October 2017 - mpiwg-persistence - mailman3.cels.anl.gov

Re: [mpiwg-persistence] Notes from meeting today 25th Oct 2017
by Anthony Skjellum 25 Oct '17

25 Oct '17

Can we do next meeting on nov 2 please I'll respond in detail on these minutes tonight ! Thx Tony Anthony Skjellum, PhD 205-807-4968 On Oct 25, 2017, at 9:44 AM, HOLMES Daniel <d.holmes(a)epcc.ed.ac.uk<mailto:[email protected]>> wrote: Hi all, Notes from today's meeting: 1) extended version of EuroMPI 2017 "Performance Planning" paper for journal special issue [Dan] add state-machine diagram and various sentences throughout section 3 - emphasise life-cycle difference between blocking/nonblocking cache (must be tied to comm) and persistent cache (can be tied to request). [Puri] change format of paper to Elsevier journal layout [Puri] update CG example to one that gave good results [Srinivas] integrate the reference implementation of persistent collectives into MLSL and run a machine learning application use-case - add plot (if possible) and text to paper in results section - focus on correctness/more-natural programming style, i.e. lack of need to duplicate comm per collective to avoid ordering issues 2) further implementation/optimisation work, aiming at new paper (possible target SC18) [Dan] scope out and begin coding work on infrastructure improvement, initially for MPI_Allreduce_init [Puri] code review and testing of PNBC new code Next meeting: Wed 1st Nov 2017 8am Central US via GoToMeeting. Cheers, Dan. -- The University of Edinburgh is a charitable body, registered in Scotland, with registration number SC005336. _______________________________________________ mpiwg-persistence mailing list mpiwg-persistence(a)lists.mpi-forum.org<mailto:[email protected]> https://lists.mpi-forum.org/mailman/listinfo/mpiwg-persistence

1 0

Notes from meeting today 25th Oct 2017
by HOLMES Daniel 25 Oct '17

25 Oct '17

Hi all, Notes from today’s meeting: 1) extended version of EuroMPI 2017 "Performance Planning" paper for journal special issue [Dan] add state-machine diagram and various sentences throughout section 3 - emphasise life-cycle difference between blocking/nonblocking cache (must be tied to comm) and persistent cache (can be tied to request). [Puri] change format of paper to Elsevier journal layout [Puri] update CG example to one that gave good results [Srinivas] integrate the reference implementation of persistent collectives into MLSL and run a machine learning application use-case - add plot (if possible) and text to paper in results section - focus on correctness/more-natural programming style, i.e. lack of need to duplicate comm per collective to avoid ordering issues 2) further implementation/optimisation work, aiming at new paper (possible target SC18) [Dan] scope out and begin coding work on infrastructure improvement, initially for MPI_Allreduce_init [Puri] code review and testing of PNBC new code Next meeting: Wed 1st Nov 2017 8am Central US via GoToMeeting. Cheers, Dan. -- The University of Edinburgh is a charitable body, registered in Scotland, with registration number SC005336.

1 0

Re: [mpiwg-persistence] Next meeting - today - at 8am Central US
by Anthony Skjellum 25 Oct '17

25 Oct '17

Go ahead without me sorry - at a site with no phones this AM Anthony Skjellum, PhD 205-807-4968 On Oct 25, 2017, at 6:42 AM, HOLMES Daniel <d.holmes(a)epcc.ed.ac.uk<mailto:[email protected]>> wrote: Hi all, We have our next teleconference meeting today in a little over 2 hours' time. Connection details below: ?????????????????????????????????????????? Please join my meeting from your computer, tablet or smartphone. https://global.gotomeeting.com/join/357844797 You can also dial in using your phone. United States: +1 (786) 535-3119 Access Code: 357-844-797 More phone numbers France: +33 170 950 590 United Kingdom: +44 330 221 0097 First GoToMeeting? Let's do a quick system check: https://link.gotomeeting.com/system-check ?????????????????????????????????????????? Cheers, Dan. -- The University of Edinburgh is a charitable body, registered in Scotland, with registration number SC005336. _______________________________________________ mpiwg-persistence mailing list mpiwg-persistence(a)lists.mpi-forum.org<mailto:[email protected]> https://lists.mpi-forum.org/mailman/listinfo/mpiwg-persistence -- The University of Edinburgh is a charitable body, registered in Scotland, with registration number SC005336. _______________________________________________ mpiwg-persistence mailing list mpiwg-persistence(a)lists.mpi-forum.org<mailto:[email protected]> https://lists.mpi-forum.org/mailman/listinfo/mpiwg-persistence

1 0

Next meeting - today - at 8am Central US
by HOLMES Daniel 25 Oct '17

25 Oct '17

Hi all, We have our next teleconference meeting today in a little over 2 hours' time. Connection details below: ────────────────────────────────────────── Please join my meeting from your computer, tablet or smartphone. https://global.gotomeeting.com/join/357844797 You can also dial in using your phone. United States: +1 (786) 535-3119 Access Code: 357-844-797 More phone numbers France: +33 170 950 590 United Kingdom: +44 330 221 0097 First GoToMeeting? Let's do a quick system check: https://link.gotomeeting.com/system-check ────────────────────────────────────────── Cheers, Dan. -- The University of Edinburgh is a charitable body, registered in Scotland, with registration number SC005336. _______________________________________________ mpiwg-persistence mailing list mpiwg-persistence(a)lists.mpi-forum.org https://lists.mpi-forum.org/mailman/listinfo/mpiwg-persistence -- The University of Edinburgh is a charitable body, registered in Scotland, with registration number SC005336.

1 0

Re: [mpiwg-persistence] Next meeting - today - at 8am Central US
by Anthony Skjellum 18 Oct '17

18 Oct '17

Ok !!! Anthony Skjellum, PhD 205-807-4968 On Oct 18, 2017, at 8:11 AM, HOLMES Daniel <d.holmes(a)epcc.ed.ac.uk<mailto:[email protected]>> wrote: Hi all, We have our next teleconference meeting today in 49 minutes' time. Connection details below: ?????????????????????????????????????????? Please join my meeting from your computer, tablet or smartphone. https://global.gotomeeting.com/join/357844797 You can also dial in using your phone. United States: +1 (786) 535-3119 Access Code: 357-844-797 More phone numbers France: +33 170 950 590 United Kingdom: +44 330 221 0097 First GoToMeeting? Let's do a quick system check: https://link.gotomeeting.com/system-check ?????????????????????????????????????????? Cheers, Dan. -- The University of Edinburgh is a charitable body, registered in Scotland, with registration number SC005336. _______________________________________________ mpiwg-persistence mailing list mpiwg-persistence(a)lists.mpi-forum.org<mailto:[email protected]> https://lists.mpi-forum.org/mailman/listinfo/mpiwg-persistence

1 0

Next meeting - today - at 8am Central US
by HOLMES Daniel 18 Oct '17

18 Oct '17

Hi all, We have our next teleconference meeting today in 49 minutes’ time. Connection details below: ────────────────────────────────────────── Please join my meeting from your computer, tablet or smartphone. https://global.gotomeeting.com/join/357844797 You can also dial in using your phone. United States: +1 (786) 535-3119 Access Code: 357-844-797 More phone numbers France: +33 170 950 590 United Kingdom: +44 330 221 0097 First GoToMeeting? Let's do a quick system check: https://link.gotomeeting.com/system-check ────────────────────────────────────────── Cheers, Dan. -- The University of Edinburgh is a charitable body, registered in Scotland, with registration number SC005336.

1 0

Re: [mpiwg-persistence] Next meeting - today at 8am Central US
by Skjellum, Anthony 11 Oct '17

11 Oct '17

Yes — Anthony Skjellum, PhD 205-807-4968 On Oct 11, 2017, at 4:20 AM, HOLMES Daniel <d.holmes(a)epcc.ed.ac.uk<mailto:[email protected]>> wrote: Hi Tony, I believe the Persistence WG is due to meet next at 8am Central US today? I have created a GoToMeeting event for that - connection details below. Cheers, Dan. ────────────────────────────────────────── Please join my meeting from your computer, tablet or smartphone. https://global.gotomeeting.com/join/758624325 You can also dial in using your phone. United States: +1 (669) 224-3319 Access Code: 758-624-325 More phone numbers France: +33 170 950 590 United Kingdom: +44 330 221 0097 First GoToMeeting? Let's do a quick system check: https://link.gotomeeting.com/system-check ────────────────────────────────────────── -- The University of Edinburgh is a charitable body, registered in Scotland, with registration number SC005336.

1 0

Next meeting - today at 8am Central US
by HOLMES Daniel 11 Oct '17

11 Oct '17

Hi Tony, I believe the Persistence WG is due to meet next at 8am Central US today? I have created a GoToMeeting event for that - connection details below. Cheers, Dan. ────────────────────────────────────────── Please join my meeting from your computer, tablet or smartphone. https://global.gotomeeting.com/join/758624325 You can also dial in using your phone. United States: +1 (669) 224-3319 Access Code: 758-624-325 More phone numbers France: +33 170 950 590 United Kingdom: +44 330 221 0097 First GoToMeeting? Let's do a quick system check: https://link.gotomeeting.com/system-check ────────────────────────────────────────── -- The University of Edinburgh is a charitable body, registered in Scotland, with registration number SC005336.

1 0

Persistent collectives progress update
by HOLMES Daniel 04 Oct '17

04 Oct '17

FYI Cheers, Dan. Begin forwarded message: From: Dan Holmes <dholmes(a)epcc.ed.ac.uk<mailto:[email protected]>> Date: 3 October 2017 at 13:50:10 BST To: "Sridharan, Srinivas" <srinivas.sridharan(a)intel.com<mailto:[email protected]>> Hi Srinivas, The paper presentation went very well. There is definitely community interest and support for this effort. Two of the other papers presented detailed implementation optimisations that require the persistent collective interface, one from a theoretical point-of-view with a proof-of-concept code, and the other with a more practical approach using the profiling intercept mechanism to override the existing MPI implementation of particular collective functions. Both demonstrated improved performance derived from performance planning - perfect use-cases for our interface work. The formal reading at the MPI Forum prompted a variety of small wording changes to the proposed text and a couple of small semantic changes - at least one which requires further design work in the working group. This will likely reset the procedure, i.e. we will probably be required to formally read the modified proposal again at the next meeting with the additional semantic change included. That issue is around freeing the persistent collective request object - the current MPI_REQUEST_FREE function is local (does not rely on MPI calls elsewhere, specifically this means *not* collective) and nonblocking (it may not have actually freed resources when it returns to user code). There is a good argument for adding a new function that gives a stronger guarantee and requires a stronger usage model, i.e. a blocking collective request free function. There is a design space of 8 or 12 possible combinations of the fundamental semantics, which we should explore. Tony/Bradley: I will let you guys comment on where we are at with sharing our reference implementation code with Srinivas and his team. Cheers, Dan. On 3 Oct 2017, at 13:30, Sridharan, Srinivas <srinivas.sridharan(a)intel.com<mailto:[email protected]>> wrote: Hi Dan, Any interesting comments on the paper at EuroMPI? Also, seems like there was supposed to be a reading of persistent collectives. How did it go? http://mpi-forum.org/meetings/2017/09/agenda Finally, please let me know if you have any updates on sharing your code with me… Srinivas The University of Edinburgh is a charitable body, registered in Scotland, with registration number SC005336.

1 0

Re: [mpiwg-persistence] mpiwg-persistence Digest, Vol 2, Issue 1
by Anthony Skjellum 04 Oct '17

04 Oct '17

Thank you. On Tue, Oct 3, 2017 at 8:26 AM, HOLMES Daniel <d.holmes(a)epcc.ed.ac.uk> wrote: > Hi Tony (et al), > > I have moved the entry in my calendar for this week to tomorrow. > > I have created a one-off GoToMeeting event - the connection details are > below. > > ────────────────────────────────────────── > Please join my meeting from your computer, tablet or smartphone. > https://global.gotomeeting.com/join/866195709 > > You can also dial in using your phone. > United States: +1 (646) 749-3129 <(646)%20749-3129> > > Access Code: 866-195-709 > > More phone numbers > France: +33 170 950 594 > United Kingdom: +44 330 221 0088 <+44%20330%20221%200088> > > First GoToMeeting? Try a test session: http://link.gotomeeting.com/ > email-welcome > ────────────────────────────────────────── > > I can ask JeffS if he will create a weekly webex (as used by all other > WGs) for us starting from Tuesday 10th Oct. > > Cheers, > Dan. > > > On 3 Oct 2017, at 13:17, Anthony Skjellum <skjellum(a)gmail.com> wrote: > > Daniel, I apologize, I can't do it this AM... let's hold it Wednesday > this time (tomorrow @ 8am central); And we will do the regular call on > Tuesday starting next week. > We are all at an annual meeting ... > > It would be great if we could use your webex... what do you think? > > Tony > > > On Tue, Oct 3, 2017 at 8:02 AM, HOLMES Daniel <d.holmes(a)epcc.ed.ac.uk> > wrote: > >> Hi Tony, >> >> > I think we are looking at 8am Central on Tuesdays right now; >> >> So, we’re meeting *today in 1 hour* from now? >> >> Do we have connection details for a teleconference? If not, I can setup a >> GoToMeeting fairly quickly. >> >> Cheers, >> Dan. >> -- >> The University of Edinburgh is a charitable body, registered in >> Scotland, with registration number SC005336. >> >> _______________________________________________ >> mpiwg-persistence mailing list >> mpiwg-persistence(a)lists.mpi-forum.org >> https://lists.mpi-forum.org/mailman/listinfo/mpiwg-persistence >> > > > > -- > Anthony Skjellum, PhD > skjellum(a)gmail.com > Cell: +1-205-807-4968 <(205)%20807-4968> > > > > > The University of Edinburgh is a charitable body, registered in > Scotland, with registration number SC005336. > > -- Anthony Skjellum, PhD skjellum(a)gmail.com Cell: +1-205-807-4968

1 0