August 2010 - mpiwg-rma - mailman3.cels.anl.gov

Re: [Mpi3-rma] MPI3 RMA Proposal 1
by Rajeev Thakur 30 Aug '10

30 Aug '10

Thanks to you and Torsten for putting this together. As discussed in the last phone call, the plan is for everyone to review this proposal and send their comments either by email or during the next phone call this Friday at 1.00pm Central. This will be the last phone call before the next Forum meeting, and we would like to have at least Proposal 1 in good shape by then. Rajeev On Aug 30, 2010, at 12:10 PM, William Gropp wrote: > A draft of "RMA Proposal 1", along with some entries for the Journal of Development, has been posted on the wiki at > > https://svn.mpi-forum.org/trac/mpi-forum-web/wiki/mpi3-rma-proposal1 . > > There are still comments and points for discussion in this document; please send your feedback ASAP. > > As a reminder, Proposal 1 is intended as a relatively small change to the MPI-2 RMA chapter that address many (but not all) of the issues that have been raised by the user community. > > Bill > > William Gropp > Deputy Director for Research > Institute for Advanced Computing Applications and Technologies > Paul and Cynthia Saylor Professor of Computer Science > University of Illinois Urbana-Champaign > > > > > _______________________________________________ > mpi3-rma mailing list > mpi3-rma(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-rma

1 0

[Mpi3-rma] MPI3 RMA Proposal 1
by William Gropp 30 Aug '10

30 Aug '10

A draft of "RMA Proposal 1", along with some entries for the Journal of Development, has been posted on the wiki at https://svn.mpi-forum.org/trac/mpi-forum-web/wiki/mpi3-rma-proposal1 . There are still comments and points for discussion in this document; please send your feedback ASAP. As a reminder, Proposal 1 is intended as a relatively small change to the MPI-2 RMA chapter that address many (but not all) of the issues that have been raised by the user community. Bill William Gropp Deputy Director for Research Institute for Advanced Computing Applications and Technologies Paul and Cynthia Saylor Professor of Computer Science University of Illinois Urbana-Champaign

1 0

[Mpi3-rma] AUTO: Richard Treumann/Poughkeepsie/IBM is out of the office until 01/02/2001. (returning 09/07/2010)
by Richard Treumann 27 Aug '10

27 Aug '10

I am out of the office until 09/07/2010. I will be out of the office on vacation the week before Labor Day. I will not see any email. Note: This is an automated response to your message "[Mpi3-rma] next RMA telecon, Friday, Sept 3, 1pm Central" sent on 8/27/10 14:39:17. This is the only notification you will receive while this person is away.

1 0

[Mpi3-rma] next RMA telecon, Friday, Sept 3, 1pm Central
by Rajeev Thakur 27 Aug '10

27 Aug '10

The next RMA phone will be next Friday, Sept 3, at 1.00pm Central. Rajeev

1 0

[Mpi3-rma] next telecon Fri Aug 27, 1pm Central
by Rajeev Thakur 16 Aug '10

16 Aug '10

The next RMA phone call will be on Friday, August 27, 1.00pm Central. Same phone number, which is also on the wiki now. Rajeev

1 0

[Mpi3-rma] next RMA phone call
by Rajeev Thakur 06 Aug '10

06 Aug '10

The next RMA phone call will be on Monday, August 16, 2.00pm Central. Same phone number. Rajeev

1 0

[Mpi3-rma] RMA Telecon postponed to Friday, 3pm Central
by Rajeev Thakur 04 Aug '10

04 Aug '10

Today's RMA phone call has been postponed to Friday, Aug 6, 3.00pm Central. Rajeev On Jul 30, 2010, at 11:39 AM, Rajeev Thakur wrote: > As we discussed, the next meeting is on Wed, Aug 4, at 3.00pm Central. > Same phone number. > > Rajeev > >> -----Original Message----- >> From: mpi3-rma-bounces(a)lists.mpi-forum.org >> [mailto:[email protected]] On Behalf Of >> Rajeev Thakur >> Sent: Tuesday, July 27, 2010 12:38 PM >> To: 'MPI 3.0 Remote Memory Access working group' >> Subject: [Mpi3-rma] RMA telecon this Friday, 10.30am Central >> >> We are having an RMA working group telecon this Friday, July 30, at >> 10.30am Central time. >> >> # Domestic dial number: 866-654-6744 >> # International dial number: 517-308-8226 >> # Passcode: 6086006 >> >> Rajeev

1 0

Re: [Mpi3-rma] Strict vs. relaxed RMA in MPI
by Pavan Balaji 02 Aug '10

02 Aug '10

And I just looked at the time-stamp of your email, and it says May 28th. Maybe we are chasing an old email chain that got buffered and resent :-). -- Pavan On 08/02/2010 10:09 AM, Pavan Balaji wrote: > Brad, > > The working group is planning to provide both relaxed (as in MPI-2.2), > as well as some form of strictness (as in ordering from the same source > to the same memory location on the same target) for MPI-3. For other > forms of strictness, the user will need to take care of that by waiting > for the operations to finish before calling other operations. > > Based on our discussion at the SC PC meeting, I thought this would be > sufficient for Chapel. And based on my discussion with Jeff, I thought > this would be sufficient for GA as well. > > Were you looking for something additional? > > -- Pavan > > On 05/28/2010 06:44 PM, Brad Chamberlain wrote: >> >> Hi MPI-3 RMA team -- >> >> I ran into Jeff Hammond at a workshop a few weeks back and we had a brief >> chat about whether, as a potential client of MPI-3 RMA, I would prefer its >> semantics to err more on the strict or relaxed side. He requested that I >> consider sending a brief note to this group with my thoughts, so this is >> that note. I hope that this opinion will be considered useful and not >> out-of-turn given how little time I've had to invest in following the work >> of the MPI-3 team. >> >> I should start with the disclaimer that I'm not an expert on memory >> consistency models -- I probably know more than the average programmer, >> but have typically been insulated from worrying about it in a great amount >> of detail, either by relying on other software layers or languages to take >> care of it for me or by having the fortune to work with codes and idioms >> that don't fall afoul of the differences. >> >> My gut response to the question is that I'd prefer things to be on the >> more relaxed side. I think one of the key benefits of single-sided >> communication is its separation of data transfer from synchronization. >> I'd worry that by trying to enforce too much strictness in the RMA >> interface, it would work break down this separation and result in >> performance overheads that couldn't be recouped. >> >> On the other hand, if MPI-3 exported a model that was more relaxed than a >> particular programmer/programming model wanted, my assumption is that they >> could increase the strictness by doing more manual synchronization/memory >> fences/etc. themselves. That is, a relaxed model would not seem to >> exclude strictness while a strict model may impact performance negatively >> without any recourse. If that's a correct interpretation, the relaxed >> approach seems like the one to take to me. >> >> I'm reluctant to speak for others, but wanted to note (if he hasn't >> already done so) that Dave Grove from IBM's X10 team was with us and >> seemed to agree with this point-of-view (though perhaps we were both >> simply falling prey to Jeff's subliminal hypnosis? :). All that said, >> owing to my lack of depth in this area, I would say that if the GASNet >> team and/or the UPC/Titanium teams who built on top of GASNet felt that >> this was clearly the wrong approach, I would tend to cast my vote with >> them since I think they've studied this issue in far more detail than most >> parallel language groups, ours included. (I do think that Kathy Yelick >> voiced a compatible opinion in another context at this same workshop, >> which gave me some reassurance that relaxed was the way to go, but again, >> these were fairly high-level conversations. More generally, I would >> encourage you to get input from the GASNet team as you consider this issue >> and others related to 1-sided communication if you haven't). >> >> If you think it would be useful for me to hear the other side of the >> debate and/or consider some specific case examples in more detail, I'd be >> happy to do so as time permits. >> >> Have a good weekend, >> -Brad >> >> _______________________________________________ >> mpi3-rma mailing list >> mpi3-rma(a)lists.mpi-forum.org >> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-rma > -- Pavan Balaji http://www.mcs.anl.gov/~balaji

1 0

Re: [Mpi3-rma] Strict vs. relaxed RMA in MPI
by Pavan Balaji 02 Aug '10

02 Aug '10

Brad, The working group is planning to provide both relaxed (as in MPI-2.2), as well as some form of strictness (as in ordering from the same source to the same memory location on the same target) for MPI-3. For other forms of strictness, the user will need to take care of that by waiting for the operations to finish before calling other operations. Based on our discussion at the SC PC meeting, I thought this would be sufficient for Chapel. And based on my discussion with Jeff, I thought this would be sufficient for GA as well. Were you looking for something additional? -- Pavan On 05/28/2010 06:44 PM, Brad Chamberlain wrote: > > Hi MPI-3 RMA team -- > > I ran into Jeff Hammond at a workshop a few weeks back and we had a brief > chat about whether, as a potential client of MPI-3 RMA, I would prefer its > semantics to err more on the strict or relaxed side. He requested that I > consider sending a brief note to this group with my thoughts, so this is > that note. I hope that this opinion will be considered useful and not > out-of-turn given how little time I've had to invest in following the work > of the MPI-3 team. > > I should start with the disclaimer that I'm not an expert on memory > consistency models -- I probably know more than the average programmer, > but have typically been insulated from worrying about it in a great amount > of detail, either by relying on other software layers or languages to take > care of it for me or by having the fortune to work with codes and idioms > that don't fall afoul of the differences. > > My gut response to the question is that I'd prefer things to be on the > more relaxed side. I think one of the key benefits of single-sided > communication is its separation of data transfer from synchronization. > I'd worry that by trying to enforce too much strictness in the RMA > interface, it would work break down this separation and result in > performance overheads that couldn't be recouped. > > On the other hand, if MPI-3 exported a model that was more relaxed than a > particular programmer/programming model wanted, my assumption is that they > could increase the strictness by doing more manual synchronization/memory > fences/etc. themselves. That is, a relaxed model would not seem to > exclude strictness while a strict model may impact performance negatively > without any recourse. If that's a correct interpretation, the relaxed > approach seems like the one to take to me. > > I'm reluctant to speak for others, but wanted to note (if he hasn't > already done so) that Dave Grove from IBM's X10 team was with us and > seemed to agree with this point-of-view (though perhaps we were both > simply falling prey to Jeff's subliminal hypnosis? :). All that said, > owing to my lack of depth in this area, I would say that if the GASNet > team and/or the UPC/Titanium teams who built on top of GASNet felt that > this was clearly the wrong approach, I would tend to cast my vote with > them since I think they've studied this issue in far more detail than most > parallel language groups, ours included. (I do think that Kathy Yelick > voiced a compatible opinion in another context at this same workshop, > which gave me some reassurance that relaxed was the way to go, but again, > these were fairly high-level conversations. More generally, I would > encourage you to get input from the GASNet team as you consider this issue > and others related to 1-sided communication if you haven't). > > If you think it would be useful for me to hear the other side of the > debate and/or consider some specific case examples in more detail, I'd be > happy to do so as time permits. > > Have a good weekend, > -Brad > > _______________________________________________ > mpi3-rma mailing list > mpi3-rma(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-rma -- Pavan Balaji http://www.mcs.anl.gov/~balaji

1 0

Re: [Mpi3-rma] Strict vs. relaxed RMA in MPI
by Krishnan, Manojkumar 02 Aug '10

02 Aug '10

Brad, I would prefer things would be more on the relaxed side due to potential performance benefits, unless someone has an argument against it. For example, put followed by fence could be a round trip and therefore expensive, when compared to simple put where the cost is more like initiating the data transfer (and the source buffer safe to reuse). In ARMCI, the model is relaxed, however it could be made strict by enforcing fence after put/acc operations. -Manoj. --------------------------------------------------------------- Manojkumar Krishnan | High Performance Computing | (509) 372-4206 http://hpc.pnl.gov/people/manoj ________________________________________ From: mpi3-rma-bounces(a)lists.mpi-forum.org [mpi3-rma-bounces(a)lists.mpi-forum.org] On Behalf Of Brad Chamberlain [bradc(a)cray.com] Sent: Friday, May 28, 2010 4:44 PM To: mpi3-rma(a)lists.mpi-forum.org Subject: [Mpi3-rma] Strict vs. relaxed RMA in MPI Hi MPI-3 RMA team -- I ran into Jeff Hammond at a workshop a few weeks back and we had a brief chat about whether, as a potential client of MPI-3 RMA, I would prefer its semantics to err more on the strict or relaxed side. He requested that I consider sending a brief note to this group with my thoughts, so this is that note. I hope that this opinion will be considered useful and not out-of-turn given how little time I've had to invest in following the work of the MPI-3 team. I should start with the disclaimer that I'm not an expert on memory consistency models -- I probably know more than the average programmer, but have typically been insulated from worrying about it in a great amount of detail, either by relying on other software layers or languages to take care of it for me or by having the fortune to work with codes and idioms that don't fall afoul of the differences. My gut response to the question is that I'd prefer things to be on the more relaxed side. I think one of the key benefits of single-sided communication is its separation of data transfer from synchronization. I'd worry that by trying to enforce too much strictness in the RMA interface, it would work break down this separation and result in performance overheads that couldn't be recouped. On the other hand, if MPI-3 exported a model that was more relaxed than a particular programmer/programming model wanted, my assumption is that they could increase the strictness by doing more manual synchronization/memory fences/etc. themselves. That is, a relaxed model would not seem to exclude strictness while a strict model may impact performance negatively without any recourse. If that's a correct interpretation, the relaxed approach seems like the one to take to me. I'm reluctant to speak for others, but wanted to note (if he hasn't already done so) that Dave Grove from IBM's X10 team was with us and seemed to agree with this point-of-view (though perhaps we were both simply falling prey to Jeff's subliminal hypnosis? :). All that said, owing to my lack of depth in this area, I would say that if the GASNet team and/or the UPC/Titanium teams who built on top of GASNet felt that this was clearly the wrong approach, I would tend to cast my vote with them since I think they've studied this issue in far more detail than most parallel language groups, ours included. (I do think that Kathy Yelick voiced a compatible opinion in another context at this same workshop, which gave me some reassurance that relaxed was the way to go, but again, these were fairly high-level conversations. More generally, I would encourage you to get input from the GASNet team as you consider this issue and others related to 1-sided communication if you haven't). If you think it would be useful for me to hear the other side of the debate and/or consider some specific case examples in more detail, I'd be happy to do so as time permits. Have a good weekend, -Brad _______________________________________________ mpi3-rma mailing list mpi3-rma(a)lists.mpi-forum.org http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-rma

1 0