- mpi-forum - mailman3.cels.anl.gov

RMA WG Vote Announcements
by Joseph Schuchart 17 Apr '23

17 Apr '23

Dear all, The RMA WG would like to request the following votes: No-no-vote and subsequent 1st vote: *RMA accumulate operations don't distinguish between throughput and latency sensitive applications* Issue: https://github.com/mpi-forum/mpi-issues/issues/640 PR: https://github.com/mpi-forum/mpi-issues/issues/640 Changes from the March meeting: https://github.com/mpi-forum/mpi-standard/pull/749/commits/82ab18050ce96975… Second votes: *Example 12.4: RMA communication calls may not block in PSCW* Issue: https://github.com/mpi-forum/mpi-issues/issues/555 PR: https://github.com/mpi-forum/mpi-standard/pull/671 *Make all RMA communication procedures local* Issue: https://github.com/mpi-forum/mpi-issues/issues/653 PR: https://github.com/mpi-forum/mpi-standard/pull/769 *RMA: Examples 12.4 and 12.5 contain advice to implementors* Issue: https://github.com/mpi-forum/mpi-issues/issues/652 PR: https://github.com/mpi-forum/mpi-standard/pull/768 No-no-vote with subsequent second vote: *Allow MPI_WIN_SHARED_QUERY on created and allocated windows* Issue: https://github.com/mpi-forum/mpi-issues/issues/23 PR: https://github.com/mpi-forum/mpi-standard/pull/708 Changes since March reading: https://github.com/mpi-forum/mpi-standard/pull/708/commits/9a62a36f4cb3c930… Thanks Joseph

1 0

Next Voting Meeting starts May 2nd - Deadline for announcements is April 18th
by Martin Schulz 16 Apr '23

16 Apr '23

Hi all, We’ll have our next voting meeting starting May 2nd – this will be a virtual meeting and we’ll use the usual timeslots from 9am to 1pm US Central time. A preliminary agenda is already online (I know a few items are still missing, they will be added soon). The deadline for any ballot or reading announcements is Tuesday April 18th. Priority will be to complete the needed discussions, (no/no and errata) readings for MPI 4.1 and the associated votes, but there should be time for other items. Hence, if any WG would like to have plenary discussions, please let me know. Thanks, Martin -- 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(a)in.tum.de

1 0

Re: [Mpi-forum] Request for Input / MPI 4.0 compliance of MPI implementations
by Martin Schulz 11 Apr '23

11 Apr '23

Thanks for digging this out – yes, that was the list. If the respective groups are still active and would like to participate, please send me a contact. Also, there are a few new additional MPI libraries from the last few years (or even older) – if you would like to be included let me know as well. Thanks! Martin -- 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(a)in.tum.de From: Wes Bland <wesley(a)wesbland.com> Date: Tuesday, 11. April 2023 at 15:23 To: Martin Schulz <schulzm(a)in.tum.de> Cc: MPI Forum <mpi-forum(a)lists.mpi-forum.org> Subject: Re: [Mpi-forum] Request for Input / MPI 4.0 compliance of MPI implementations The last version of this slide that I could find easily was from 2018 (https://www.mpi-forum.org/bofs/2018-11-sc/intro.pdf slide 3). At that time, we had these implementations: MPICH MVAPICH Open MPI Cray Tianhe Intel MPI IBM (BG/Q, PE, Spectrum) HPE Fujitsu Microsoft MPC NEC Sunway RIKEN Adaptive MPI I’m sure there are some on here that would drop off, but maybe we can start with these folks? Thanks, Wes On Apr 11, 2023, at 4:06 AM, Martin Schulz via mpi-forum <mpi-forum(a)lists.mpi-forum.org> wrote: Hi all, We have gotten a few requests to restart the tradition on reporting on the status of MPI implementations at the traditional SC and ISC BoFs, in particular their compliance with the newest standard on the various new features. I would, therefore, like to try to get that information together for ISC, but I will need from the MPI implementors. For the main open source implementations I can guess the right contacts, but then we have a wide range of other implementations and variants, which would be great to include as well. Therefore, I would like to ask any MPI implementation team who would like to be included in this chart to let me know and send me a contact I can connect with. Thanks! Martin -- 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(a)in.tum.de _______________________________________________ mpi-forum mailing list mpi-forum(a)lists.mpi-forum.org https://lists.mpi-forum.org/mailman/listinfo/mpi-forum

1 0

Re: [Mpi-forum] Another announcement for the May'23 MPI Forum meeting
by Holmes, Daniel John 11 Apr '23

11 Apr '23

Hi Wes (et al), Yes, apologies - that's a copy&paste error on my part - issue 639 and PR 750 is the intended combination. Best wishes, Dan. From: Wes Bland <wesley(a)wesbland.com> Sent: 11 April 2023 16:38 To: Holmes, Daniel John <daniel.john.holmes(a)intel.com> Cc: MPI Forum <mpi-forum(a)lists.mpi-forum.org> Subject: Re: [Mpi-forum] Another announcement for the May'23 MPI Forum meeting Hi Dan, I assume the PR for this should be https://github.com/mpi-forum/mpi-standard/pull/750 instead? That PR points to https://github.com/mpi-forum/mpi-issues/issues/669. Thanks, Wes On Apr 11, 2023, at 5:44 AM, Holmes, Daniel John via mpi-forum <mpi-forum(a)lists.mpi-forum.org<mailto:[email protected]>> wrote: Hi all, I'd also like to announce a 1st vote: Title: MPI_WAIT is nonlocal, except when it isn't Issue: https://github.com/mpi-forum/mpi-issues/issues/639 PR: https://github.com/mpi-forum/mpi-standard/pull/788 Best wishes, Dan. From: mpi-forum <mpi-forum-bounces(a)lists.mpi-forum.org<mailto:[email protected]>> On Behalf Of Holmes, Daniel John via mpi-forum Sent: 11 April 2023 11:38 To: MPI Forum <mpi-forum(a)lists.mpi-forum.org<mailto:[email protected]>> <mpi-forum(a)lists.mpi-forum.org<mailto:[email protected]>> Subject: [Mpi-forum] Errata reading announcement Hi all, If we still have time in the procedure for MPI-4.1, I'd like to read an errata ticket at the next meeting. Title: Correct the MPI_TESTANY 'as if' wording Issue: https://github.com/mpi-forum/mpi-issues/issues/681 PR: https://github.com/mpi-forum/mpi-standard/pull/818 Best wishes, Dan. _______________________________________________ mpi-forum mailing list mpi-forum(a)lists.mpi-forum.org<mailto:[email protected]> https://lists.mpi-forum.org/mailman/listinfo/mpi-forum

1 0

Another announcement for the May'23 MPI Forum meeting
by Holmes, Daniel John 11 Apr '23

11 Apr '23

Hi all, I'd also like to announce a 1st vote: Title: MPI_WAIT is nonlocal, except when it isn't Issue: https://github.com/mpi-forum/mpi-issues/issues/639 PR: https://github.com/mpi-forum/mpi-standard/pull/788 Best wishes, Dan. From: mpi-forum <mpi-forum-bounces(a)lists.mpi-forum.org> On Behalf Of Holmes, Daniel John via mpi-forum Sent: 11 April 2023 11:38 To: MPI Forum <mpi-forum(a)lists.mpi-forum.org> <mpi-forum(a)lists.mpi-forum.org> Subject: [Mpi-forum] Errata reading announcement Hi all, If we still have time in the procedure for MPI-4.1, I'd like to read an errata ticket at the next meeting. Title: Correct the MPI_TESTANY 'as if' wording Issue: https://github.com/mpi-forum/mpi-issues/issues/681 PR: https://github.com/mpi-forum/mpi-standard/pull/818 Best wishes, Dan.

1 0

Errata reading announcement
by Holmes, Daniel John 11 Apr '23

11 Apr '23

Hi all, If we still have time in the procedure for MPI-4.1, I'd like to read an errata ticket at the next meeting. Title: Correct the MPI_TESTANY 'as if' wording Issue: https://github.com/mpi-forum/mpi-issues/issues/681 PR: https://github.com/mpi-forum/mpi-standard/pull/818 Best wishes, Dan.

1 0

Request for Input / MPI 4.0 compliance of MPI implementations
by Martin Schulz 11 Apr '23

11 Apr '23

Hi all, We have gotten a few requests to restart the tradition on reporting on the status of MPI implementations at the traditional SC and ISC BoFs, in particular their compliance with the newest standard on the various new features. I would, therefore, like to try to get that information together for ISC, but I will need from the MPI implementors. For the main open source implementations I can guess the right contacts, but then we have a wide range of other implementations and variants, which would be great to include as well. Therefore, I would like to ask any MPI implementation team who would like to be included in this chart to let me know and send me a contact I can connect with. Thanks! Martin -- 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(a)in.tum.de

1 0

Announcement of no-no vote and first vote for REQUEST_GET_STATUS_XXX
by William Williams 11 Apr '23

11 Apr '23

Hi all, Based on the March reading, I have no-no changes for #799 and would, provided that those pass, like to bring it to a first vote in May. PR: https://github.com/mpi-forum/mpi-standard/pull/799/ No-no diff: https://github.com/mpi-forum/mpi-standard/pull/799/files/5391d71c0c14a0fdac… Issue: https://github.com/mpi-forum/mpi-issues/issues/519 <https://github.com/mpi-forum/mpi-issues/issues/519> Cheers, Bill -- Bill Williams Wiss. Mitarbeiter ZIH-IAK, TU-Dresden Falkenbrunnen, Room 009 Chemnitzer Straße 50 01187 Dresden Phone: +49 351 463-39976 e-mail: william.williams(a)mailbox.tu-dresden.de

1 0

May 2023 MPI Forum Meeting Registration and Announcements
by Wes Bland 06 Apr '23

06 Apr '23

Hi all, The next MPI Forum meeting is now about a month away. That means it’s time to register and get out voting announcements. Please register at this link: https://forms.gle/ae8TyARWi7Hchcqb6 Please also make sure you are preparing your tickets for readings and votes. The deadline for announcements is April 18th. There were a number of announcements that were already made. You can see the current agenda here: https://www.mpi-forum.org/meetings/2023/05/agenda There are also a number of issues that have either had readings or votes already that still need announcements for the current meeting. You can see everything we’re tracking for MPI 4.1 that isn’t announced in one form or another here: https://github.com/orgs/mpi-forum/projects/1/views/9 We’re still on track for completing MPI 4.1 later this year. Our plan for that is: * May - Final 1st votes * July - Final 2nd votes and errata votes * September - Release Candidate Meeting (RCM) * October - Final Ratification Meeting (FRM) That means that everything that hasn’t already had a reading at this point that isn’t an errata or chapter committee change is too late for MPI 4.1 and will need to move out to target MPI 5.0 (or, very narrowly, 4.2). I’ll be moving those tickets out to MPI 5.0 soon. You can confirm whether your issue is in that bucket here: https://github.com/orgs/mpi-forum/projects/1/views/10 As always, let me know if you have any questions. Thanks, Wes

1 0

question about MPI_UNDEFINED and MPI_DISPLACEMENT_CURRENT
by Jeff Hammond 03 Apr '23

03 Apr '23

Does anyone know the history for either of these values? MPI_UNDEFINED = -32766 MPI_DISPLACEMENT_CURRENT = -54278278 MPI_UNDEFINED is close enough to SHRT_MIN but the other one doesn't make sense to me. Thanks, Jeff -- Jeff Hammond jeff.science(a)gmail.com http://jeffhammond.github.io/

1 0