May 2022 - mpi-forum - mailman3.cels.anl.gov

1st Future of MPI RMA Workshop: Call for Short Talks and Participation
by Joseph Schuchart 29 May '22

29 May '22

[Apologies if you got multiple copies of this email.] *1st Future of MPI RMA Workshop (FoRMA'22)* https://mpiwg-rma.github.io/forma22/ The MPI RMA Working Group is organizing a workshop aimed at gathering inputs from users and implementors of MPI RMA with past experiences and ideas for improvements, as well as hardware vendors to discuss future developments of high-performance network hardware relevant to MPI RMA. The goal is to evaluate the current design of MPI RMA, to learn from the current state of practice, and to start the process of rethinking the design of one-sided communication in the MPI standard. The workshop will be held entirely virtual on *June 16 & 17* and the results and contributions will be combined into a joint white paper. *Call for Short Talks* We are seeking input from users of MPI RMA willing to share their experiences, results, and lessons learned in short talks of 10-15 minutes. No stand-alone publications are required. In particular, the following topics for contributions are of interest: - Successful and unsuccessful attempts of using MPI RMA to improve the communication efficiency of applications and middle-ware; - Challenges in implementing and porting applications and middle-ware on top of MPI RMA; - Features that are missing from MPI RMA. If interested, please send a short email with the title of the talk toschuchart(a)icl.utk.edu. *Call for Participation* We encourage everyone interested in one-sided communication models in general and MPI RMA in particular to join this two day workshop and contribute to its success through questions and comments. We encourage a lively and open exchange of ideas and discussions between the speakers and the audience. The connections information will be posted before the workshop athttps://mpiwg-rma.github.io/forma22/. Please direct any questions toschuchart(a)icl.utk.edu. *Registration* To register for the workshop (and receive the access link) please use the registration site at https://tennessee.zoom.us/meeting/register/tJ0qduChrDgsGNdEG3MQeLB-DH3lZ6r-…. All participation is free. *Organizing committee* Joseph Schuchart, University of Tennessee, Knoxville James Dinan, Nvidia Inc. Bill Gropp, University of Illinois Urbana Champaign

1 0

FEEDBACK REQUESTED: Upcoming MPI Forum Meetings
by Martin Schulz 28 May '22

28 May '22

Hi all, At the end of the forum meeting we also had an initial discussion on how to proceed with the MPI Forum meeting series, now that the pandemic is at least (hopefully) nearing its end. First, the next meeting will be 9/28-9/30 and it will be held as a hybrid meeting, i.e., both online and in person after EuroMPI in Chattanooga – following our rules, this will again be an “exceptional virtual voting meeting”, i.e., the same rules will apply as for the fully virtual meeting this week. What will change, though, is that we will follow a timing for the sessions that matches the local setup. However, we do know that this will create issues for folks joining from other timezones and we will take this into account. As for the following meetings: there seemed to be a consensus that our normal cadence of 4 voting meetings per year is still appropriate (please speak up if you disagree). Following our usual timeline, the following dates/weeks could work: Winter 2022: December 5-8, 2022 Spring 2023: February 27-March 2, 2023 Summer 2023: May 29-June 1, 2023 Fall 2023: together with EuroMPI in September (Dates tbd.) Note, this are just proposals! Please check if there are any substantial conflicts – we will adjust the dates if there are such and we will send a final proposal for dates before deciding on final dates. However, please send possible conflicts rather sooner than later, to give us time to react. Third, we had a discussion on whether to return to all or part in Person meetings – both for the tail end of the pandemic, but also afterwards. The general consensus from the attendees this week was that we need some in person meetings again (1-2 per year), but neither fully virtual nor fully only in person are good options. Also, there was a strong push towards allowing hybrid participation of some sort, possibly coupled with different voting roles. However, it was also clear that hybrid meetings will cause technical challenges. As a consequence, we decided to wait out the first hybrid meeting in September and then decide on how to proceed after that. The December meeting will be virtual for sure, with a possible hybrid option for the Spring meeting. If we do want to go hybrid for the spring, though, we would need volunteers that could host us in their organization and that have good video conference setups (similar to what we had at Cisco or Microsoft) in order to support such a meeting. If you are interested and willing to host the MPI forum (for the spring meeting or also in general), please let Wesley and me know. I think this covers the discussion at the forum meeting – if I missed something, or if you were not able to attend the discussion this week and would like to comment, please follow up here. 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

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