[mpiwg-sessions] note from off-cadence WG meeting to discuss issue 435, etc.
schulzm at in.tum.de
Wed Feb 24 16:35:12 CST 2021
Hi Howard, all,
Strange, the webex issue – sorry we couldn’t connect.
Rolf and I had a discussion as well and we agree that PR48 looks good as is – Rolf’s only suggestion was to add the examples after that to make the text more clear. Not sure which PR he had that in, though.
Talk to you tomorrow,
Prof. Dr. Martin Schulz, Chair of Computer Architecture and Parallel Systems
Department of Informatics, TU-Munich, Boltzmannstraße 3, D-85748 Garching
Member of the Board of Directors at the Leibniz Supercomputing Centre (LRZ)
Email: schulzm at in.tum.de
From: mpiwg-sessions <mpiwg-sessions-bounces at lists.mpi-forum.org> on behalf of "Pritchard Jr., Howard via mpiwg-sessions" <mpiwg-sessions at lists.mpi-forum.org>
Reply-To: MPI Sessions working group <mpiwg-sessions at lists.mpi-forum.org>
Date: Wednesday, 24. February 2021 at 22:07
To: MPI Sessions working group <mpiwg-sessions at lists.mpi-forum.org>
Cc: "Pritchard Jr., Howard" <howardp at lanl.gov>
Subject: [mpiwg-sessions] note from off-cadence WG meeting to discuss issue 435, etc.
Dan and I met after the forum today. Sorry about the webex issues others had, not sure what was wrong.
Anyway here’s what we discussed/did.
We agreed to not change any of the text that would currently be added by
but would add an item in the normative text linking back to examples in section 6.14 and the dangers of having arbitrary association of communicators with sessions handles across MPI processes in a MPI job.
I added Rolf’s example case but haven’t been able to push up to github owing to the github issue with private repos that’s happening today. Hopefully it will clear up by tomorrow. Dan and I agreed that the “rank” usage in the example would be sure to cause issues, so replaced ranks with X, Y, Z.
We also discussed issue 424 - https://github.com/mpi-forum/mpi-issues/issues/424. It’s been added to the project board. I will also go thru the associated PR if we have time tomorrow.
We also briefly discussed the Dynamic Process Model/Sessions and decided to defer further discussion of that topic till after the meeting tomorrow.
Los Alamos National Laboratory
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the mpiwg-sessions