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

MPI Forum Starting Tomorrow / Wednesday: Missing Reviews / Chapter Readings
by Martin Schulz 12 Sep '23

12 Sep '23

Hi all, We will be starting tomorrow afternoon (local time) with our MPI Forum RC Meeting. The core will be the reading of the changes made to all chapters, as well as the creation of a list of all outstanding issues. For this, we are still missing a few reviews of the MPI standard – the procedure is laid out below in a previous email by Wesley. If you haven’t had the chance to do this, yet, please complete your section as soon as possible, so that we can get a complete overview. If you would like to volunteer for an open section, please let us know as well. Also, just a reminder for all chapter committee chairs: please let us know if you won’t be reading your chapter diffs who from your committee you have designated to do so. Thanks and see you tomorrow (in Bristol or online), 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 <work(a)wesbland.com> Date: Wednesday, 23. August 2023 at 17:36 To: Amit Ruhela <aruhela(a)tacc.utexas.edu>, Aurelien Bouteiller <bouteill(a)icl.utk.edu>, "benson.muite(a)kichakatokizito.solutions" <benson.muite(a)kichakatokizito.solutions>, William Williams <william.williams(a)mailbox.tu-dresden.de>, "bsmith(a)cornelisnetworks.com" <bsmith(a)cornelisnetworks.com>, Christoph Niethammer <niethammer(a)hlrs.de>, Claudia Blaas-Schenner <claudia.blaas-schenner(a)tuwien.ac.at>, Dan Holmes <daniel.john.holmes(a)intel.com>, "dschafer1(a)unm.edu" <dschafer1(a)unm.edu>, "edgar.gabriel(a)gmail.com" <edgar.gabriel(a)gmail.com>, "mercier(a)labri.fr" <mercier(a)labri.fr>, "Howard Pritchard Jr." <howardp(a)lanl.gov>, Hugo Taboada <Hugo.TABOADA(a)cea.fr>, Ignacio Laguna <ilaguna(a)llnl.gov>, "jbbesnard(a)paratools.com" <jbbesnard(a)paratools.com>, James Dinan <jdinan(a)nvidia.com>, Joseph Schuchart <schuchart(a)icl.utk.edu>, "Julien.JAEGER(a)cea.fr" <Julien.JAEGER(a)cea.fr>, "raffenet(a)anl.gov" <raffenet(a)anl.gov>, "capelli.ludovic(a)gmail.com" <capelli.ludovic(a)gmail.com>, Marc-André Hermanns <hermanns(a)itc.rwth-aachen.de>, Maria Garzaran <maria.garzaran(a)intel.com>, Martin Rüefenacht <m.a.ruefenacht(a)gmail.com>, "martin.schreiber(a)univ-grenoble-alpes.fr" <martin.schreiber(a)univ-grenoble-alpes.fr>, Martin Schulz <schulzm(a)in.tum.de>, Matthew Dosanjh <mdosanj(a)sandia.gov>, "michael.klemm(a)amd.com" <michael.klemm(a)amd.com>, "m.knobloch(a)fz-juelich.de" <m.knobloch(a)fz-juelich.de>, "shineman.5(a)osu.edu" <shineman.5(a)osu.edu>, "jnysal(a)in.ibm.com" <jnysal(a)in.ibm.com>, Puri Bangalore <pvbangalore(a)ua.edu>, Quincey Koziol <qkoziol(a)amazon.com>, "Shipley, Riley Prescott" <riley-shipley(a)utc.edu>, Ryan Grant <ryan.grant(a)queensu.ca>, "sumimoto(a)cc.u-tokyo.ac.jp" <sumimoto(a)cc.u-tokyo.ac.jp>, Thomas Naughton <naughtont(a)ornl.gov>, Tobias Haas <tobias.haas(a)hlrs.de>, "eijkhout(a)tacc.utexas.edu" <eijkhout(a)tacc.utexas.edu>, Wes Bland <work(a)wesbland.com>, William Gropp <wgropp(a)illinois.edu> Subject: Re: Action Required -- MPI 4.1 RC Review Hi folks, I’m sorry that I left out the detail about the page numbers. A couple of you have already asked about this. The document essentially has two sets of page numbers. The first is the one that your PDF reader will likely give and starts at 1. The second is what is printed in the document and starts at i (lowercase “eye”). The original numbers in the sheet were the first set, but I’ve now added another set of columns with the second as well. So you can hopefully decipher which ones are assigned to you from that. When you fill out the form with your response, please use the first set of numbers. Thanks, Wes On Aug 23, 2023, at 10:57 AM, Wes Bland <work(a)wesbland.com> wrote: Hi all, Thanks to each of you for being an active participant in the MPI Forum this year. As we come to the end of the MPI 4.1 ratification process, we have one more ask for each of you who have been most involved. As we discussed during the July 2023 meeting now is the time that we will collectively review the MPI 4.1 RC Document before we go through the final steps for ratification. Each of you who have been active participants over the last year have been assigned a chunk of the document (each chunk is 30 pages). Your assignments can be found on this Google Sheet: https://docs.google.com/spreadsheets/d/16o1Zf0YGGEIHIn9ULp9IodBok2TF2YNKEFH… You can find the document itself here: https://github.com/mpi-forum/mpi-forum.github.io/raw/master/docs/mpi-4.1/mp… As you are reviewing, these are the guidelines for things you should be looking for: 1. Formatting and whitespace problems 2. Spelling errors or other typos 3. Logical inconsistencies in the overall document 4. Problems with the way approved changes were applied You are welcome to note any other issues you see as well. As you finish your review, please use this form (https://forms.gle/93hXANvVkp3MhrCD9) to provide your feedback. I’ll be able to triage issues from there. If you have no notes, please still fill out that form and mention that so we know each chunk has been covered. These reviews must be completed by September 8th (there are no review deadline extensions 😀) so we can discuss them at the meeting the following week. If you cannot complete your assignment by then, please let Martin and I know as soon as possible so we can try to find someone else who is willing to take over your section. If you are willing to look through an extra section, please let us know that too so we can line up extra reviewers just in case. If you have any questions, feel free to reach out to us for clarification. Thanks, Wes

1 0

September 2023 MPI Forum Registration
by Wes Bland 10 Sep '23

10 Sep '23

Hi all, As of now, we still haven’t met quorum yet for the September meeting. I’m sure this is largely because of the confusion of having two registration pages for this meeting. Please double check if you’ve registered for the meeting here: https://www.mpi-forum.org/meetings/2023/09/attendance If you haven’t registered, please do so ASAP here: https://forms.gle/sEeNtbyXxnJndGrU9 Thanks, Wes

1 0

RMA WG Reading Announcement September'23 Meeting
by Joseph Schuchart 31 Aug '23

31 Aug '23

The RMA WG would like to read the following errata: RMA chapter committee changes Issue: https://github.com/mpi-forum/mpi-issues/issues/745 PR: https://github.com/mpi-forum/mpi-standard/pull/857 These changes were somehow missed for the 4.1rc and will likely address many of the issues reviewers will have found. Thanks Joseph

1 0

Reading announcement for ULFM slices 2, 3: AGREE and SHRINK
by Aurelien Bouteiller 31 Aug '23

31 Aug '23

The Error Handling and Fault TOlerance Working Group would like to read the following slices of the ULFM fine-grain fault tolerance ULFM Fault Tolerance (slice 2: agree) #582 https://github.com/mpi-forum/mpi-issues/issues/582 ULFM Fault Tolerance (slice 3: shrink) #583 https://github.com/mpi-forum/mpi-issues/issues/583 This work is slated for MPI-5 and follows upon the (already voted-in) ULFM slice1 Please refer to the diff of the latest commit when reviewing: only the last commit is relevant for each PR (the prior commits pertain to the prior slices). I will try to produce a latexdiff as soon as possible. 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

September MPI Forum Registration and Readings
by Wes Bland 30 Aug '23

30 Aug '23

Hi all, Quick reminder to register for the upcoming meeting. Remember that you need to register twice (I know it’s annoying…). First, register for the conference as an in-person or remote attendee: https://www.eventbrite.co.uk/e/iwomp-2023-eurompi-and-mpi-forum-tickets-666… Second, register for the MPI Forum meeting so I can count you for voting eligibility: https://forms.gle/EMHZgSq7nSJ7gFwk6 It seems like there’s a lot of folks who have registered for the conference, but not for the meeting. So please make sure you do both! Lastly, if you haven’t already looked at it, remember that if you’ve been participating in the forum over the last year or so, you probably have a review assignment for MPI 4.1. I sent those out a little bit ago with instructions so go find that email and make sure you’re completing those reviews ASAP so the chapter chairs can make any edits, if necessary. Thanks, Wes

1 0

reading announcement for pull 875: standard ABI
by Jeff Hammond 29 Aug '23

29 Aug '23

See https://github.com/mpi-forum/mpi-standard/pull/875. I expect to read this more than once but we have to start somewhere. This PR defines the overall structure and specifies many of the ABI features. Many of the choices have strong consensus in the working group. The constants are not listed in tables yet because that's a bunch of LaTeX tedium for which I have not yet had time. Nothing in this PR depends on the values of the constants, and thus they can be added in a second PR. There are open issues regarding Fortran that will be added later, possibly as a different PR, depending on the working group discussion later today. The current proposal assumes one choice of MPI_Aint, which is only relevant in the context of exotic platforms like CHERI that have wide pointers (pointers are bigger than addresses). Jeff -- Jeff Hammond jeff.science(a)gmail.com http://jeffhammond.github.io/

1 0

has anyone ever implemented MPI for a 16-bit OS or x86 real mode or DOS?
by Jeff Hammond 25 Aug '23

25 Aug '23

I am trying to figure if there is any justification for the continued mention of segmented addressing in the standard. Also, it is not clear to me that we have ever had any implementation experience that gives us the expertise to standardize features for it. x86 real mode support was dropped in Windows 3.1. Linux never supported it because it was created for 386. I cannot find any evidence that anyone has ever supported MPI on a 286, or any operating system that supported 16-bit real mode. https://en.wikipedia.org/wiki/Coherent_(operating_system) ran on 286 but didn't support TCP/IP so it's not clear how MPI would have achieved parallelism. https://en.wikipedia.org/wiki/Xenix was discontinued in 1987. What operating system are we using as the reference for reasoning about how MPI works on such systems? https://citeseerx.ist.psu.edu/document?repid=rep1&type=pdf&doi=06e491ccb3ce… has some relevant history. Jeff -- Jeff Hammond jeff.science(a)gmail.com http://jeffhammond.github.io/

1 0

Re: [Mpi-forum] reading announcement for issue 738 / pull 872 (MPI_GET_ADDRESS and C &)
by Jeff Hammond 25 Aug '23

25 Aug '23

I rewrote the text in response to feedback. I just removed the part about segmented addressing, which is no longer relevant. If you think we need to say something about not supporting 16-bit x86 real mode, feel free to propose a change. Jeff On Thu, Aug 24, 2023 at 12:36 PM Jeff Hammond <jeff.science(a)gmail.com> wrote: > https://github.com/mpi-forum/mpi-standard/pull/872, which addresses > https://github.com/mpiwg-abi/mpi-standard/tree/issue-738 > >> > If you want to fight for continued support for segmented addressing or > deny the behavior of & in C99, now is your chance. > > Jeff > > -- > Jeff Hammond > jeff.science(a)gmail.com > http://jeffhammond.github.io/ > -- Jeff Hammond jeff.science(a)gmail.com http://jeffhammond.github.io/

1 0

reading announcement for issue 738 / pull 872 (MPI_GET_ADDRESS and C &)
by Jeff Hammond 24 Aug '23

24 Aug '23

https://github.com/mpi-forum/mpi-standard/pull/872, which addresses https://github.com/mpiwg-abi/mpi-standard/tree/issue-738 > If you want to fight for continued support for segmented addressing or deny the behavior of & in C99, now is your chance. Jeff -- Jeff Hammond jeff.science(a)gmail.com http://jeffhammond.github.io/

1 0

MPI 4.1 Git Repository
by Wes Bland 23 Aug '23

23 Aug '23

Hi all, I want to update folks on the status of the MPI standard document repository. As discussed in the July 2023 meeting, we’re now in “release candidate mode”, which means that between now and the release of the MPI 4.1 document, we won’t be merging any more changes into the mpi-4.x branch until the document has been ratified (or we start the process over). This means that all changes from this point on need to be based on the mpi-41-rc branch. I created the branch this morning and it’s now available in the GitHub repository. From a practical perspective, this means that if you’re making any further edits to address minor issues, you should start from the mpi-41-rc branch on your laptop. When you push your changes to GitHub as a pull request, it will automatically use the mpi-41-rc branch as I’ve made it the default. If there are any outstanding changes that are already in pull requests, I haven’t changed any existing PRs. You can do that yourself, if you’d like, by opening your pull requests’s page, clicking the “edit” button in the top right corner, then clicking the “base” dropdown underneath the PR title. Choose mpi-41-rc and GitHub should do the rest for you and let you know if there’s a problem. If you have any issues, please let me know. If you’d like to see the compiled MPI 4.1 RC1 document, it’s available (or will be in a few minutes as the site was just published) on the docs page on the website: https://www.mpi-forum.org/docs/ Thanks, Wes

1 0