May 2010 - mpiwg-hybridpm - mailman3.cels.anl.gov

Re: [Mpi3-hybridpm] Reminder for the telecon tomorrow
by Pavan Balaji 25 May '10

25 May '10

Folks, Apologies. I was unexpectedly stuck in the German consulate for more than 3 hours, and couldn't get out on time for this call. Doug: I looked through the proposal and I think we covered all of the details at the last Forum. So, other than an implementation, I don't think there's anything new we need to discuss, correct? The only touchy issue is the use of this call when MPI_THREAD_MULTIPLE is not provided (e.g., using multiple helper threads in MPI_THREAD_FUNNELED mode). I plan to bring this up at the Forum next time. -- Pavan On 05/25/2010 07:39 AM, Douglas Miller wrote: > I would like to finalize the helper threads proposal, so if everyone has > looked at the latest copy it might be worthwhile to talk about it briefly. > I would not expect that the entire hour is needed. > > > _______________________________________________ > Douglas Miller BlueGene Messaging Development > IBM Corp., Rochester, MN USA Bldg 030-2 A410 > dougmill(a)us.ibm.com Douglas Miller/Rochester/IBM > > > > Pavan Balaji > <balaji(a)mcs.anl.g > ov> To > Sent by: mpi3-hybridpm(a)lists.mpi-forum.org > mpi3-hybridpm-bou cc > nces(a)lists.mpi-fo > rum.org Subject > Re: [Mpi3-hybridpm] Reminder for > the telecon tomorrow > 05/24/2010 06:48 > PM > > > Please respond to > mpi3-hybridpm@lis > ts.mpi-forum.org > > > > > > > > In that case shall we cancel the telecon? > > Doug: would you like to discuss anything on the helper threads proposal? > > -- Pavan > > On 05/24/2010 06:34 PM, Snir, Marc wrote: >> I shall be on a plane -- sorry. >> >> On May 24, 2010, at 4:35 PM, Pavan Balaji wrote: >> >>> We'll have our usual telecon tomorrow. The telecon details are available > >>> here: https://svn.mpi-forum.org/trac/mpi-forum-web/wiki/MPI3Hybrid >>> >>> (the telecon information has changed from the previous meetings). >>> >>> -- Pavan >>> >>> -- >>> Pavan Balaji >>> http://www.mcs.anl.gov/~balaji >>> _______________________________________________ >>> Mpi3-hybridpm mailing list >>> Mpi3-hybridpm(a)lists.mpi-forum.org >>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-hybridpm >> Marc Snir >> >> >> >> >> >> _______________________________________________ >> Mpi3-hybridpm mailing list >> Mpi3-hybridpm(a)lists.mpi-forum.org >> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-hybridpm > > -- > Pavan Balaji > http://www.mcs.anl.gov/~balaji > _______________________________________________ > Mpi3-hybridpm mailing list > Mpi3-hybridpm(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-hybridpm > > > _______________________________________________ > Mpi3-hybridpm mailing list > Mpi3-hybridpm(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-hybridpm -- Pavan Balaji http://www.mcs.anl.gov/~balaji

1 0

Re: [Mpi3-hybridpm] Reminder for the telecon tomorrow
by Douglas Miller 25 May '10

25 May '10

I would like to finalize the helper threads proposal, so if everyone has looked at the latest copy it might be worthwhile to talk about it briefly. I would not expect that the entire hour is needed. _______________________________________________ Douglas Miller BlueGene Messaging Development IBM Corp., Rochester, MN USA Bldg 030-2 A410 dougmill(a)us.ibm.com Douglas Miller/Rochester/IBM Pavan Balaji <balaji(a)mcs.anl.g ov> To Sent by: mpi3-hybridpm(a)lists.mpi-forum.org mpi3-hybridpm-bou cc nces(a)lists.mpi-fo rum.org Subject Re: [Mpi3-hybridpm] Reminder for the telecon tomorrow 05/24/2010 06:48 PM Please respond to mpi3-hybridpm@lis ts.mpi-forum.org In that case shall we cancel the telecon? Doug: would you like to discuss anything on the helper threads proposal? -- Pavan On 05/24/2010 06:34 PM, Snir, Marc wrote: > I shall be on a plane -- sorry. > > On May 24, 2010, at 4:35 PM, Pavan Balaji wrote: > >> We'll have our usual telecon tomorrow. The telecon details are available >> here: https://svn.mpi-forum.org/trac/mpi-forum-web/wiki/MPI3Hybrid >> >> (the telecon information has changed from the previous meetings). >> >> -- Pavan >> >> -- >> Pavan Balaji >> http://www.mcs.anl.gov/~balaji >> _______________________________________________ >> Mpi3-hybridpm mailing list >> Mpi3-hybridpm(a)lists.mpi-forum.org >> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-hybridpm > > Marc Snir > > > > > > _______________________________________________ > Mpi3-hybridpm mailing list > Mpi3-hybridpm(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-hybridpm -- Pavan Balaji http://www.mcs.anl.gov/~balaji _______________________________________________ Mpi3-hybridpm mailing list Mpi3-hybridpm(a)lists.mpi-forum.org http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-hybridpm

1 0

Re: [Mpi3-hybridpm] Reminder for the telecon tomorrow
by Pavan Balaji 24 May '10

24 May '10

In that case shall we cancel the telecon? Doug: would you like to discuss anything on the helper threads proposal? -- Pavan On 05/24/2010 06:34 PM, Snir, Marc wrote: > I shall be on a plane -- sorry. > > On May 24, 2010, at 4:35 PM, Pavan Balaji wrote: > >> We'll have our usual telecon tomorrow. The telecon details are available >> here: https://svn.mpi-forum.org/trac/mpi-forum-web/wiki/MPI3Hybrid >> >> (the telecon information has changed from the previous meetings). >> >> -- Pavan >> >> -- >> Pavan Balaji >> http://www.mcs.anl.gov/~balaji >> _______________________________________________ >> Mpi3-hybridpm mailing list >> Mpi3-hybridpm(a)lists.mpi-forum.org >> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-hybridpm > > Marc Snir > > > > > > _______________________________________________ > Mpi3-hybridpm mailing list > Mpi3-hybridpm(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-hybridpm -- Pavan Balaji http://www.mcs.anl.gov/~balaji

1 0

Re: [Mpi3-hybridpm] Reminder for the telecon tomorrow
by Snir, Marc 24 May '10

24 May '10

I shall be on a plane -- sorry. On May 24, 2010, at 4:35 PM, Pavan Balaji wrote: > > We'll have our usual telecon tomorrow. The telecon details are available > here: https://svn.mpi-forum.org/trac/mpi-forum-web/wiki/MPI3Hybrid > > (the telecon information has changed from the previous meetings). > > -- Pavan > > -- > Pavan Balaji > http://www.mcs.anl.gov/~balaji > _______________________________________________ > Mpi3-hybridpm mailing list > Mpi3-hybridpm(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-hybridpm Marc Snir

1 0

[Mpi3-hybridpm] Reminder for the telecon tomorrow
by Pavan Balaji 24 May '10

24 May '10

We'll have our usual telecon tomorrow. The telecon details are available here: https://svn.mpi-forum.org/trac/mpi-forum-web/wiki/MPI3Hybrid (the telecon information has changed from the previous meetings). -- Pavan -- Pavan Balaji http://www.mcs.anl.gov/~balaji

1 0

Re: [Mpi3-hybridpm] endpoint proposal
by Anthony Skjellum 22 May '10

22 May '10

MPI_Comm_merge is very useful esp. On intercommunicators, or as a form of inverse to split (K inputs, K bigger than 2) Tony Skjellum, PhD RunTime Computing Solutions, LLC tony(a)runtimecomputing.com direct: +1-205-314-3595 cell: +1-205-807-4968 -----Original Message----- From: Pavan Balaji <balaji(a)mcs.anl.gov> Date: Sat, 22 May 2010 11:57:08 To: <mpi3-hybridpm(a)lists.mpi-forum.org> Subject: Re: [Mpi3-hybridpm] endpoint proposal On 05/19/2010 06:55 PM, Snir, Marc wrote: > Page 20 -- do we want MPI_COMM_MERGE? Useful but takes a while to explain FWIW, I've talked to Torsten about pursuing this in the collectives working group, as that's where it really belongs. Using an MPI_COMM_MERGE function is *much* more cleaner than the extraction of endpoints from each process like was previously proposed, IMHO. The only major issue raised was that this is probably the only function that's collective over two communicators, instead of one. -- Pavan -- Pavan Balaji http://www.mcs.anl.gov/~balaji _______________________________________________ Mpi3-hybridpm mailing list Mpi3-hybridpm(a)lists.mpi-forum.org http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-hybridpm

1 0

Re: [Mpi3-hybridpm] endpoint proposal
by Pavan Balaji 22 May '10

22 May '10

On 05/19/2010 06:55 PM, Snir, Marc wrote: > Page 20 -- do we want MPI_COMM_MERGE? Useful but takes a while to explain FWIW, I've talked to Torsten about pursuing this in the collectives working group, as that's where it really belongs. Using an MPI_COMM_MERGE function is *much* more cleaner than the extraction of endpoints from each process like was previously proposed, IMHO. The only major issue raised was that this is probably the only function that's collective over two communicators, instead of one. -- Pavan -- Pavan Balaji http://www.mcs.anl.gov/~balaji

1 0

Re: [Mpi3-hybridpm] endpoint proposal [helper threads proposal updated]
by Douglas Miller 20 May '10

20 May '10

I have also attached an updated version of the Helper Threads proposal to Ticket #217 , to which there is a link from the MPI3 Hybrid Programming wiki. This includes some performance graphs for the BG/P Collective network showing its need for multiple threads. _______________________________________________ Douglas Miller BlueGene Messaging Development IBM Corp., Rochester, MN USA Bldg 030-2 A410 dougmill(a)us.ibm.com Douglas Miller/Rochester/IBM "Snir, Marc" <snir(a)illinois.ed u> To Sent by: "mpi3-hybridpm(a)lists.mpi-forum.org" mpi3-hybridpm-bou <mpi3-hybridpm(a)lists.mpi-forum.org> nces(a)lists.mpi-fo cc rum.org Subject [Mpi3-hybridpm] endpoint proposal 05/19/2010 06:55 PM Please respond to mpi3-hybridpm@lis ts.mpi-forum.org I attached a new version of the proposal on the wiki: main.2.pdf< https://svn.mpi-forum.org/trac/mpi-forum-web/attachment/wiki/MPI3Hybrid/mai… > Main changes: Changed the initialization functions as discussed in Chicago Added the COMM_MERGE functionality we discussed (page 20-23) Got rid of the new header files for OpenMP and PGAS Points to discuss: Page 10 The issue of threads being attached by default to endpoint 0 for backward compatibility. Need to discuss the current solution (an extra initialization argument) vs. the option of having all threads attached by default to endpoint 0. The issue of common level of thread support: common to all endpoints, to all endpoints at a process, different for each endpoint. Page 12 -- do we want cached attributes with endpoints? Page 20 -- do we want MPI_COMM_MERGE? Useful but takes a while to explain Page 35 & 37 - PGAS initialization (we never went over this text) _______________________________________________ Mpi3-hybridpm mailing list Mpi3-hybridpm(a)lists.mpi-forum.org http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-hybridpm

1 0

[Mpi3-hybridpm] endpoint proposal
by Snir, Marc 19 May '10

19 May '10

I attached a new version of the proposal on the wiki: main.2.pdf<https://svn.mpi-forum.org/trac/mpi-forum-web/attachment/wiki/MPI3Hybrid/mai…> Main changes: Changed the initialization functions as discussed in Chicago Added the COMM_MERGE functionality we discussed (page 20-23) Got rid of the new header files for OpenMP and PGAS Points to discuss: Page 10 The issue of threads being attached by default to endpoint 0 for backward compatibility. Need to discuss the current solution (an extra initialization argument) vs. the option of having all threads attached by default to endpoint 0. The issue of common level of thread support: common to all endpoints, to all endpoints at a process, different for each endpoint. Page 12 -- do we want cached attributes with endpoints? Page 20 -- do we want MPI_COMM_MERGE? Useful but takes a while to explain Page 35 & 37 - PGAS initialization (we never went over this text)

1 0

[Mpi3-hybridpm] No telecon today
by Pavan Balaji 11 May '10

11 May '10

Folks, I forgot to send out an email, but I assumed that since we just met last week, there'll not be any telecon today. Unless someone has something pressing to discuss, we'll be canceling today's telecon and will meet in two weeks. Also, for future telecons, please note that the telecon number has changed on the WG wiki. Thanks, -- Pavan -- Pavan Balaji http://www.mcs.anl.gov/~balaji

1 0