February 2012 - mpiwg-hybridpm - mailman3.cels.anl.gov

Re: [Mpi3-hybridpm] INIT/FINALIZE, with corrections suggested by Darius
by Jim Dinan 06 Feb '12

06 Feb '12

Thanks for the feedback. I created a ticket to track this: https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/320 ~Jim. On 2/5/12 7:09 PM, Martin Schulz wrote: > Hi Jim, > > Just as a comment: if you want to go down this route, it may be worth to think about a different approach to describing this list, since it is starting to get large. In addition to the three routines mentioned in 2.2, MPI_Get_library_version and all routines starting with MPI_T_ will be allowed as well in MPI 3 (based on the second votes for tickets #204 and #266 at the last forum). > > Martin > > > > > On Feb 3, 2012, at 3:17 PM, Jim Dinan wrote: > >> Right, that change wouldn't go in this ticket. My question was more >> along the lines of: Is this something worthwhile in the broader MPI >> sense? If so, I can create a ticket so we don't lose it. >> >> ~Jim. >> >> On 2/3/12 2:48 PM, Pavan Balaji wrote: >>> >>> On 02/03/2012 02:36 PM, Jim Dinan wrote: >>>> As long as we're updating this text, do we want to allow any additional >>>> routines before MPI_Init/after MPI_Finalize (e.g. MPI_Wtime)? >>> >>> Why? The changes made so far are all addressing the limitations that >>> need to be addressed for the endpoints proposal. They are not being made >>> lightly. Adding MPI_WTIME is irrelevant for this. >>> >>> That could be a separate ticket if you like. But there's no reason to >>> mix these two together. >>> >>> -- Pavan >>> >> _______________________________________________ >> Mpi3-hybridpm mailing list >> Mpi3-hybridpm(a)lists.mpi-forum.org >> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-hybridpm > > ________________________________________________________________________ > Martin Schulz, schulzm(a)llnl.gov, http://people.llnl.gov/schulzm > CASC @ Lawrence Livermore National Laboratory, Livermore, USA > > > > > _______________________________________________ > 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] INIT/FINALIZE, with corrections suggested by Darius
by Martin Schulz 05 Feb '12

05 Feb '12

Hi Jim, Just as a comment: if you want to go down this route, it may be worth to think about a different approach to describing this list, since it is starting to get large. In addition to the three routines mentioned in 2.2, MPI_Get_library_version and all routines starting with MPI_T_ will be allowed as well in MPI 3 (based on the second votes for tickets #204 and #266 at the last forum). Martin On Feb 3, 2012, at 3:17 PM, Jim Dinan wrote: > Right, that change wouldn't go in this ticket. My question was more > along the lines of: Is this something worthwhile in the broader MPI > sense? If so, I can create a ticket so we don't lose it. > > ~Jim. > > On 2/3/12 2:48 PM, Pavan Balaji wrote: >> >> On 02/03/2012 02:36 PM, Jim Dinan wrote: >>> As long as we're updating this text, do we want to allow any additional >>> routines before MPI_Init/after MPI_Finalize (e.g. MPI_Wtime)? >> >> Why? The changes made so far are all addressing the limitations that >> need to be addressed for the endpoints proposal. They are not being made >> lightly. Adding MPI_WTIME is irrelevant for this. >> >> That could be a separate ticket if you like. But there's no reason to >> mix these two together. >> >> -- Pavan >> > _______________________________________________ > Mpi3-hybridpm mailing list > Mpi3-hybridpm(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-hybridpm ________________________________________________________________________ Martin Schulz, schulzm(a)llnl.gov, http://people.llnl.gov/schulzm CASC @ Lawrence Livermore National Laboratory, Livermore, USA

1 0

Re: [Mpi3-hybridpm] INIT/FINALIZE, with corrections suggested by Darius
by Jim Dinan 03 Feb '12

03 Feb '12

Right, that change wouldn't go in this ticket. My question was more along the lines of: Is this something worthwhile in the broader MPI sense? If so, I can create a ticket so we don't lose it. ~Jim. On 2/3/12 2:48 PM, Pavan Balaji wrote: > > On 02/03/2012 02:36 PM, Jim Dinan wrote: >> As long as we're updating this text, do we want to allow any additional >> routines before MPI_Init/after MPI_Finalize (e.g. MPI_Wtime)? > > Why? The changes made so far are all addressing the limitations that > need to be addressed for the endpoints proposal. They are not being made > lightly. Adding MPI_WTIME is irrelevant for this. > > That could be a separate ticket if you like. But there's no reason to > mix these two together. > > -- Pavan >

1 0

Re: [Mpi3-hybridpm] INIT/FINALIZE, with corrections suggested by Darius
by Pavan Balaji 03 Feb '12

03 Feb '12

On 02/03/2012 02:36 PM, Jim Dinan wrote: > As long as we're updating this text, do we want to allow any additional > routines before MPI_Init/after MPI_Finalize (e.g. MPI_Wtime)? Why? The changes made so far are all addressing the limitations that need to be addressed for the endpoints proposal. They are not being made lightly. Adding MPI_WTIME is irrelevant for this. That could be a separate ticket if you like. But there's no reason to mix these two together. -- Pavan -- Pavan Balaji http://www.mcs.anl.gov/~balaji

1 0

Re: [Mpi3-hybridpm] INIT/FINALIZE, with corrections suggested by Darius
by Jim Dinan 03 Feb '12

03 Feb '12

As long as we're updating this text, do we want to allow any additional routines before MPI_Init/after MPI_Finalize (e.g. MPI_Wtime)? On 2/3/12 11:49 AM, Jim Dinan wrote: > p. 318, l. 17: "_must_ have been matched" > > On 2/3/12 10:46 AM, Marc Snir wrote: >> >> >> >> >> _______________________________________________ >> 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

1 0

Re: [Mpi3-hybridpm] Next Hybrid telecon
by Pavan Balaji 03 Feb '12

03 Feb '12

Folks, Here are the minutes for the meeting today: https://svn.mpi-forum.org/trac/mpi-forum-web/wiki/MPI3Hybrid/notes-2012-02-… Thanks, -- Pavan On 01/24/2012 04:21 PM, Pavan Balaji wrote: > Hi all, > > The next hybrid telecon will be from 11-12 central time on Friday 02/03. > Telecon details are on the wiki. > > https://svn.mpi-forum.org/trac/mpi-forum-web/wiki/MPI3Hybrid > > -- Pavan > -- Pavan Balaji http://www.mcs.anl.gov/~balaji

1 0

Re: [Mpi3-hybridpm] INIT/FINALIZE, with corrections suggested by Darius
by Jim Dinan 03 Feb '12

03 Feb '12

p. 318, l. 17: "_must_ have been matched" On 2/3/12 10:46 AM, Marc Snir wrote: > > > > > _______________________________________________ > 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] INIT/FINALIZE, with corrections suggested by Darius
by Marc Snir 03 Feb '12

03 Feb '12

1 0

Re: [Mpi3-hybridpm] Next Hybrid telecon
by Jim Dinan 03 Feb '12

03 Feb '12

Hi All, We've had some discussion on naming of the helper threads routines and have generated the following interfaces for discussion in today's telecon: Original Version I Version II ======================================================================= MPI_Team_join MPI_Team_register MPI_Team_register MPI_Team_leave MPI_Team_execute_deregister MPI_Team_execute(int coord) MPI_Team_sync MPI_Team_execute MPI_Team_break MPI_Team_deregister Where "coordination" argument can be one of: { MPI_TEAM_EXECUTE_ONCE, MPI_TEAM_EXECUTE_REREGISTER, MPI_TEAM_NO_EXECUTE } Version I adds symmetry between join/break and similarity between leave/sync. Version II merges leave, sync, and break into a single call. This version would also allow an implementation to define other coordination mechanisms that could be specific to a given system. For example "MPI_TEAM_EXECUTE_WHILE_IDLE" could give a thread back to the application if a message arrives. Cheers, ~Jim. On 1/24/12 4:21 PM, Pavan Balaji wrote: > Hi all, > > The next hybrid telecon will be from 11-12 central time on Friday 02/03. > Telecon details are on the wiki. > > https://svn.mpi-forum.org/trac/mpi-forum-web/wiki/MPI3Hybrid > > -- Pavan >

1 0

Re: [Mpi3-hybridpm] new text for MPI_INIT/MPI_FINALIZE
by Pavan Balaji 03 Feb '12

03 Feb '12

On 02/02/2012 09:57 AM, Douglas Miller wrote: > I still disagree with the idea that MPI_INIT/FINALIZE is called > per-MPI-Process. I know that this ticket does not deal with multiple MPI > processes, but unless this sort of wording is stated correctly, ticket > #310 won't work. We should really discuss this in the telecon, rather than on mailing lists, but each MPI process calling INIT/FINALIZE should still work with ticket #310. Yes, the implementation will have to do some extra work, but maybe you can clarify why you think this is not implementable? > The big brick wall that ticket #310 will run into if we promote the idea > that "every MPI process has its own MPI_INIT/FINALIZE" is that if there > are multiple MPI processes then there is no clear way to initialize them > since there is no argument to MPI_INIT/FINALIZE that specifies the MPI > Process and there is no way to know what MPI process a thread might > represent. My understanding is that a thread does not represent any MPI process before it calls MPI_INIT. After it calls MPI_INIT, it represents an MPI process -- at this point, my expectation is that the MPI library will add this information to the thread private storage. -- Pavan -- Pavan Balaji http://www.mcs.anl.gov/~balaji

1 0