June 2020 - mpi-forum - mailman3.cels.anl.gov

Agenda / Votes for 2020-07-01
by Wesley Bland 30 Jun '20

30 Jun '20

Hi all, Thanks for another good virtual meeting today. The schedule of plenaries and votes for tomorrow has been updated after today’s readings to include necessary no-no readings and to add and remove votes as necessary. So tomorrow morning before the voting block, there will be these no-no readings: #167: Dan - Error - Section 3.3.2 - p. 38 - Replace C process (no-no) #252: Dan - Clarification - Section 3.2.2 - p. 27 - Datatype Footnote (no-no) #190: Dan - Update - Section 3.7.5 - p. 60 - Replace C-bias with LIS (no-no) These readings are considered complete along with the no-no changes made during the reading itself (do not require a no-no reading): #263: Wesley - Clarification - Section 11.7 - p. 482 - MPI Exception Clarification (no-no) #181: George - Oversight - Section 8.7 - p. 383 - Escape Spaces in Arguments (no-no) #254: George - Clarification - Example 4.17 - p. 128 - Example Comment (no-no) These readings will be voted on as originally read: #179: Joseph - Omission - p. 440 - WIN_DETATCH Binding Fix (errata) #178: Joseph - Omission - p. 439 - WIN_ATTACH_ Binding Fix (errata) #218: Wesley - Update - Section 17.1 - p. 635 - Update Backward Incompatibility Text (errata) #262: Wesley - Clarification - Section 11.3.5 - p. 458 - Error in Status Field Clarification (errata) #171: George - Error in text - Section 4.1.1 - p. 87 - INTEGER*8 Extension (errata) #257: Julien - Clarification - Section 5.1 - p. 144 - Collective Synchronization (errata) #182: Julien - Question - Section 5.12.4 - p. 206 - MPI_ISCATTERV sendcounts and displs Consistency (errata) You can find the sample ballot for tomorrow here: https://www.jotform.com/build/201614072992151 <https://www.jotform.com/build/201614072992151> After the voting block starts, we’ll do the following plenaries in this order: Reading - 120 - MPI_Cart_create_weighted / Topology aware Cartesian communicators - Rolf Errata Readings - Read all errata tickets up for vote in the order of above Reading - 57 - Deprecate MPI_HOST - Julien Reading - 259 - Clarification - Section 8.5 - p. 377 - Error Class and Code Advice to Implementors - Wesley Discussion - Next Meeting(s) and Timeline - Martin As you can see, we still have quite a few things we need to complete tomorrow so we’ll be starting on time and moving quickly. I’d encourage minor comments (e.g. typos and small word-smithing discussions) to use GitHub to allow time for all of the topics. If anyone has any corrections for the list above, please let Martin and I know and we’ll get everything updated. Thanks, Wes

1 0

Call for short talks at the Q3 2020 PMIx Standard Quarterly Administrative Steering Committee (ASC) Meeting
by Mohror, Kathryn 29 Jun '20

29 Jun '20

Hi all, The PMIx Administrative Steering Committee (ASC) is soliciting short talks from researchers and practitioners whose work is related to the PMIx standard interface for inclusion in the second-quarter PMIx ASC meeting (meeting information below) on July 22. We encourage presentations demonstrating novel application or optimization of the PMIx interface, as well as presentations that introduce new use cases or other issues for broad discussion. Selected speakers will share their work and ideas with the PMIx community which could result in shaping the future direction of the PMIx standard or related efforts. The objective of the PMIx standard is to provide an implementation-independent specification to facilitate interaction between applications, tools, and middleware and system runtime environments. The interface focuses on supporting application launch and wire-up; supporting tools that wish to inspect, steer, and/or debug parallel applications; providing interfaces to manage dynamic workload applications, to support fault tolerance and cross-library coordination; and facilitate communication across container boundaries. We encourage anyone interested in sharing ideas or research related to these topics at our upcoming meeting to email Josh Hursey (jhursey(a)us.ibm.com<mailto:[email protected]>) and Kathryn Mohror (kathryn(a)llnl.gov<mailto:[email protected]>) by July 13, 2020. Include a title, presenter information, and a short abstract in your email. Meeting information: * Date: July 22, 2020 * Time: 11 am - 5 pm US Central Time * Location: Virtual (online) Meeting (details on how to join will be available soon) * Meeting Link: https://github.com/pmix/pmix-standard/wiki/ASC-Q3-2020-Meeting For more information about the PMIx standardization body and its activities please join the mailing list [1] and check out the activity on the GitHub repository [2]. [1] PMIx Forum Mailing list: https://groups.google.com/forum/#!forum/pmix-forum [2] PMIx Standard GitHub: https://github.com/pmix/pmix-standard Kathryn Mohror and Josh Hursey PMIx ASC Co-Chairs _________________________________________________________________ Kathryn Mohror, kathryn(a)llnl.gov<mailto:[email protected]>, https://people.llnl.gov/kathryn Data Analysis Group<https://computation.llnl.gov/casc/data-analysis-group> @ Lawrence Livermore National Laboratory, Livermore, CA, USA

1 0

Re: [Mpi-forum] Agenda and Ballots for Tomorrow
by Wesley Bland 29 Jun '20

29 Jun '20

One quick addendum / clarification to this: We will not have the first vote on "#120 - MPI_Cart_create_weighted / Topology aware Cartesian communicators” on Tuesday. We will only have the “no-no” vote on Tuesday. If the “no-no” vote passes on Tuesday, we will have the first vote for "#120 - MPI_Cart_create_weighted / Topology aware Cartesian communicators” on Wednesday. If there’s any confusion, you can see the two ballots here: Tuesday - https://www.jotform.com/build/201804215921143 <https://www.jotform.com/build/201804215921143> Wednesday - https://www.jotform.com/build/201614072992151 <https://www.jotform.com/build/201614072992151> Note that you can’t actually vote yet. Your personalized link for voting will be sent out tomorrow during the voting block. Thanks, Wes > On Jun 29, 2020, at 2:00 PM, Martin Schulz via mpi-forum <mpi-forum(a)lists.mpi-forum.org> wrote: > > Hi all, > > Thanks for a first, quite productive session today. > > Just a quick preview for tomorrow: we will start with a discussion on ticket #120 by Rolf and the proposed NoNo Vote text. > > The voting block will open at 9:30 and will stay open until 10:00, at which time Wesley will announce the results - for the block tomorrow, we’ll include the procedural votes, the NoNos, as well as all first and second votes. If you are the owner of one of those, please be prepared to briefly introduce it, if requested. > > For the rest of the day, we will then focus on errata readings, so we can vote on them on Wednesday. The agenda lists the order - we grouped them by owner, so we don’t have to switch speakers that often - please take a look at this if you own an errata ticket. > > Thanks and talk to you all tomorrow, > > 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

Agenda and Ballots for Tomorrow
by Martin Schulz 29 Jun '20

29 Jun '20

Hi all, Thanks for a first, quite productive session today. Just a quick preview for tomorrow: we will start with a discussion on ticket #120 by Rolf and the proposed NoNo Vote text. The voting block will open at 9:30 and will stay open until 10:00, at which time Wesley will announce the results - for the block tomorrow, we’ll include the procedural votes, the NoNos, as well as all first and second votes. If you are the owner of one of those, please be prepared to briefly introduce it, if requested. For the rest of the day, we will then focus on errata readings, so we can vote on them on Wednesday. The agenda lists the order - we grouped them by owner, so we don’t have to switch speakers that often - please take a look at this if you own an errata ticket. Thanks and talk to you all tomorrow, 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

Reminder: MPI Forum (virtual voting) meeting Today (Monday) - Wednesday
by Martin Schulz 29 Jun '20

29 Jun '20

Hi all, Just as a reminder, we will have our next full (voting) MPI Forum meeting starting today until Wednesday. We will meet at 9am Central time and the forum is scheduled to last for 3 hours each day. The agenda and the Webex links are at the usual place. We will only do plenary meetings, but we do have a lot of ground to cover (in particular many erratas and a few last big tickets for MPI 4.0). The order on the agenda should be roughly correct, but we will adjust as needed to make sure we get the critical things done. Thanks and talk to you later today! 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] Semantic Terms (Issue #96) Reading Announcement
by Bangalore, Purushotham 28 Jun '20

28 Jun '20

The Semantic Terms WG would like to announce a no "no vote" for Issue #: https://github.com/mpi-forum/mpi-issues/issues/96 and PR: https://github.com/mpi-forum/mpi-standard/pull/116. The changes since the original announcement before the two week deadline can be found here<https://github.com/mpi-forum/mpi-standard/pull/116/files/f51be453bd3ca2f810…>. The corresponding annotated PDF file can be found here<https://github.com/mpi-forum/mpi-issues/files/4841073/mpi40-report-june2020…>. There is also a short presentation that explains these changes here<https://github.com/mpi-forum/mpi-issues/files/4842762/SemanticTerms-4.0-no-…>. We also have a presentation on additional discussions related to this topic to be considered for MPI 4.1 here<https://github.com/mpi-forum/mpi-issues/files/4842764/SemanticTerms-4.1-Dis…>. Thanks, Puri on behalf of the Semantic Terms WG ________________________________ From: Bangalore, Purushotham <puri(a)uab.edu> Sent: Monday, June 15, 2020 8:15 AM To: Wesley Bland via mpi-forum <mpi-forum(a)lists.mpi-forum.org> Subject: Semantic Terms (Issue #96) Reading Announcement Here is the official announcement for reading issue #96 related to Semantic Terms: Issue #: https://github.com/mpi-forum/mpi-issues/issues/96 PR: https://github.com/mpi-forum/mpi-standard/pull/116 There are two versions currently provided for reading, based on the feedback from the Forum we will only read one version, the PDF file for the two versions can be found here: Version-1: https://github.com/mpi-forum/mpi-standard/files/4779984/mpi40-report-520db9… Version-2: https://github.com/mpi-forum/mpi-standard/files/4779988/mpi40-report-f51be4… The difference between the two versions is in the locality of MPI_RSEND and MPI_MRECV functions (that can be found here<https://github.com/mpi-forum/mpi-standard/pull/116/files/520db9c157951605f3…>) and corresponding entries in the table in Appendix A.2 (page 723) in the PDF files above. Annotated PDF files will be available soon under this issue on github. Thanks, Puri (on behalf of the WG)

1 0

Meeting Registration Reminder and Voting Guidelines
by Wesley Bland 24 Jun '20

24 Jun '20

Action Items: 1. Register for the meeting next week if you haven’t already. 2. Confirm that you are able to use the voting link and arrange who will vote from your org. === Hi all, I just wanted to send out one final reminder for people to register for next week’s MPI Forum meeting. There are still a few orgs who usually attend but have not registered. https://forms.gle/65bMTDiUQBmbJgjs9 <https://forms.gle/65bMTDiUQBmbJgjs9> The MPI Forum procedures state that you must both register and attend the meeting before the first voting block in order to be officially counted and to be eligible to vote. The first voting block is Tuesday at 9:30 AM Central US so that means you will need to have attended at least some portion of the Monday meeting and/or the very beginning of the Tuesday meeting in order to be eligible to vote in either voting block at the meeting. If you’re not sure if you’ve registered, you can find the list of people here: https://www.mpi-forum.org/meetings/2020/06/attendance <https://www.mpi-forum.org/meetings/2020/06/attendance> We will also be using the same voting system as from the last exceptional virtual meeting in May (JotForm). My understanding is that you will need to enable Javascript in your browser in order for this to work. If you’d like to double check that the voting page will work for you this time, you can use the voting page from last time using this link: https://form.jotform.com/201253678123047?participantId=a1b2c3d4&name=Test_N… <https://form.jotform.com/201253678123047?participantId=a1b2c3d4&name=Test_N…> When the voting block opens, I’ll be sending an automated email to the address you used to register with a personalized link for your ballot. If you didn’t provide an email address (I forgot to add an email address to the original registration form), I’ll be using the one you’ve usually been using for the MPI Forum. If you want to change the email address that I use, email me directly and let me know. Remember that each organization only gets one vote (not one per person). So for those organizations with more than one representative, please make arrangements ahead of time to determine who will be casting the ballot for your organization. If I get more than one ballot from an organization, I will discard all but one of them (probably the first, maybe the last). As always, if you have any questions or comments or think I’ve made any mistakes, please let me know on email or Slack. Thanks, Wes

1 0

Re: [Mpi-forum] Hardware Topologies Working Group Announcements for the June Meeting
by Rolf Rabenseifner 24 Jun '20

24 Jun '20

Dear all, Ken just detected, that the provided pdf was not annotated. It contained all what is needed for the reading, but annotated is of course better. Therefore here the annotated file: https://github.com/mpi-forum/mpi-issues/files/4823780/mpi-report-issue120-t… My apologies for any inconvenience. Best regards Rolf ----- Original Message ----- > From: "Rolf Rabenseifner" <rabenseifner(a)hlrs.de> > To: "Main MPI Forum mailing list" <mpi-forum(a)lists.mpi-forum.org> > Sent: Monday, June 15, 2020 3:40:35 PM > Subject: Re: [Mpi-forum] Hardware Topologies Working Group Announcements for the June Meeting > Hi all, > >> * A reading for issue #120 >> (https://github.com/mpi-forum/mpi-issues/issues/120) >> (PR#98 https://github.com/mpi-forum/mpi-standard/pull/98) : >> MPI_Cart_create_weighted / Topology aware Cartesian communicators >> This reading will deal with the last changes made since the Portland >> meeting. > > An annotated pdf for the reading can be found at > https://github.com/mpi-forum/mpi-issues/files/4776307/mpi-report-issue120-t… > > Slides with a short summary are at > https://github.com/mpi-forum/mpi-issues/files/4758843/MPIX_2020-06-10-Munic… > > Best regards > Rolf > > ----- Original Message ----- >> From: "Main MPI Forum mailing list" <mpi-forum(a)lists.mpi-forum.org> >> To: "Main MPI Forum mailing list" <mpi-forum(a)lists.mpi-forum.org> >> Cc: "Guillaume Mercier" <guillaume.mercier(a)u-bordeaux.fr> >> Sent: Friday, June 12, 2020 3:26:16 PM >> Subject: [Mpi-forum] Hardware Topologies Working Group Announcements for the >> June Meeting > >> Hi all, >> >> I'd like to make a couple of announcements for the upcoming >> Voting Meeting (aka Exceptional Virtual Meeting) : >> >> *second votes for: >> -PR1#42 (https://github.com/mpi-forum/mpi-standard/pull/142) : Guided >> Mode for Hardware Communicators Creation >> -PR#148 (https://github.com/mpi-forum/mpi-standard/pull/148) Unguided >> Mode for Hardware Communicators Creation >> >> * A reading for issue #120 >> (https://github.com/mpi-forum/mpi-issues/issues/120) (PR#98 >> https://github.com/mpi-forum/mpi-standard/pull/98) : >> MPI_Cart_create_weighted / Topology aware Cartesian communicators >> This reading will deal with the last changes made since the Portland >> meeting. >> >> We will request a no-no votes for this last item too. >> >> Regards >> Guillaume >> _______________________________________________ >> mpi-forum mailing list >> mpi-forum(a)lists.mpi-forum.org >> https://lists.mpi-forum.org/mailman/listinfo/mpi-forum > > -- > Dr. Rolf Rabenseifner . . . . . . . . . .. email rabenseifner(a)hlrs.de . > High Performance Computing Center (HLRS) . phone ++49(0)711/685-65530 . > University of Stuttgart . . . . . . . . .. fax ++49(0)711 / 685-65832 . > Head of Dpmt Parallel Computing . . . www.hlrs.de/people/rabenseifner . > Nobelstr. 19, D-70550 Stuttgart, Germany . . . . (Office: Room 1.307) . -- Dr. Rolf Rabenseifner . . . . . . . . . .. email rabenseifner(a)hlrs.de . High Performance Computing Center (HLRS) . phone ++49(0)711/685-65530 . University of Stuttgart . . . . . . . . .. fax ++49(0)711 / 685-65832 . Head of Dpmt Parallel Computing . . . www.hlrs.de/people/rabenseifner . Nobelstr. 19, D-70550 Stuttgart, Germany . . . . (Office: Room 1.307) .

1 0

No virtual MPI forum tomorrow / Wednesday this week
by Martin Schulz 23 Jun '20

23 Jun '20

Hi all, As Wesley already mentioned last week, the plan is to cancel this week’s virtual meeting. If there are any issues left that need to be addressed before the full voting forum meeting next week, please let Wesley and me know. I hope to see/hear all of you Monday in our brave new world of virtual voting meetings, 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

Reading of issue #160: Clarify behavior for invalid MPI_T binding arguments
by Marc-André Hermanns 15 Jun '20

15 Jun '20

Dear all, I would like to read ticket #160 at the June meeting: Title: https://github.com/mpi-forum/mpi-standard/pull/161 Issue: https://github.com/mpi-forum/mpi-issues/issues/160 PR: https://github.com/mpi-forum/mpi-standard/pull/161 Cheers, Marc-Andre -- Dr. rer. nat. Marc-André Hermanns IT Center Group: High Performance Computing Division: Computational Science and Engineering RWTH Aachen University Seffenter Weg 23 52074 Aachen Phone: +49 241 80-24381 hermanns(a)itc.rwth-aachen.de www.itc.rwth-aachen.de

1 0