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

Re: [Mpi3-hybridpm] External interfaces chapter updates
by Pavan Balaji 30 Oct '10

30 Oct '10

On 10/30/2010 01:49 PM, Bronis R. de Supinski wrote: >> I think it's time for us to start dumping in text into the chapter and >> start discussing the exact wording. I've included the helper threads and >> shared memory extensions proposals into the chapter and uploaded it to >> the wiki >> (https://svn.mpi-forum.org/trac/mpi-forum-web/attachment/wiki/MPI3Hybrid/ei-…) >> Please take a look at it and let me know your comments. > > I'll try to make a detailed reading next week. In the I had initially incorrectly uploaded a change with respect to the MPI_THREAD_SERIALIZED semantics which we decided to drop last time. I've now uploaded v0.2 of the document with this correction as well as the other changes suggested by Bronis. -- Pavan -- Pavan Balaji http://www.mcs.anl.gov/~balaji

1 0

Re: [Mpi3-hybridpm] External interfaces chapter updates
by Bronis R. de Supinski 30 Oct '10

30 Oct '10

Bill: Doesn't it seem like it would be useful to partition the changes into the issues that motivate them? It would seem we could use the ticket number as a way to do that. Otherwise, it will be hard to tell minor wording changes that can be considered individually from those that need acceptance of a more significant overall change. Bronis On Sat, 30 Oct 2010, William Gropp wrote: > We don't need tickets for edits to the document - tickets were the > appropriate way to manage minor corrections to the document but don't > help the process of developing a new or greatly revised chapter. The > RMA group has two proposals on its wiki, both of which are in the > context of complete chapters. A drawback to tickets was they made it > possible to think that by reading the ticket, the issue was > understood. This caused several problems in the prior rounds, and > would be more serious here. > > So, no ticket is needed. Following the written instructions, a zero > (0) may be used. > > Bill > > On Oct 30, 2010, at 3:32 PM, Bronis R. de Supinski wrote: > >> >> Pavan: >> >> Re: >>>> 2. You need to annotate your proposed changes with ticket >>>> numbers. I suggest that we need separate tickets for each >>>> of the major additions. Isee that ticket 217: >>>> >>>> https://BLOCKEDsvn.mpi-forum.org/trac/mpi-forum-web/ticket/217 >>>> >>>> covers the helper threads proposal. We will need ticekts >>>> for the shared memory extensions and endpoints. >>> >>> I thought the process was changing out of using tickets for the >>> changes >>> for MPI-3. If not we can update this once all the tickets have been >>> created. >> >> The change tracking macros include a ticket number field. >> My understanding is that for simple wording changes for >> things like grammatical issues on which the chapter >> reaches consensus then we can use "ticket 0". For more >> significant changes, like major additions, my understanding >> is that they are still supposed to use a separate ticket. >> However, the revised text should be in LaTex so that >> we vote on it as part of the ticket. >> >> Bill and/or Rich should confirm or correct. >> >> Bronis >> > > 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

Re: [Mpi3-hybridpm] External interfaces chapter updates
by William Gropp 30 Oct '10

30 Oct '10

We don't need tickets for edits to the document - tickets were the appropriate way to manage minor corrections to the document but don't help the process of developing a new or greatly revised chapter. The RMA group has two proposals on its wiki, both of which are in the context of complete chapters. A drawback to tickets was they made it possible to think that by reading the ticket, the issue was understood. This caused several problems in the prior rounds, and would be more serious here. So, no ticket is needed. Following the written instructions, a zero (0) may be used. Bill On Oct 30, 2010, at 3:32 PM, Bronis R. de Supinski wrote: > > Pavan: > > Re: >>> 2. You need to annotate your proposed changes with ticket >>> numbers. I suggest that we need separate tickets for each >>> of the major additions. Isee that ticket 217: >>> >>> https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/217 >>> >>> covers the helper threads proposal. We will need ticekts >>> for the shared memory extensions and endpoints. >> >> I thought the process was changing out of using tickets for the >> changes >> for MPI-3. If not we can update this once all the tickets have been >> created. > > The change tracking macros include a ticket number field. > My understanding is that for simple wording changes for > things like grammatical issues on which the chapter > reaches consensus then we can use "ticket 0". For more > significant changes, like major additions, my understanding > is that they are still supposed to use a separate ticket. > However, the revised text should be in LaTex so that > we vote on it as part of the ticket. > > Bill and/or Rich should confirm or correct. > > Bronis > 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

Re: [Mpi3-hybridpm] External interfaces chapter updates
by Bronis R. de Supinski 30 Oct '10

30 Oct '10

Pavan: Re: >> 2. You need to annotate your proposed changes with ticket >> numbers. I suggest that we need separate tickets for each >> of the major additions. Isee that ticket 217: >> >> https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/217 >> >> covers the helper threads proposal. We will need ticekts >> for the shared memory extensions and endpoints. > > I thought the process was changing out of using tickets for the changes > for MPI-3. If not we can update this once all the tickets have been created. The change tracking macros include a ticket number field. My understanding is that for simple wording changes for things like grammatical issues on which the chapter reaches consensus then we can use "ticket 0". For more significant changes, like major additions, my understanding is that they are still supposed to use a separate ticket. However, the revised text should be in LaTex so that we vote on it as part of the ticket. Bill and/or Rich should confirm or correct. Bronis

1 0

Re: [Mpi3-hybridpm] External interfaces chapter updates
by Pavan Balaji 30 Oct '10

30 Oct '10

Bronis, On 10/30/2010 01:49 PM, Bronis R. de Supinski wrote: > 1. You need to modify section 12.1 in light of the additions. > In the least, you should describe the proposed section 12.5 > there although it would probably be good to mention helper > threads in some way also. Fixed. > 2. You need to annotate your proposed changes with ticket > numbers. I suggest that we need separate tickets for each > of the major additions. Isee that ticket 217: > > https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/217 > > covers the helper threads proposal. We will need ticekts > for the shared memory extensions and endpoints. I thought the process was changing out of using tickets for the changes for MPI-3. If not we can update this once all the tickets have been created. Thanks, -- Pavan -- Pavan Balaji http://www.mcs.anl.gov/~balaji

1 0

Re: [Mpi3-hybridpm] External interfaces chapter updates
by Bronis R. de Supinski 30 Oct '10

30 Oct '10

Pavan: Re: > I think it's time for us to start dumping in text into the chapter and > start discussing the exact wording. I've included the helper threads and > shared memory extensions proposals into the chapter and uploaded it to > the wiki > (https://svn.mpi-forum.org/trac/mpi-forum-web/attachment/wiki/MPI3Hybrid/ei-…) > Please take a look at it and let me know your comments. I'll try to make a detailed reading next week. In the meantime, I have some immediate comments: 1. You need to modify section 12.1 in light of the additions. In the least, you should describe the proposed section 12.5 there although it would probably be good to mention helper threads in some way also. 2. You need to annotate your proposed changes with ticket numbers. I suggest that we need separate tickets for each of the major additions. Isee that ticket 217: https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/217 covers the helper threads proposal. We will need ticekts for the shared memory extensions and endpoints. Bronis > Marc is still making some changes to the endpoints proposal. I'll add it > in once the changes are done. > > -- Pavan > > -- > Pavan Balaji > http://BLOCKEDwww.BLOCKEDmcs.anl.gov/~balaji > _______________________________________________ > Mpi3-hybridpm mailing list > Mpi3-hybridpm(a)lists.mpi-forum.org > http://BLOCKEDlists.mpi-forum.org/mailman/listinfo.cgi/mpi3-hybridpm > >

1 0

[Mpi3-hybridpm] External interfaces chapter updates
by Pavan Balaji 29 Oct '10

29 Oct '10

Folks, I think it's time for us to start dumping in text into the chapter and start discussing the exact wording. I've included the helper threads and shared memory extensions proposals into the chapter and uploaded it to the wiki (https://svn.mpi-forum.org/trac/mpi-forum-web/attachment/wiki/MPI3Hybrid/ei-…) Please take a look at it and let me know your comments. Marc is still making some changes to the endpoints proposal. I'll add it in once the changes are done. -- Pavan -- Pavan Balaji http://www.mcs.anl.gov/~balaji

1 0