November 2010 - mpiwg-tools - mailman3.cels.anl.gov

Re: [Mpi3-tools] Next WG meeting on 11/29 cancelled
by Martin Schulz 25 Nov '10

25 Nov '10

Hi Andreas, The two major changes that I can think about now are changes to the taxonomy section, which Dave Goodell is currently preparing, and changes to the search and discovery functions to enable MPIT to safely remove variables or variable association. Bast on the last discussion, the latter will probably introduce a handles concept. Besides that there will be many small changes to address the various comments. The next few days will be Thanksgiving holidays in the US, but my plan is to work on this early next week. Martin PS: Regarding your other email and the error codes: sure, that makes sense and I will add this. Thanks for the feedback. On Nov 25, 2010, at 3:05 AM, Andreas Kuntze wrote: > Hi Martin, > > On Thursday 25 November 2010 08:58:32 Martin Schulz wrote: > >> Due to SC and the holidays I will not have a new draft of MPIT >> ready by next Monday to discuss. Since there is, I believe, no >> other current topic, I am proposing to cancel the next meeting >> and get together during the next MPI forum meeting in December >> (then with a new MPIT draft). > > can you tell me which big changes are expected to be included in the next > draft? I wonder if it is reasonable to wait for the next draft or to keep my > current implementation for my diploma thesis. > > Thanks, > > Andreas > > > _______________________________________________ > Mpi3-tools mailing list > Mpi3-tools(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-tools ________________________________________________________________________ Martin Schulz, schulzm(a)llnl.gov, http://people.llnl.gov/schulzm CASC @ Lawrence Livermore National Laboratory, Livermore, USA

1 0

Re: [Mpi3-tools] Next WG meeting on 11/29 cancelled
by Andreas Kuntze 25 Nov '10

25 Nov '10

Hi Martin, On Thursday 25 November 2010 08:58:32 Martin Schulz wrote: > Due to SC and the holidays I will not have a new draft of MPIT > ready by next Monday to discuss. Since there is, I believe, no > other current topic, I am proposing to cancel the next meeting > and get together during the next MPI forum meeting in December > (then with a new MPIT draft). can you tell me which big changes are expected to be included in the next draft? I wonder if it is reasonable to wait for the next draft or to keep my current implementation for my diploma thesis. Thanks, Andreas

1 0

[Mpi3-tools] MPIT_CTRLVAR_WRITE return codes
by Andreas Kuntze 25 Nov '10

25 Nov '10

Hi all, in the current draft, MPIT_CTRLVAR_WRITE can return the error codes PIT_ERR_SETNOTNOW and PIT_ERR_SETNEVER only. For my implementation it would be useful, if you could return something indicating that the value you want to write is out of range or that not all elements could be changed (if a control variable contains more than one element). Andreas

1 0

[Mpi3-tools] Next WG meeting on 11/29 cancelled
by Martin Schulz 25 Nov '10

25 Nov '10

Hi all, Due to SC and the holidays I will not have a new draft of MPIT ready by next Monday to discuss. Since there is, I believe, no other current topic, I am proposing to cancel the next meeting and get together during the next MPI forum meeting in December (then with a new MPIT draft). If someone has a topic for the next meeting, though, please let me and Jeff know and we can schedule another Webex session. Happy Thanksgiving! Martin ________________________________________________________________________ Martin Schulz, schulzm(a)llnl.gov, http://people.llnl.gov/schulzm CASC @ Lawrence Livermore National Laboratory, Livermore, USA

1 0

Re: [Mpi3-tools] Final MPI-3 Tools BOF Slides
by Jeff Squyres 07 Nov '10

07 Nov '10

Points - - Slide 3: "Main philosophy" seems (mostly) redundant with prior slide - Slide 3: "Information provided as a set of variables" seems redundant with prior slide - Too many words on slide 4 Slightly modified the title slide I deleted slide 3 and modified/extended slide 2. I changed the MPIR slide to add visuals, reduced number of words On Nov 6, 2010, at 8:42 PM, Martin Schulz wrote: > Hi all, > > Thanks to all for the comments and the feedback. I tried to > include all comments (more detailed responses summarized > below). Attached are the final slides that I will send to Rich. > > Martin > <2010-11-mpi3-toolswg-status-sc.pptx> > > Marc-Andre: > >> I think the slides are too full. Maybe I am missing what the BoF session is for. Am I correct that it is meant to reach out beyond the Forum? > > I tried to cut it down even more. I think we need a bit more than > in your proposal, but I understand your direction. > >> >> I think it is best to have the user relate the info to his own experience. That's why in my proposal I just wrote "it like PAPI just for 'Software Counters' and with read--write access". A single sentence and whoever is listening should have a clear view on what we want to accomplish. If we directly want comments on the internals, people have to read the draft first anyway. > > I added the PAPI example back. > >> >> Also. I would not cite the direct link to the drafts. People will not be able to memorize the whole url, so I think the critical info is: svn.mpi-forum.org and "Tools WG". Then, I would place a link to each draft prominently on the beginning of the page (maybe below "List of …") > > Good point - I changed the slides and I will change the Wiki > accordingly. > > > Bronis: > >> The animated MPIT slide is good. I tweaked the animation >> slightly in the attached so that each examples text box >> appears just after the relevant animation. > > That actually changed the meaning - the example had the > setup phase to the left and then an example for a performance > variable on the right. Control variables were not in. I changed > the text to avoid this misunderstanding when Rich is presenting. > >> >> Also, I thought we agreed that we would not use hidden slides >> as it is confusing. In any event, I think you need some words > > Those were really intended only for the Web version only. > >> on MPIT even with the animation. Further, the hidden MPIR >> texxt slide was almost identical to the unhidden one. I unhid >> all of the slides in the attached and merged the two MPIR >> text slides. A good tweak would be to animate the MPIR process >> figure but that would be above and beyond. > > If we don't want any hidden slides - the last one seems too much. > I got rid of it - people can look at the document. > >> >> Overall the current version does put us a couple slides over >> the request. However, it goes over due to the graphic slides, >> whichwill help to make the talk more interesting so I would >> hope that Rich would be OK with that. > > We are now 1 slide over, so I hope this is OK. > > > Kathryn: > >> I really like the examples and animation you added to slide 2! Now that you have that, I think you can remove some of the text from slide 3 to make it less dense without losing information, e.g. the examples of variables. I edited slide 3 so that it's somewhat less dense but still something that I think Rich can read from and get the information we want across (attached slide 3). > > Slide 3 was not supposed to be shown and only intended for the > web version. Since both you and Bronis seemed to not like this, > I added your modified slide as an unhidden one (in which case > I like the reduced text). > >> >> Also, there was a typo in slide 3 that I think you will want to fix in any case: "if the MPI implementation support this" change support to supports. > > That's in there now. > > > > > > > ________________________________________________________________________ > Martin Schulz, schulzm(a)llnl.gov, http://people.llnl.gov/schulzm > CASC @ Lawrence Livermore National Laboratory, Livermore, USA > > > > _______________________________________________ > Mpi3-tools mailing list > Mpi3-tools(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-tools -- Jeff Squyres jsquyres(a)cisco.com For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/

1 0

Re: [Mpi3-tools] Final MPI-3 Tools BOF Slides
by Martin Schulz 06 Nov '10

06 Nov '10

Hi Bronis, On Nov 6, 2010, at 6:00 PM, Bronis R. de Supinski wrote: > > Martin: > > Re: >> Thanks to all for the comments and the feedback. I tried to >> include all comments (more detailed responses summarized >> below). Attached are the final slides that I will send to Rich. > > This version is fine but I have no idea what you meant by > "more detailed responses summarized below" -- perhaps you > forgot to include something into your email. This is very strange - somehow Apple Mail or Exchange put all text in my email that was below the attachment into a separate HTML attachment, which should be attached to the original email (at least that's how my outfolder has it). I have not seen this before - the text is appended below again just to make sure (not that it is overly important at this point). Martin Marc-Andre: > I think the slides are too full. Maybe I am missing what the BoF session is for. Am I correct that it is meant to reach out beyond the Forum? I tried to cut it down even more. I think we need a bit more than in your proposal, but I understand your direction. > > I think it is best to have the user relate the info to his own experience. That's why in my proposal I just wrote "it like PAPI just for 'Software Counters' and with read--write access". A single sentence and whoever is listening should have a clear view on what we want to accomplish. If we directly want comments on the internals, people have to read the draft first anyway. I added the PAPI example back. > > Also. I would not cite the direct link to the drafts. People will not be able to memorize the whole url, so I think the critical info is: svn.mpi-forum.org and "Tools WG". Then, I would place a link to each draft prominently on the beginning of the page (maybe below "List of …") Good point - I changed the slides and I will change the Wiki accordingly. Bronis: > The animated MPIT slide is good. I tweaked the animation > slightly in the attached so that each examples text box > appears just after the relevant animation. That actually changed the meaning - the example had the setup phase to the left and then an example for a performance variable on the right. Control variables were not in. I changed the text to avoid this misunderstanding when Rich is presenting. > > Also, I thought we agreed that we would not use hidden slides > as it is confusing. In any event, I think you need some words Those were really intended only for the Web version only. > on MPIT even with the animation. Further, the hidden MPIR > texxt slide was almost identical to the unhidden one. I unhid > all of the slides in the attached and merged the two MPIR > text slides. A good tweak would be to animate the MPIR process > figure but that would be above and beyond. If we don't want any hidden slides - the last one seems too much. I got rid of it - people can look at the document. > > Overall the current version does put us a couple slides over > the request. However, it goes over due to the graphic slides, > whichwill help to make the talk more interesting so I would > hope that Rich would be OK with that. We are now 1 slide over, so I hope this is OK. Kathryn: > I really like the examples and animation you added to slide 2! Now that you have that, I think you can remove some of the text from slide 3 to make it less dense without losing information, e.g. the examples of variables. I edited slide 3 so that it's somewhat less dense but still something that I think Rich can read from and get the information we want across (attached slide 3). Slide 3 was not supposed to be shown and only intended for the web version. Since both you and Bronis seemed to not like this, I added your modified slide as an unhidden one (in which case I like the reduced text). > > Also, there was a typo in slide 3 that I think you will want to fix in any case: "if the MPI implementation support this" change support to supports. That's in there now. ________________________________________________________________________ Martin Schulz, schulzm(a)llnl.gov, http://people.llnl.gov/schulzm CASC @ Lawrence Livermore National Laboratory, Livermore, USA > > Bronis > > _______________________________________________ > Mpi3-tools mailing list > Mpi3-tools(a)lists.mpi-forum.org > http://BLOCKEDlists.mpi-forum.org/mailman/listinfo.cgi/mpi3-tools > ________________________________________________________________________ Martin Schulz, schulzm(a)llnl.gov, http://people.llnl.gov/schulzm CASC @ Lawrence Livermore National Laboratory, Livermore, USA

1 0

Re: [Mpi3-tools] Final MPI-3 Tools BOF Slides
by Bronis R. de Supinski 06 Nov '10

06 Nov '10

Martin: Re: > Thanks to all for the comments and the feedback. I tried to > include all comments (more detailed responses summarized > below). Attached are the final slides that I will send to Rich. This version is fine but I have no idea what you meant by "more detailed responses summarized below" -- perhaps you forgot to include something into your email. Bronis

1 0

[Mpi3-tools] Final MPI-3 Tools BOF Slides
by Martin Schulz 06 Nov '10

06 Nov '10

Hi all, Thanks to all for the comments and the feedback. I tried to include all comments (more detailed responses summarized below). Attached are the final slides that I will send to Rich. Martin Marc-Andre: > I think the slides are too full. Maybe I am missing what the BoF session is for. Am I correct that it is meant to reach out beyond the Forum? I tried to cut it down even more. I think we need a bit more than in your proposal, but I understand your direction. > > I think it is best to have the user relate the info to his own experience. That's why in my proposal I just wrote "it like PAPI just for 'Software Counters' and with read--write access". A single sentence and whoever is listening should have a clear view on what we want to accomplish. If we directly want comments on the internals, people have to read the draft first anyway. I added the PAPI example back. > > Also. I would not cite the direct link to the drafts. People will not be able to memorize the whole url, so I think the critical info is: svn.mpi-forum.org and "Tools WG". Then, I would place a link to each draft prominently on the beginning of the page (maybe below "List of …") Good point - I changed the slides and I will change the Wiki accordingly. Bronis: > The animated MPIT slide is good. I tweaked the animation > slightly in the attached so that each examples text box > appears just after the relevant animation. That actually changed the meaning - the example had the setup phase to the left and then an example for a performance variable on the right. Control variables were not in. I changed the text to avoid this misunderstanding when Rich is presenting. > > Also, I thought we agreed that we would not use hidden slides > as it is confusing. In any event, I think you need some words Those were really intended only for the Web version only. > on MPIT even with the animation. Further, the hidden MPIR > texxt slide was almost identical to the unhidden one. I unhid > all of the slides in the attached and merged the two MPIR > text slides. A good tweak would be to animate the MPIR process > figure but that would be above and beyond. If we don't want any hidden slides - the last one seems too much. I got rid of it - people can look at the document. > > Overall the current version does put us a couple slides over > the request. However, it goes over due to the graphic slides, > whichwill help to make the talk more interesting so I would > hope that Rich would be OK with that. We are now 1 slide over, so I hope this is OK. Kathryn: > I really like the examples and animation you added to slide 2! Now that you have that, I think you can remove some of the text from slide 3 to make it less dense without losing information, e.g. the examples of variables. I edited slide 3 so that it's somewhat less dense but still something that I think Rich can read from and get the information we want across (attached slide 3). Slide 3 was not supposed to be shown and only intended for the web version. Since both you and Bronis seemed to not like this, I added your modified slide as an unhidden one (in which case I like the reduced text). > > Also, there was a typo in slide 3 that I think you will want to fix in any case: "if the MPI implementation support this" change support to supports. That's in there now. ________________________________________________________________________ Martin Schulz, schulzm(a)llnl.gov, http://people.llnl.gov/schulzm CASC @ Lawrence Livermore National Laboratory, Livermore, USA

1 0

Re: [Mpi3-tools] New Version of MPI BoF Slides
by Mohror, Kathryn 05 Nov '10

05 Nov '10

Hi Martin, I really like the examples and animation you added to slide 2! Now that you have that, I think you can remove some of the text from slide 3 to make it less dense without losing information, e.g. the examples of variables. I edited slide 3 so that it's somewhat less dense but still something that I think Rich can read from and get the information we want across (attached slide 3). Also, there was a typo in slide 3 that I think you will want to fix in any case: "if the MPI implementation support this" change support to supports. Kathryn > -----Original Message----- > From: mpi3-tools-bounces(a)lists.mpi-forum.org [mailto:mpi3-tools- > bounces(a)lists.mpi-forum.org] On Behalf Of Bronis R. de Supinski > Sent: Friday, November 05, 2010 3:58 AM > To: mpi3-tools(a)lists.mpi-forum.org > Subject: Re: [Mpi3-tools] New Version of MPI BoF Slides > > > Martin: > > Re: > > It took me a while to get back to it, but here is a new version of the > > slides for the MPI BoF. Note that slides 3,5 and 6 are hidden and only > > for the web version. > > > > I hope this includes most/all of the suggestions - let me know if you > > have any further comments (we need to wrap this up very soon, though). > > The animated MPIT slide is good. I tweaked the animation slightly in the > attached so that each examples text box appears just after the relevant > animation. > > Also, I thought we agreed that we would not use hidden slides as it is > confusing. In any event, I think you need some words on MPIT even with the > animation. Further, the hidden MPIR texxt slide was almost identical to the > unhidden one. I unhid all of the slides in the attached and merged the two > MPIR text slides. A good tweak would be to animate the MPIR process figure > but that would be above and beyond. > > Overall the current version does put us a couple slides over the request. > However, it goes over due to the graphic slides, whichwill help to make the > talk more interesting so I would hope that Rich would be OK with that. > > Bronis

1 0

Re: [Mpi3-tools] New Version of MPI BoF Slides
by Bronis R. de Supinski 05 Nov '10

05 Nov '10

Martin: Re: > It took me a while to get back to it, but here is a new version of > the slides for the MPI BoF. Note that slides 3,5 and 6 are hidden > and only for the web version. > > I hope this includes most/all of the suggestions - let me know > if you have any further comments (we need to wrap this up > very soon, though). The animated MPIT slide is good. I tweaked the animation slightly in the attached so that each examples text box appears just after the relevant animation. Also, I thought we agreed that we would not use hidden slides as it is confusing. In any event, I think you need some words on MPIT even with the animation. Further, the hidden MPIR texxt slide was almost identical to the unhidden one. I unhid all of the slides in the attached and merged the two MPIR text slides. A good tweak would be to animate the MPIR process figure but that would be above and beyond. Overall the current version does put us a couple slides over the request. However, it goes over due to the graphic slides, whichwill help to make the talk more interesting so I would hope that Rich would be OK with that. Bronis

1 0