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

Reading request for Session WG issue related to clarifying err handling fallback
by Aurelien Bouteiller 21 Nov '22

21 Nov '22

Hello All, I’d like to put in a request to read the following pull request at the December ’22 MPI Forum: ## Clarification of errhandler fallback with Sessions Issue: https://github.com/mpi-forum/mpi-issues/issues/588 PR: https://github.com/mpi-forum/mpi-standard/pull/644 PR 644 was already read (as-is) in prior meetings, but it was labeled under the wrong issue number and that caused confusion. In the spirit of transparency we will restart the process with a dedicated issue number. This is an MPI 4.1 item. Best Regards, Aurelien -- Aurelien Bouteiller, Ph.D. Innovative Computing Laboratory; The University of Tennessee; 1122 Volunteer Blvd.; Claxton suite 203; 37996, Knoxville, TN, USA +1 865 974 9308 (p); +1 865 974 8296 (f); Claxton 316 (f2f)

1 0

Reading requests: language interoperability concerns
by Joseph Schuchart 21 Nov '22

21 Nov '22

On behalf of JeffH I'd like to request the first reading of the following issues/PRs addressing language interoperability issues at the December 2022 meeting: *add MPI_Status_{set,get} for 3 status fields* Issue: https://github.com/mpi-forum/mpi-issues/issues/645 PR: https://github.com/mpi-forum/mpi-standard/pull/767 *Remove the handle conversion functions from the list of allowed macros* Issue: https://github.com/mpi-forum/mpi-issues/issues/647 PR: https://github.com/mpi-forum/mpi-standard/pull/770 *Remove (P)MPI_Wtime and (P)MPI_Wtick from list of allowed macros* Issue: https://github.com/mpi-forum/mpi-issues/issues/646 PR: https://github.com/mpi-forum/mpi-standard/pull/771 Cheers Joseph

1 0

Re: [Mpi-forum] RMA WG reading/vote announcements for December 2022 meeting
by Joseph Schuchart 21 Nov '22

21 Nov '22

I'd like to push three more items to the heap of first reading requests for the RMA chapter: *RMA communication procedures should have local semantics* Issue: https://github.com/mpi-forum/mpi-issues/issues/653 PR: https://github.com/mpi-forum/mpi-standard/pull/769 This change depends on the larger change adjusting the RMA chapter to the new semantics. *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 This change depends on the larger change adjusting the RMA chapter to the new semantics. *Wrap text about blocking MPI functions in Example 12.4 and 12.5 in AtoI* Issue: https://github.com/mpi-forum/mpi-issues/issues/652 PR: https://github.com/mpi-forum/mpi-standard/pull/768 Thanks, Joseph On 11/18/22 18:14, Joseph Schuchart wrote: > Dear all, > > On behalf of the RMA WG I'd like to request a first reading of the > following tickets: > > *Add info key for RMA accumulate behavior preference* > Issue: https://github.com/mpi-forum/mpi-issues/issues/640 > PR: https://github.com/mpi-forum/mpi-standard/pull/749 > PDF: > https://github.com/mpi-forum/mpi-standard/files/10036637/mpi41-report-accum… > > *Rephrase MPI_WIN_SYNC to make it clear what it's used for* > Issue: https://github.com/mpi-forum/mpi-issues/issues/634 > PR: https://github.com/mpi-forum/mpi-standard/pull/747 > PDF: > https://github.com/mpi-forum/mpi-standard/files/10036667/mpi41-report-win-s… > > The following tickets had a reading and subsequent (minor) changes so > I'd like to request no-no-votes and the first vote for them: > > *RMA: Add missing win-sync calls to Example 12.21* > Issue: https://github.com/mpi-forum/mpi-issues/issues/553 > PR: https://github.com/mpi-forum/mpi-standard/pull/663 > PDF: > https://github.com/mpi-forum/mpi-standard/files/10042342/mpi41-report-rma-w… > Changes since the reading in September 2022: > https://github.com/mpi-forum/mpi-standard/pull/663/commits/d754108182a2e5f4… > > *MPI_Win_free is synchronizing, except when it isn't* > Issue: https://github.com/mpi-forum/mpi-issues/issues/551 > PR: https://github.com/mpi-forum/mpi-standard/pull/659 > PDF: > https://github.com/mpi-forum/mpi-standard/files/10042443/mpi41-report-mpi-w… > Changes since the reading in September 2022: > https://github.com/mpi-forum/mpi-standard/pull/659/files/81e277f3ec81587c49… > > *RMA: Replace 'operation' with 'operator' where appropriate* > Issue: https://github.com/mpi-forum/mpi-issues/issues/429 > PR: https://github.com/mpi-forum/mpi-standard/pull/637/files > PDF: > https://github.com/mpi-forum/mpi-standard/files/10045184/mpi41-report-rma-o… > Changes since the reading in September 2022: > https://github.com/mpi-forum/mpi-standard/pull/637/files/203e21355cc13054db… > > *Fix semantic terms in RMA chapter* > Issue: https://github.com/mpi-forum/mpi-issues/issues/407 > PR: https://github.com/mpi-forum/mpi-standard/pull/611 > PDF: > https://github.com/mpi-forum/mpi-standard/files/10045382/mpi41-report-rma-s… > Changes since last reading in May 2022: > https://github.com/mpi-forum/mpi-standard/pull/611/files/21489fa5db19f03474… > I should note that this one had quite some additions but most changes > are scattered and independent of other changes throughout the chapter. > If the Forum feels like this is too much for a no-no-vote then I'll be > happy to do a full reading. > > > The following ticket had a first reading and significant changes > since, so I believe another full reading is warranted: > > *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/files > PDF: > https://github.com/mpi-forum/mpi-standard/files/10045238/mpi41-report-win-s… > > > I'd like to request a first vote for the following ticket: > > *RMA: both MPI_WIN_ALLOCATE and MPI_WIN_ALLOCATE_SHARED allocate memory* > Issue: https://github.com/mpi-forum/mpi-issues/issues/554 > PR: https://github.com/mpi-forum/mpi-standard/pull/664 > PDF: > https://github.com/mpi-forum/mpi-standard/files/8652275/mpi40-report-rma-wi… > > I hope I did not forget any ticket. I may add another ticket to make > RMA communication operations local before the deadline on Monday but > wanted to get this list out first. > > Thanks > Joseph

1 0

Announcment reading/vote for issues improving the topology chapter
by Christoph Niethammer 21 Nov '22

21 Nov '22

Hello all, For the upcoming December ‘22 MPI forum, I’d like to reserve a view minutes for presenting updates to the topology chapter. One suggested change is updating the chapter title - which the forum likely wants to have a (short) discussion and - reading/vote on #644 "Update "Process Topology" chapter title to use "MPI process" Issue: https://github.com/mpi-forum/mpi-issues/issues/644 PR: https://github.com/mpi-forum/mpi-standard/pull/759 Beside this there are the following updates I'd like to read if they are not passed as chapter committee change. If they pass as chapter committee change, then please take this only as a notice about those changes: - #648: Topology LIS text improvements Issue: https://github.com/mpi-forum/mpi-issues/issues/648 PR: https://github.com/mpi-forum/mpi-standard/pull/763 - #649: Topology - use of \MPI process where necessary Issue: https://github.com/mpi-forum/mpi-issues/issues/649 PR: https://github.com/mpi-forum/mpi-standard/pull/764 - #650: Topology improvement around the term virtual topology Issue: https://github.com/mpi-forum/mpi-issues/issues/650 PR: https://github.com/mpi-forum/mpi-standard/pull/765 - #651: Topology typos and smaller text improvements Issue: https://github.com/mpi-forum/mpi-issues/issues/651 PR: https://github.com/mpi-forum/mpi-standard/pull/744 Thanks to Dan, Rolf, and Guillaume as well as the HW-topology WG for feedback and discussions! Best regards Christoph Niethammer -- Christoph Niethammer High Performance Computing Center Stuttgart (HLRS) Nobelstrasse 19 70569 Stuttgart Tel: ++49(0)711-685-87203 email: christoph.niethammer(a)hlrs.de https://www.hlrs.de/people/christoph-niethammer

1 0

Readings announcements for the December meeting of the MPI Forum
by Guillaume Mercier 21 Nov '22

21 Nov '22

Dear all, I'd like to make the following readings on behalf of the Topologies working Group: 1- Allow pset names to guide splitting in MPI_Comm_split_type Issue: https://github.com/mpi-forum/mpi-issues/issues/538 PR: https://github.com/mpi-forum/mpi-standard/pull/761 2- New functions to query hardware resource names Issue: https://github.com/mpi-forum/mpi-issues/issues/154 PR: https://github.com/mpi-forum/mpi-standard/pull/762 Christoph will also send a message to announce readings that deal with the "Virtual Topologies" chapter of the standard. Best regards Guillaume

1 0

Re: [Mpi-forum] RMA WG reading/vote announcements for December 2022 meeting
by Joseph Schuchart 21 Nov '22

21 Nov '22

Yes, if the Forum agrees that it's an errata then we can vote on it right away. Thanks Joseph On 11/21/22 09:21, Wes Bland wrote: > #634 is marked as errata. Should I schedule this for a vote as well? > > Thanks, > Wes > >> On Nov 18, 2022, at 5:14 PM, Joseph Schuchart via mpi-forum <mpi-forum(a)lists.mpi-forum.org> wrote: >> >> Dear all, >> >> On behalf of the RMA WG I'd like to request a first reading of the following tickets: >> >> *Add info key for RMA accumulate behavior preference* >> Issue: https://github.com/mpi-forum/mpi-issues/issues/640 >> PR: https://github.com/mpi-forum/mpi-standard/pull/749 >> PDF: https://github.com/mpi-forum/mpi-standard/files/10036637/mpi41-report-accum… >> >> *Rephrase MPI_WIN_SYNC to make it clear what it's used for* >> Issue: https://github.com/mpi-forum/mpi-issues/issues/634 >> PR: https://github.com/mpi-forum/mpi-standard/pull/747 >> PDF: https://github.com/mpi-forum/mpi-standard/files/10036667/mpi41-report-win-s… >> >> The following tickets had a reading and subsequent (minor) changes so I'd like to request no-no-votes and the first vote for them: >> >> *RMA: Add missing win-sync calls to Example 12.21* >> Issue: https://github.com/mpi-forum/mpi-issues/issues/553 >> PR: https://github.com/mpi-forum/mpi-standard/pull/663 >> PDF: https://github.com/mpi-forum/mpi-standard/files/10042342/mpi41-report-rma-w… >> Changes since the reading in September 2022: https://github.com/mpi-forum/mpi-standard/pull/663/commits/d754108182a2e5f4… >> >> *MPI_Win_free is synchronizing, except when it isn't* >> Issue: https://github.com/mpi-forum/mpi-issues/issues/551 >> PR: https://github.com/mpi-forum/mpi-standard/pull/659 >> PDF: https://github.com/mpi-forum/mpi-standard/files/10042443/mpi41-report-mpi-w… >> Changes since the reading in September 2022: https://github.com/mpi-forum/mpi-standard/pull/659/files/81e277f3ec81587c49… >> >> *RMA: Replace 'operation' with 'operator' where appropriate* >> Issue: https://github.com/mpi-forum/mpi-issues/issues/429 >> PR: https://github.com/mpi-forum/mpi-standard/pull/637/files >> PDF: https://github.com/mpi-forum/mpi-standard/files/10045184/mpi41-report-rma-o… >> Changes since the reading in September 2022: https://github.com/mpi-forum/mpi-standard/pull/637/files/203e21355cc13054db… >> >> *Fix semantic terms in RMA chapter* >> Issue: https://github.com/mpi-forum/mpi-issues/issues/407 >> PR: https://github.com/mpi-forum/mpi-standard/pull/611 >> PDF: https://github.com/mpi-forum/mpi-standard/files/10045382/mpi41-report-rma-s… >> Changes since last reading in May 2022: https://github.com/mpi-forum/mpi-standard/pull/611/files/21489fa5db19f03474… >> I should note that this one had quite some additions but most changes are scattered and independent of other changes throughout the chapter. If the Forum feels like this is too much for a no-no-vote then I'll be happy to do a full reading. >> >> >> The following ticket had a first reading and significant changes since, so I believe another full reading is warranted: >> >> *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/files >> PDF: https://github.com/mpi-forum/mpi-standard/files/10045238/mpi41-report-win-s… >> >> >> I'd like to request a first vote for the following ticket: >> >> *RMA: both MPI_WIN_ALLOCATE and MPI_WIN_ALLOCATE_SHARED allocate memory* >> Issue: https://github.com/mpi-forum/mpi-issues/issues/554 >> PR: https://github.com/mpi-forum/mpi-standard/pull/664 >> PDF: https://github.com/mpi-forum/mpi-standard/files/8652275/mpi40-report-rma-wi… >> >> I hope I did not forget any ticket. I may add another ticket to make RMA communication operations local before the deadline on Monday but wanted to get this list out first. >> >> Thanks >> Joseph >> _______________________________________________ >> 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] why is MPI_Status not opaque?
by Jeff Hammond 21 Nov '22

21 Nov '22

Yes, but every time the MPI Forum breaks backwards compatibility, an angel loses its wings, so we can’t fix design flaws like this. Everyone must suffer forever. More seriously, Lisandro pointer out that we’d need allocate and deallocate functions for status if they were handles to hidden state, so we can’t just fix this by making Status opaque. I guess we have no choice but to standardize the MPI_Status object size and layout. Jeff > On 21Nov 2022, at 2:43 PM, Skjellum, Anthony <tony-skjellum(a)utc.edu> wrote: > > It should be an opaque object :-) > > Anthony Skjellum, PhD > 205-807-4968 > > > On Nov 21, 2022, at 7:29 AM, Jeff Hammond via mpi-forum <mpi-forum(a)lists.mpi-forum.org> wrote: > > > I assume that MPI_Status is not opaque because somebody asserted that function call overhead was too much for some use cases. Was there more to it than this? > > Why does the standard say there is an opaque part for elements and cancelled, but not make those visible? The lack of consistency here doesn't make a lot of sense to me. > > MPI_Status not being opaque was a horrible mistake but I would like to be less mad about it by learning what possible reasons for it existed in 1995. > > Thanks, > > Jeff > > -- > Jeff Hammond > jeff.science(a)gmail.com <mailto:[email protected]> > http://jeffhammond.github.io/ <https://nam11.safelinks.protection.outlook.com/?url=http%3A%2F%2Fjeffhammon…>_______________________________________________ > mpi-forum mailing list > mpi-forum(a)lists.mpi-forum.org > https://lists.mpi-forum.org/mailman/listinfo/mpi-forum

1 0

Announcment reading/vote on #589
by Christoph Niethammer 21 Nov '22

21 Nov '22

Hello all, For the upcoming December ‘22 MPI forum, I’d like to announce a * reading/vote on errata/issue #589 "Fixes for partitioned communication example 4.4" issue: https://github.com/mpi-forum/mpi-issues/issues/589 PR: https://github.com/mpi-forum/mpi-standard/pull/726 Best, Christoph Niethammer -- Christoph Niethammer High Performance Computing Center Stuttgart (HLRS) Nobelstrasse 19 70569 Stuttgart Tel: ++49(0)711-685-87203 email: christoph.niethammer(a)hlrs.de https://www.hlrs.de/people/christoph-niethammer

1 0

Announcements for Dec'22 MPI Forum meeting
by Holmes, Daniel John 21 Nov '22

21 Nov '22

Hi all, I'd like to request time at the Dec'22 meeting for the following issues/PRs: Reading: Enabled is a semantic term Issue 421: https://github.com/mpi-forum/mpi-issues/issues/421 PR 760: https://github.com/mpi-forum/mpi-standard/pull/760 Reading: Unsafe example relies on MPI providing buffering Issue 188: https://github.com/mpi-forum/mpi-issues/issues/188 PR: https://github.com/mpi-forum/mpi-standard/pull/203 Reading: Fix usage of rank/task/etc in point-to-point chapter Issue 472: https://github.com/mpi-forum/mpi-issues/issues/472 PR: 606: https://github.com/mpi-forum/mpi-standard/pull/606 Reading: INFO key for strict start ordering of persistent collectives Issue 83: https://github.com/mpi-forum/mpi-issues/issues/83 PR 666: https://github.com/mpi-forum/mpi-standard/pull/666 Reading: Logically concurrent isn't Issue 117: https://github.com/mpi-forum/mpi-issues/issues/117 PR 748: https://github.com/mpi-forum/mpi-standard/pull/748 Reading: MPI_WAIT is nonlocal, except when it isn't Issue 639: https://github.com/mpi-forum/mpi-issues/issues/639 PR 750: https://github.com/mpi-forum/mpi-standard/pull/750 Best wishes, Dan.

1 0

Re: [Mpi-forum] why is MPI_Status not opaque?
by Skjellum, Anthony 21 Nov '22

21 Nov '22

It should be an opaque object :-) Anthony Skjellum, PhD 205-807-4968 On Nov 21, 2022, at 7:29 AM, Jeff Hammond via mpi-forum <mpi-forum(a)lists.mpi-forum.org> wrote: I assume that MPI_Status is not opaque because somebody asserted that function call overhead was too much for some use cases. Was there more to it than this? Why does the standard say there is an opaque part for elements and cancelled, but not make those visible? The lack of consistency here doesn't make a lot of sense to me. MPI_Status not being opaque was a horrible mistake but I would like to be less mad about it by learning what possible reasons for it existed in 1995. Thanks, Jeff -- Jeff Hammond jeff.science(a)gmail.com<mailto:[email protected]> http://jeffhammond.github.io/<https://nam11.safelinks.protection.outlook.com/?url=http%3A%2F%2Fjeffhammon…> _______________________________________________ mpi-forum mailing list mpi-forum(a)lists.mpi-forum.org https://lists.mpi-forum.org/mailman/listinfo/mpi-forum

1 0