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

ACTION ITEM: timeline for MPI 4.1
by Martin Schulz 28 May '22

28 May '22

Hi all, As mentioned during the forum meeting, we should start thinking about the timeline for MPI 4.1. Our goal was to produce a relatively quick turn-around version of the standard, with a big focus on clean-up and fixed terminology and only minor technical additions. While we have made good progress in this direction, several important items are still outstanding. I would like to use the September meeting to make a final collection of these items and with that start fixing a timeline that we can plan with (keep in mind, we need several voting meetings before the last idea being read and when a standard can be ratified). To enable this, it would be helpful if each working group and chapter committee could provide a list of items they still consider outstanding and needed for MPI 4.1. Also I would like everyone to take a look at the MPI 4.1 project board at https://github.com/mpi-forum/mpi-issues/projects/3 and check whether you have items there and whether they are still in the MPI 4.1 range. I would like to get a preliminary assessment on this by mid July, so we can have a first discussion at the July 20th virtual meeting. I hope this gives enough time – please let me know if you have questions or concerns! Thanks, Martin PS: Please also check the chapter committee membership, which is listed on the MPI 4.1 page of the forum website. If would like to make any changes or sign up for particular CCs, please let me know. Thanks! -- 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

ACTION ITEM: Virtual MPI Forum Meetings until September
by Martin Schulz 28 May '22

28 May '22

Hi all, Thanks to all for a productive MPI Forum meeting this week! I think we made good progress on several large items. At the end of the forum, though, we ran out of time to discuss the upcoming virtual meetings. As we have seen in the past, these are more effective when we plan them long term pre-assign slots. Hence, I would like to ask everyone to think about whether you would like a slot until September and if so, please send that request to me (topic, date, speaker). The following slots are currently available: June 15th June 22nd June 29th July 6th July 13th July 27th August 3rd August 10th August 17th August 24th August 31st September 7th September 14th (after the two week deadline) September 21th (after the two week deadline) Please send requests early, so everyone can prepare and note this on their calendars. 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] Collecting Slides From This Week's Meeting
by Aurelien Bouteiller 26 May '22

26 May '22

Wesley, here’s what I presented during the Fine Grain Recovery reading > On May 26, 2022, at 11:09, Wes Bland via mpi-forum <mpi-forum(a)lists.mpi-forum.org> wrote: > > Hi all, > > For anyone who has presented slides this week, please pass them along to me so I can post them in the repository. > > Thanks, > Wes > _______________________________________________ > mpi-forum mailing list > mpi-forum(a)lists.mpi-forum.org > https://lists.mpi-forum.org/mailman/listinfo/mpi-forum

1 0

Collecting Slides From This Week's Meeting
by Wes Bland 26 May '22

26 May '22

Hi all, For anyone who has presented slides this week, please pass them along to me so I can post them in the repository. Thanks, Wes

1 0

Re: [Mpi-forum] Rusty Lusk
by Bronis R. de Supinski 23 May '22

23 May '22

I agree completely. His passing is a loss for us all. On Tue, 24 May 2022, Panda, Dhabaleswar via mpi-forum wrote: > Sorry to know about this sad news. > > As Jack correctly indicated, without Rusty’s contributions, MPI wouldn’t be what it is today. Through the MPI standard, he had immense contributions to the field of Programming Models for HPC systems. I had the pleasure of many years working with him on the PModels1 and PModels2 projects. > > May his soul rest in peace. > > The HPC and MPI communities will miss him. > > Best Regards, > > DK Panda > > > > From: mpi-forum <mpi-forum-bounces(a)lists.mpi-forum.org> On Behalf Of William Emmanuel S. Yu via mpi-forum > Sent: Monday, May 23, 2022 6:36 PM > To: Main MPI Forum mailing list <mpi-forum(a)lists.mpi-forum.org> > Cc: William Emmanuel S. Yu <wyu(a)ateneo.edu> > Subject: Re: [Mpi-forum] Rusty Lusk > > May he rest in peace. On Tue, May 24, 2022, 6:34 AM Jack Dongarra via mpi-forum <mpi-forum(a)lists.mpi-forum.org<mailto:[email protected]>> wrote: It's with a heavy heart I pass along some sad news. My dear friend, Rusty Lusk passed away on May 16. He began > > May he rest in peace. > > On Tue, May 24, 2022, 6:34 AM Jack Dongarra via mpi-forum <mpi-forum(a)lists.mpi-forum.org<mailto:[email protected]>> wrote: > It's with a heavy heart I pass along some sad news. My dear friend, Rusty Lusk passed away on May 16. He began his career as an assistant professor of mathematics at Northern Illinois University. He later moved into the Computer Science Department, where he eventually became a full professor and the department's acting chairman. In 1982, Rusty joined Argonne National Laboratory as a computer scientist, and he became a senior computer scientist in 1989. Along with Paul Messina and me, Rusty was instrumental in founding Argonne's Advanced Computing Research Facility, which signaled the beginning of Argonne's commitment to parallel computing research. > > Rusty was one of the founders of the MPI Forum. He Chaired MPI-2, started the MPICH project with Bill Gropp, and was a close friend, collaborator, and mentor to many of us. > > Rusty has said, "I think that the development of the MPI Standard is probably the most important thing that I have worked on. Developing a community standard in cooperation with vendors, computer scientists, and users was fascinating to participate in. The experience was social and technical and succeeded because of the many people who worked hard to achieve a common goal. It was then possible to work on implementation research for MPI to convince a broad range of people that the standard was efficient to implement and use. We realized that we would be impacting a whole generation of parallel programmers if we succeeded." > > Without Rusty's influence, MPI wouldn't be what it is today. I miss my friend. > > Jack Dongarra > > https://urldefense.us/v3/__https://obituaries.neptunesociety.com/obituaries… <https://urldefense.us/v3/__https://urldefense.com/v3/__https:/obituaries.ne… > > > ********************************************************** > Prof. Jack Dongarra; Innovative Computing Laboratory; EECS Department; > 1122 Volunteer Blvd; University of Tennessee; Knoxville TN 37996-3450; > +1-865-974-8295; dongarra(a)icl.utk.edu<mailto:[email protected]>; https://urldefense.us/v3/__http://www.cs.utk.edu/*dongarra/__;fg!!G2kpM7uM-… <https://urldefense.us/v3/__https://urldefense.com/v3/__http:/www.cs.utk.edu… > > > > > _______________________________________________ > mpi-forum mailing list > mpi-forum(a)lists.mpi-forum.org<mailto:[email protected]> > https://urldefense.us/v3/__https://lists.mpi-forum.org/mailman/listinfo/mpi… <https://urldefense.us/v3/__https://urldefense.com/v3/__https:/lists.mpi-for… > > > ________________________________ > > CONFIDENTIALITY NOTICE: This email, including any attachments, is only intended for the use of the addressee. It may contain confidential or privileged information, including personal data, which should not be used or disclosed without proper authorization. If you received this email in error, contact the sender immediately and permanently delete it from your system. Unless otherwise indicated, the contents of this email do not necessarily reflect the views or policies of the Ateneo de Manila University. > > General questions or inquiries (info(a)ateneo.edu<mailto:[email protected]>) | Data privacy concerns (info.udpo(a)ateneo.edu<mailto:[email protected]>) | Complaints or security incidents involving data privacy (alert.udpo(a)ateneo.edu<mailto:[email protected]>) >

1 0

Re: [Mpi-forum] Rusty Lusk
by Panda, Dhabaleswar 23 May '22

23 May '22

Sorry to know about this sad news. As Jack correctly indicated, without Rusty’s contributions, MPI wouldn’t be what it is today. Through the MPI standard, he had immense contributions to the field of Programming Models for HPC systems. I had the pleasure of many years working with him on the PModels1 and PModels2 projects. May his soul rest in peace. The HPC and MPI communities will miss him. Best Regards, DK Panda From: mpi-forum <mpi-forum-bounces(a)lists.mpi-forum.org> On Behalf Of William Emmanuel S. Yu via mpi-forum Sent: Monday, May 23, 2022 6:36 PM To: Main MPI Forum mailing list <mpi-forum(a)lists.mpi-forum.org> Cc: William Emmanuel S. Yu <wyu(a)ateneo.edu> Subject: Re: [Mpi-forum] Rusty Lusk May he rest in peace. On Tue, May 24, 2022, 6:34 AM Jack Dongarra via mpi-forum <mpi-forum(a)lists.mpi-forum.org<mailto:[email protected]>> wrote: It's with a heavy heart I pass along some sad news. My dear friend, Rusty Lusk passed away on May 16. He began May he rest in peace. On Tue, May 24, 2022, 6:34 AM Jack Dongarra via mpi-forum <mpi-forum(a)lists.mpi-forum.org<mailto:[email protected]>> wrote: It's with a heavy heart I pass along some sad news. My dear friend, Rusty Lusk passed away on May 16. He began his career as an assistant professor of mathematics at Northern Illinois University. He later moved into the Computer Science Department, where he eventually became a full professor and the department's acting chairman. In 1982, Rusty joined Argonne National Laboratory as a computer scientist, and he became a senior computer scientist in 1989. Along with Paul Messina and me, Rusty was instrumental in founding Argonne's Advanced Computing Research Facility, which signaled the beginning of Argonne's commitment to parallel computing research. Rusty was one of the founders of the MPI Forum. He Chaired MPI-2, started the MPICH project with Bill Gropp, and was a close friend, collaborator, and mentor to many of us. Rusty has said, "I think that the development of the MPI Standard is probably the most important thing that I have worked on. Developing a community standard in cooperation with vendors, computer scientists, and users was fascinating to participate in. The experience was social and technical and succeeded because of the many people who worked hard to achieve a common goal. It was then possible to work on implementation research for MPI to convince a broad range of people that the standard was efficient to implement and use. We realized that we would be impacting a whole generation of parallel programmers if we succeeded." Without Rusty's influence, MPI wouldn't be what it is today. I miss my friend. Jack Dongarra https://obituaries.neptunesociety.com/obituaries/downers-grove-il/ewing-lus…<https://urldefense.com/v3/__https:/obituaries.neptunesociety.com/obituaries…> ********************************************************** Prof. Jack Dongarra; Innovative Computing Laboratory; EECS Department; 1122 Volunteer Blvd; University of Tennessee; Knoxville TN 37996-3450; +1-865-974-8295; dongarra(a)icl.utk.edu<mailto:[email protected]>; http://www.cs.utk.edu/~dongarra/<https://urldefense.com/v3/__http:/www.cs.utk.edu/*dongarra/__;fg!!KGKeukY!3…> _______________________________________________ mpi-forum mailing list mpi-forum(a)lists.mpi-forum.org<mailto:[email protected]> https://lists.mpi-forum.org/mailman/listinfo/mpi-forum<https://urldefense.com/v3/__https:/lists.mpi-forum.org/mailman/listinfo/mpi…> ________________________________ CONFIDENTIALITY NOTICE: This email, including any attachments, is only intended for the use of the addressee. It may contain confidential or privileged information, including personal data, which should not be used or disclosed without proper authorization. If you received this email in error, contact the sender immediately and permanently delete it from your system. Unless otherwise indicated, the contents of this email do not necessarily reflect the views or policies of the Ateneo de Manila University. General questions or inquiries (info(a)ateneo.edu<mailto:[email protected]>) | Data privacy concerns (info.udpo(a)ateneo.edu<mailto:[email protected]>) | Complaints or security incidents involving data privacy (alert.udpo(a)ateneo.edu<mailto:[email protected]>)

1 0

Re: [Mpi-forum] Rusty Lusk
by William Emmanuel S. Yu 23 May '22

23 May '22

May he rest in peace. On Tue, May 24, 2022, 6:34 AM Jack Dongarra via mpi-forum < mpi-forum(a)lists.mpi-forum.org> wrote: > It's with a heavy heart I pass along some sad news. My dear friend, Rusty > Lusk passed away on May 16. He began his career as an assistant professor > of mathematics at Northern Illinois University. He later moved into the > Computer Science Department, where he eventually became a full professor > and the department's acting chairman. In 1982, Rusty joined Argonne > National Laboratory as a computer scientist, and he became a senior > computer scientist in 1989. Along with Paul Messina and me, Rusty was > instrumental in founding Argonne's Advanced Computing Research Facility, > which signaled the beginning of Argonne's commitment to parallel computing > research. > > Rusty was one of the founders of the MPI Forum. He Chaired MPI-2, started > the MPICH project with Bill Gropp, and was a close friend, collaborator, > and mentor to many of us. > > Rusty has said, "I think that the development of the MPI Standard is > probably the most important thing that I have worked on. Developing a > community standard in cooperation with vendors, computer scientists, and > users was fascinating to participate in. The experience was social and > technical and succeeded because of the many people who worked hard to > achieve a common goal. It was then possible to work on implementation > research for MPI to convince a broad range of people that the standard was > efficient to implement and use. We realized that we would be impacting a > whole generation of parallel programmers if we succeeded." > > Without Rusty's influence, MPI wouldn't be what it is today. I miss my > friend. > > Jack Dongarra > > > https://obituaries.neptunesociety.com/obituaries/downers-grove-il/ewing-lus… > > > ********************************************************** > Prof. Jack Dongarra; Innovative Computing Laboratory; EECS Department; > 1122 Volunteer Blvd; University of Tennessee; Knoxville TN 37996-3450; > +1-865-974-8295; dongarra(a)icl.utk.edu; http://www.cs.utk.edu/~dongarra/ > > > > > _______________________________________________ > mpi-forum mailing list > mpi-forum(a)lists.mpi-forum.org > https://lists.mpi-forum.org/mailman/listinfo/mpi-forum > -- CONFIDENTIALITY NOTICE: This email, including any attachments, is only intended for the use of the addressee. It may contain confidential or privileged information, including personal data, which should not be used or disclosed without proper authorization. If you received this email in error, contact the sender immediately and permanently delete it from your system. Unless otherwise indicated, the contents of this email do not necessarily reflect the views or policies of the Ateneo de Manila University. General questions or inquiries (info(a)ateneo.edu <mailto:[email protected]>) | Data privacy concerns (info.udpo(a)ateneo.edu <mailto:[email protected]>) | Complaints or security incidents involving data privacy (alert.udpo(a)ateneo.edu <mailto:[email protected]>)

1 0

Rusty Lusk
by Jack Dongarra 23 May '22

23 May '22

It's with a heavy heart I pass along some sad news. My dear friend, Rusty Lusk passed away on May 16. He began his career as an assistant professor of mathematics at Northern Illinois University. He later moved into the Computer Science Department, where he eventually became a full professor and the department's acting chairman. In 1982, Rusty joined Argonne National Laboratory as a computer scientist, and he became a senior computer scientist in 1989. Along with Paul Messina and me, Rusty was instrumental in founding Argonne's Advanced Computing Research Facility, which signaled the beginning of Argonne's commitment to parallel computing research. Rusty was one of the founders of the MPI Forum. He Chaired MPI-2, started the MPICH project with Bill Gropp, and was a close friend, collaborator, and mentor to many of us. Rusty has said, "I think that the development of the MPI Standard is probably the most important thing that I have worked on. Developing a community standard in cooperation with vendors, computer scientists, and users was fascinating to participate in. The experience was social and technical and succeeded because of the many people who worked hard to achieve a common goal. It was then possible to work on implementation research for MPI to convince a broad range of people that the standard was efficient to implement and use. We realized that we would be impacting a whole generation of parallel programmers if we succeeded." Without Rusty's influence, MPI wouldn't be what it is today. I miss my friend. Jack Dongarra https://obituaries.neptunesociety.com/obituaries/downers-grove-il/ewing-lus… <https://obituaries.neptunesociety.com/obituaries/downers-grove-il/ewing-lus…> ********************************************************** Prof. Jack Dongarra; Innovative Computing Laboratory; EECS Department; 1122 Volunteer Blvd; University of Tennessee; Knoxville TN 37996-3450; +1-865-974-8295; dongarra(a)icl.utk.edu; http://www.cs.utk.edu/~dongarra/

1 0

MPI Forum This Week!
by Martin Schulz 22 May '22

22 May '22

Hi all, Just as a quick reminder: our next (exceptional voting) MPI Forum meeting is this week, starting tomorrow May 23 and will run until May 26. On all days, we will run from 9am US CT to 1pm US CT. The details, including dial in data, is at: https://www.mpi-forum.org/meetings/2022/05/agenda Note: the link is a different link specifically for this meeting. On May 23 we will have WG meetings – this time, this will be in one track. The rough split is: 9:00 HW Topologies 10:00 Hybrid/Accelerators 11:00 Sessions However, slight shifts may occur at the common discretion of the WG leads. The other three days are intended for plenaries, as listed on the agenda. We have several large reading scheduled we would like to get through. Hence, it would be good if for those readings everyone could work with the github versions of the PRs during the reading, so that small change suggestions can be made directly into the PR and will not derail the overall readings. I am sure the “readers” will appreciate this and it gives us a better chance to focus on the conceptual issues in the discussion. 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

No Virtual MPI Forum Meeting tomorrow / September dates changed
by Martin Schulz 17 May '22

17 May '22

Hi all, Tomorrow’s virtual MPI forum meeting is cancelled. We will meet for the voting meeting the following week. The agenda is online – Monday is WG day, Tue-Thu plenaries. Further, due to a conflict with another large event, the EuroMPI conference (and with that the forum) had to pushed back one week. The conference now starts on September 26, the MPI forum on September 28 and then runs to September 30. The webpages should all be updated by now. See you all next week! 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