July 2012 - mpiwg-rma - mailman3.cels.anl.gov

Re: [Mpi3-rma] notes from 5pm Tuesday meeting @ July 2012 Forum
by Jeff Hammond 17 Jul '12

17 Jul '12

Are we going to force implementers to support 8-byte atomics when the load-store granularity is 4-bytes? I guess it doesn't matter so much but I thought it prudent to ask. I guess this was addressed already. Jeff On Tue, Jul 17, 2012 at 6:26 PM, Jeff Hammond <jhammond(a)alcf.anl.gov> wrote: > I don't know that floating-point compare is well-defined. You really > have ask "if abs(x-y)<tolerance" and not "if x==y". > > I think only fixed-point types should be valid for CAS. > > Jeff > > On Tue, Jul 17, 2012 at 6:14 PM, Dave Goodell <goodell(a)mcs.anl.gov> wrote: >> Discussed the complex types in COMPARE_AND_SWAP issue. "Fortran Integer" category is permitted, but "Complex" category is not, primarily because of width. Since "Fortran Integer" contains wide types, shouldn't we just permit "Complex" and "Floating point" as well? Consensus was to stick with the existing text which permits only "C integer, Fortran integer, Logical, Multi-language types, or Byte". >> >> Group review (esp. Jim & Sreeram): https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/187 >> >> incorporate Jim suggested change (Torsten): https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/271 >> >> we think we are unaffected, but need a second check (Jim): https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/278 >> >> Double-check that C++ is not referenced in the RMA chapter (Pavan): https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/281 >> >> Needs review (Dave): https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/300 >> >> Think unaffected, but slim chance of Rput/Rget being affected (Pavan): https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/317 >> >> Need to check implementation of various (4?) "flush is non-local" changes (Dave): https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/270 >> >> Need to check disp_unit change (Jim & Sreeram): https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/284 >> >> After the above items have all been dealt with, all chapter committee members should re-read the whole chapter in the *clean* document (the one _without_ the changebars and colored text) to look for obvious typos and inconsistencies. >> >> -Dave >> >> >> _______________________________________________ >> mpi3-rma mailing list >> mpi3-rma(a)lists.mpi-forum.org >> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-rma > > > > -- > Jeff Hammond > Argonne Leadership Computing Facility > University of Chicago Computation Institute > jhammond(a)alcf.anl.gov / (630) 252-5381 > http://www.linkedin.com/in/jeffhammond > https://wiki.alcf.anl.gov/parts/index.php/User:Jhammond -- Jeff Hammond Argonne Leadership Computing Facility University of Chicago Computation Institute jhammond(a)alcf.anl.gov / (630) 252-5381 http://www.linkedin.com/in/jeffhammond https://wiki.alcf.anl.gov/parts/index.php/User:Jhammond

1 0

Re: [Mpi3-rma] notes from 5pm Tuesday meeting @ July 2012 Forum
by Jeff Hammond 17 Jul '12

17 Jul '12

I don't know that floating-point compare is well-defined. You really have ask "if abs(x-y)<tolerance" and not "if x==y". I think only fixed-point types should be valid for CAS. Jeff On Tue, Jul 17, 2012 at 6:14 PM, Dave Goodell <goodell(a)mcs.anl.gov> wrote: > Discussed the complex types in COMPARE_AND_SWAP issue. "Fortran Integer" category is permitted, but "Complex" category is not, primarily because of width. Since "Fortran Integer" contains wide types, shouldn't we just permit "Complex" and "Floating point" as well? Consensus was to stick with the existing text which permits only "C integer, Fortran integer, Logical, Multi-language types, or Byte". > > Group review (esp. Jim & Sreeram): https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/187 > > incorporate Jim suggested change (Torsten): https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/271 > > we think we are unaffected, but need a second check (Jim): https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/278 > > Double-check that C++ is not referenced in the RMA chapter (Pavan): https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/281 > > Needs review (Dave): https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/300 > > Think unaffected, but slim chance of Rput/Rget being affected (Pavan): https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/317 > > Need to check implementation of various (4?) "flush is non-local" changes (Dave): https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/270 > > Need to check disp_unit change (Jim & Sreeram): https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/284 > > After the above items have all been dealt with, all chapter committee members should re-read the whole chapter in the *clean* document (the one _without_ the changebars and colored text) to look for obvious typos and inconsistencies. > > -Dave > > > _______________________________________________ > mpi3-rma mailing list > mpi3-rma(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-rma -- Jeff Hammond Argonne Leadership Computing Facility University of Chicago Computation Institute jhammond(a)alcf.anl.gov / (630) 252-5381 http://www.linkedin.com/in/jeffhammond https://wiki.alcf.anl.gov/parts/index.php/User:Jhammond

1 0

Re: [Mpi3-rma] notes from 5pm Tuesday meeting @ July 2012 Forum
by Torsten Hoefler 17 Jul '12

17 Jul '12

Hi all, > incorporate Jim suggested change (Torsten): https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/271 Done! --> http://www.unixer.de/sec/mpi-report.pdf Torsten -- ### qreharg rug ebs fv crryF ------------- http://www.unixer.de/ ----- Torsten Hoefler | Performance Modeling and Simulation Lead Blue Waters Directorate | University of Illinois (UIUC) 1205 W Clark Street | Urbana, IL, 61801 NCSA Building | +01 (217) 244-7736

1 0

Re: [Mpi3-rma] RMA Reviews for yesterday's passed tickets and WG meeting
by Torsten Hoefler 17 Jul '12

17 Jul '12

Howard, Thanks! Change committed! Torsten On Tue, Jul 17, 2012 at 11:06:46PM +0000, Howard Pritchard wrote: > Torsten, > > I noticed the MPI_Address_kind problem (should be MPI_ADDRESS_KIND) > On page 97, line 1, page 413 line 21 and page 415 line 16 of the > Rma chapter in this version. > > Howard > > > -----Original Message----- > From: mpi3-rma-bounces(a)lists.mpi-forum.org [mailto:[email protected]] On Behalf Of Torsten Hoefler > Sent: Tuesday, July 17, 2012 3:11 PM > To: mpi3-rma(a)lists.mpi-forum.org > Subject: Re: [Mpi3-rma] RMA Reviews for yesterday's passed tickets and WG meeting > > Hi again, > > I also rolled in a bunch of smaller changes from Jim's marked up > document. Those all count as ticket 0 changes and don't change > semantics, thus, I didn't use the change macros (they're already > three-levels deep in some places). > > Here's the diff: http://www.unixer.de/sec/jims_edits.diff > Here's the marked up doc: http://www.unixer.de/sec/mpi-report.pdf > Here's the clean copy: http://www.unixer.de/sec/mpi30-report.pdf > > Committed as r1365! > > Please review carefully. Also I think we should stop correcting very > tiny issues since there is the danger to introduce bigger problems. > > Thanks, > Torsten > > On Tue, Jul 17, 2012 at 02:53:59PM -0500, Torsten Hoefler wrote: > > Hi all, > > > > I implemented all tickets that were passed yesterday. Please review the > > implementations (r1343) http://www.unixer.de/sec/mpi-report.pdf > > > > I would like to have an RMA meeting around 5pm today. > > > > Thanks & Best, > > Torsten > > > > -- > > ### qreharg rug ebs fv crryF ------------- http://www.unixer.de/ ----- > > Torsten Hoefler | Performance Modeling and Simulation Lead > > Blue Waters Directorate | University of Illinois (UIUC) > > 1205 W Clark Street | Urbana, IL, 61801 > > NCSA Building | +01 (217) 244-7736 > > _______________________________________________ > > mpi3-rma mailing list > > mpi3-rma(a)lists.mpi-forum.org > > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-rma > > > > -- > ### qreharg rug ebs fv crryF ------------- http://www.unixer.de/ ----- > Torsten Hoefler | Performance Modeling and Simulation Lead > Blue Waters Directorate | University of Illinois (UIUC) > 1205 W Clark Street | Urbana, IL, 61801 > NCSA Building | +01 (217) 244-7736 > _______________________________________________ > mpi3-rma mailing list > mpi3-rma(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-rma > > _______________________________________________ > mpi3-rma mailing list > mpi3-rma(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-rma > -- ### qreharg rug ebs fv crryF ------------- http://www.unixer.de/ ----- Torsten Hoefler | Performance Modeling and Simulation Lead Blue Waters Directorate | University of Illinois (UIUC) 1205 W Clark Street | Urbana, IL, 61801 NCSA Building | +01 (217) 244-7736

1 0

[Mpi3-rma] notes from 5pm Tuesday meeting @ July 2012 Forum
by Dave Goodell 17 Jul '12

17 Jul '12

Discussed the complex types in COMPARE_AND_SWAP issue. "Fortran Integer" category is permitted, but "Complex" category is not, primarily because of width. Since "Fortran Integer" contains wide types, shouldn't we just permit "Complex" and "Floating point" as well? Consensus was to stick with the existing text which permits only "C integer, Fortran integer, Logical, Multi-language types, or Byte". Group review (esp. Jim & Sreeram): https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/187 incorporate Jim suggested change (Torsten): https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/271 we think we are unaffected, but need a second check (Jim): https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/278 Double-check that C++ is not referenced in the RMA chapter (Pavan): https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/281 Needs review (Dave): https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/300 Think unaffected, but slim chance of Rput/Rget being affected (Pavan): https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/317 Need to check implementation of various (4?) "flush is non-local" changes (Dave): https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/270 Need to check disp_unit change (Jim & Sreeram): https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/284 After the above items have all been dealt with, all chapter committee members should re-read the whole chapter in the *clean* document (the one _without_ the changebars and colored text) to look for obvious typos and inconsistencies. -Dave

1 0

Re: [Mpi3-rma] RMA Reviews for yesterday's passed tickets and WG meeting
by Howard Pritchard 17 Jul '12

17 Jul '12

Torsten, I noticed the MPI_Address_kind problem (should be MPI_ADDRESS_KIND) On page 97, line 1, page 413 line 21 and page 415 line 16 of the Rma chapter in this version. Howard -----Original Message----- From: mpi3-rma-bounces(a)lists.mpi-forum.org [mailto:[email protected]] On Behalf Of Torsten Hoefler Sent: Tuesday, July 17, 2012 3:11 PM To: mpi3-rma(a)lists.mpi-forum.org Subject: Re: [Mpi3-rma] RMA Reviews for yesterday's passed tickets and WG meeting Hi again, I also rolled in a bunch of smaller changes from Jim's marked up document. Those all count as ticket 0 changes and don't change semantics, thus, I didn't use the change macros (they're already three-levels deep in some places). Here's the diff: http://www.unixer.de/sec/jims_edits.diff Here's the marked up doc: http://www.unixer.de/sec/mpi-report.pdf Here's the clean copy: http://www.unixer.de/sec/mpi30-report.pdf Committed as r1365! Please review carefully. Also I think we should stop correcting very tiny issues since there is the danger to introduce bigger problems. Thanks, Torsten On Tue, Jul 17, 2012 at 02:53:59PM -0500, Torsten Hoefler wrote: > Hi all, > > I implemented all tickets that were passed yesterday. Please review the > implementations (r1343) http://www.unixer.de/sec/mpi-report.pdf > > I would like to have an RMA meeting around 5pm today. > > Thanks & Best, > Torsten > > -- > ### qreharg rug ebs fv crryF ------------- http://www.unixer.de/ ----- > Torsten Hoefler | Performance Modeling and Simulation Lead > Blue Waters Directorate | University of Illinois (UIUC) > 1205 W Clark Street | Urbana, IL, 61801 > NCSA Building | +01 (217) 244-7736 > _______________________________________________ > mpi3-rma mailing list > mpi3-rma(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-rma > -- ### qreharg rug ebs fv crryF ------------- http://www.unixer.de/ ----- Torsten Hoefler | Performance Modeling and Simulation Lead Blue Waters Directorate | University of Illinois (UIUC) 1205 W Clark Street | Urbana, IL, 61801 NCSA Building | +01 (217) 244-7736 _______________________________________________ mpi3-rma mailing list mpi3-rma(a)lists.mpi-forum.org http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-rma

1 0

Re: [Mpi3-rma] RMA Reviews for yesterday's passed tickets and WG meeting
by Torsten Hoefler 17 Jul '12

17 Jul '12

Hi again, I also rolled in a bunch of smaller changes from Jim's marked up document. Those all count as ticket 0 changes and don't change semantics, thus, I didn't use the change macros (they're already three-levels deep in some places). Here's the diff: http://www.unixer.de/sec/jims_edits.diff Here's the marked up doc: http://www.unixer.de/sec/mpi-report.pdf Here's the clean copy: http://www.unixer.de/sec/mpi30-report.pdf Committed as r1365! Please review carefully. Also I think we should stop correcting very tiny issues since there is the danger to introduce bigger problems. Thanks, Torsten On Tue, Jul 17, 2012 at 02:53:59PM -0500, Torsten Hoefler wrote: > Hi all, > > I implemented all tickets that were passed yesterday. Please review the > implementations (r1343) http://www.unixer.de/sec/mpi-report.pdf > > I would like to have an RMA meeting around 5pm today. > > Thanks & Best, > Torsten > > -- > ### qreharg rug ebs fv crryF ------------- http://www.unixer.de/ ----- > Torsten Hoefler | Performance Modeling and Simulation Lead > Blue Waters Directorate | University of Illinois (UIUC) > 1205 W Clark Street | Urbana, IL, 61801 > NCSA Building | +01 (217) 244-7736 > _______________________________________________ > mpi3-rma mailing list > mpi3-rma(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-rma > -- ### qreharg rug ebs fv crryF ------------- http://www.unixer.de/ ----- Torsten Hoefler | Performance Modeling and Simulation Lead Blue Waters Directorate | University of Illinois (UIUC) 1205 W Clark Street | Urbana, IL, 61801 NCSA Building | +01 (217) 244-7736

1 0

[Mpi3-rma] RMA Reviews for yesterday's passed tickets and WG meeting
by Torsten Hoefler 17 Jul '12

17 Jul '12

Hi all, I implemented all tickets that were passed yesterday. Please review the implementations (r1343) http://www.unixer.de/sec/mpi-report.pdf I would like to have an RMA meeting around 5pm today. Thanks & Best, Torsten -- ### qreharg rug ebs fv crryF ------------- http://www.unixer.de/ ----- Torsten Hoefler | Performance Modeling and Simulation Lead Blue Waters Directorate | University of Illinois (UIUC) 1205 W Clark Street | Urbana, IL, 61801 NCSA Building | +01 (217) 244-7736

1 0

Re: [Mpi3-rma] Chapter Reviews for One Sided Chapter
by Jim Dinan 15 Jul '12

15 Jul '12

Hi All, Sorry for taking so long to send my review. I've attached a marked-up pdf with lots of edits. Most of these should be ticket 0, but I feel that all of them are important enough to resolve somehow for MPI 3.0. You will need real (free) adobe acrobat to view the edits (evince, for example, will not work at all). In acrobat, select view->navigation panels->comments to show a list of all edits and notes. Don't just scan the document to find them or you will miss some of the small edits. I also attached notes to explain some edits and another pdf viewer may not be able to show them to you. If pdf markup isn't going to work, let me know and I'll type up these edits up as text. One set of changes that may need some discussion are the flush operations edits. These operations are currently described as local, which contradicts the progress semantics for synchronization operations in 11.7.3 (highlighted in the pdf). If you have to wait for another process to give up the lock, a flush can become nonlocal as described in this section. I think the best option is to classify flushes as synchronization operations and remove the offending sentences from the flush text since it's already covered in 11.7.3. Cheers, ~Jim. On 7/12/12 1:50 PM, Torsten Hoefler wrote: > Review reminder! The meeting is next Monday! > > Thanks & Best, > Torsten > > On Thu, Jul 05, 2012 at 12:40:51PM -0500, Torsten Hoefler wrote: >> Ticket #283 is now also committed (r1244). Please review it or include >> it in your chapter committee review! >> >> I put a copy of the document for your convenience at >> http://www.unixer.de/sec/mpi-report-r1244.pdf . >> >> Thanks, >> Torsten >> >> On Wed, Jul 04, 2012 at 05:29:14PM -0500, Torsten Hoefler wrote: >>> [This message got stuck in mailman because the chapter committee is too >>> large (max recipients exceeded :-) -- here is a copy for the working >>> group -- Thanks, Torsten] >>> >>> Dear One Sided Chapter Committee and Working Group, >>> >>> Nearly all commits to the One Sided chapter for MPI-3.0 should be done >>> by now (most of them since quite a while). Ticket #283 is unclear, so I >>> didn't commit anything yet. Jim, can you elaborate? >>> >>> Please review the correct implementation of the following tickets *and* >>> sign off *at each ticket* that you reviewed it before the next Forum (in >>> two weeks): >>> >>> #140, #270, #284, #298, #308, #309 >>> >>> Everybody is welcome to review, there can never be enough eyes checking >>> the implementation. >>> >>> You can either pull a copy from subversion or use a build of r1243 that >>> I put at: http://www.unixer.de/sec/mpi-report-r1243.pdf . >>> >>> Please let me know if you have any questions! >>> >>> Thanks, >>> Torsten >>> >>> -- >>> ### qreharg rug ebs fv crryF ------------- http://www.unixer.de/ ----- >>> Torsten Hoefler | Performance Modeling and Simulation Lead >>> Blue Waters Directorate | University of Illinois (UIUC) >>> 1205 W Clark Street | Urbana, IL, 61801 >>> NCSA Building | +01 (217) 244-7736 >>> _______________________________________________ >>> mpi3-rma mailing list >>> mpi3-rma(a)lists.mpi-forum.org >>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-rma >>> >> >> -- >> ### qreharg rug ebs fv crryF ------------- http://www.unixer.de/ ----- >> Torsten Hoefler | Performance Modeling and Simulation Lead >> Blue Waters Directorate | University of Illinois (UIUC) >> 1205 W Clark Street | Urbana, IL, 61801 >> NCSA Building | +01 (217) 244-7736 >> _______________________________________________ >> mpi3-rma mailing list >> mpi3-rma(a)lists.mpi-forum.org >> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-rma >> >

1 0

Re: [Mpi3-rma] [EXTERNAL] RMA chapter discussions at next forum
by Torsten Hoefler 12 Jul '12

12 Jul '12

On Thu, Jul 12, 2012 at 07:07:01PM +0000, Barrett, Brian W wrote: > I unfortunately have a 9:00 flight out of ORD, so I won't be able to stay > very long if we meet. Ok > Do we really want to focus on things that are post MPI-3 at this > meeting (keeping in mind, we can't add info arguments to MPI-3 at this > point)? Sure, we can postpone this discussion (I just hoped that we get something technical done at this meeting :-)). However, we need to talk about #284 in any case. Torsten -- ### qreharg rug ebs fv crryF ------------- http://www.unixer.de/ ----- Torsten Hoefler | Performance Modeling and Simulation Lead Blue Waters Directorate | University of Illinois (UIUC) 1205 W Clark Street | Urbana, IL, 61801 NCSA Building | +01 (217) 244-7736

1 0