/ompi
-
Notifications
You must be signed in to change notification settings - Fork 1
WeeklyTelcon_20160202
Jeff Squyres edited this page Nov 18, 2016
·
1 revision
- Dialup Info: (Do not post to public mailing list or public wiki)
- Edgar Gabriel
- Geoffroy Vallee
- Geoffrey Paulsen
- Sylvain Jeaugey
- Howard
- Joshua Hursey
- Joshua Ladd
- Brad Benton
- Ralph Castain
- Todd Kordenbrock
- George
- Milestones: https://github.com/open-mpi/ompi-release/milestones/v1.10.3
- Unless someone needs someone faster, next cycle will probably be April.
-
Issue 1323 - F08 bindings
- Fixed by master F08 bindings commit - PR 937
- 1.10 C Strided mutex lock issue. (Nathan) - ask again next week.
- Wiki: https://github.com/open-mpi/ompi/wiki/Releasev20
- Blocker Issues: https://github.com/open-mpi/ompi/issues?utf8=%E2%9C%93&q=is%3Aopen+milestone%3Av2.0.0+label%3Ablocker
-
Issue 1215 https://github.com/open-mpi/ompi/pull/1335: grpcomm errors
- Ralph is unable to replicate. Didn't see on Trinity and elsewhere at scale. Found where the problem is, but trying to figure out why solution isn't working. Ralph-and-Jeff-are-iterating phase.
-
https://github.com/open-mpi/ompi/issues/1252: bad perf caused by openib
- Nathan and Geoff WebX today?
-
Issue 1215 https://github.com/open-mpi/ompi/pull/1335: grpcomm errors
-
PR 927 - need a Ralph review
- (the X / test fail was due to github being down -- it's a false failure)
- Issue 1299 - Nathan Hang osc pt2pt.
-
PR 932
- Ralph Thinks it's okay, but worried it might fix this use case, but cause regression for other use cases.
- If any hardware thread on any core is allowed, then we should allow hardware threads. Should be okay.
- Howard will merge today.
-
PR 922
- This feels like an enhancement, and is likely to be deferred to v2.1.0 (we're feature complete for v2.0.0)
- Milestones: https://github.com/open-mpi/ompi-release/milestones/v2.0.0
- Jeff filed a PR 936 for F08 bindings commit
- Do we need a PR against 2.0 for Orte OOB fix Issue 1301
- Master commit: https://github.com/artpol84/ompi/commit/39bbfdda9cf257cb90fb2265471e2a4d2851dea8
- No, already Fixed on 2.0.
- OMPIO default issues (Edgar)
- Solaris wasn't building? Should be fixed. Went into 2.x branch.
- Still working on component, but not coming into 2.0. Maybe 2.0.1
- Indiana has changed open-mpi.org to 100% https.
- Geoffroy Vallee - Started looking at .dbm package. Supposed to support every platform .dbm supports.
- Just checking out dbm code is getting permission issue.
- At face2face in Oregon. Has new Python client.
- Python will require new ini file.
- New features - in .ini can specify a specific PR or branches.
- Each section tells you what key/values it will accept.
- completely plugable, so parsing of .ini file is a plugin.
- Going to start talking about new web based reporter.
- Josh will put new stuff he has locally pushed upto MTT site.
- has a monitoring system
- Have the ability for reporter to report that data.
- Want to be able to drill down to the monitoring data behind that test.
- Mellanox
- nothing much on our end, other than some PMI-x work.
- Artem has been doing a good job cleaning
- Sandia
- Been backed up on other projects. Not much new on OMPI. waiting for 2.0
- Intel
- Working on Group Comm problem. MTT re-write.
- DVM code - PR sitting out there for some significant changes.
- can execute 350 jobs / minute (0 time jobs). Launch and recover.
- created a new directory ORTE-bindings. Created Python bindings for ORTE. ORTE submit tool is put into library. So you don't keep opening a port if you're going to submit multiple jobs. Already asked for Python bindings for PMI-x
- PMI-x - Working group started up to look at Error responce APIs. So can negotiate with Resource manager.
- Error notification stuff for debuggers. Hope to get this done this week.
- Mellanox, Sandia, Intel
- LANL, Houston, HLRS, IBM
- Cisco, ORNL, UTK, NVIDIA