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

Re: [Mpi3-tools] MPIT profiling
by Marty Itzkowitz 18 Oct '10

18 Oct '10

On 10/18/10 9:13 AM, Jeff Squyres wrote: > 2. Marty: It is worthwhile to be able to profile the impact of MPIT itself -- e.g., see if MPIT itself is performing poorly. I extrapolate this to mean that it can be useful to help quantify the at least part of the MPIT's own overhead (as I typed that sentence, I'm not sure how much of MPIT's overhead will be measurable by simply timing how long its individual function calls...? I guess there's other metrics that can be measured, too...?). Statistical clock-profiling could do it without any interface in the MPIT APIs. So it is probably not worth putting it in. It's a trade-off between observability and performance and complexity. Even having argued that it's nice to be able to observe, I'm happy with not adding the extra complexity. Marty

1 0

[Mpi3-tools] MPIT profiling
by Jeff Squyres 18 Oct '10

18 Oct '10

We ran out of time before finishing the discussion today... My take on section 1.2.8 of the MPIT document is that it's weird/inconsistent that we're saying that you can use a PMPIT interface to profile the MPIT interface, and that it basically behaves exactly like PMPI. I claim that this is inconsistent with the MPIT's goal of allowing multiple tools to be active in a single executable -- what if multiple tools want to use the PMPIT interface? You can't do that without using an external thingy (like PNMPI -- which would need to be expanded to support the MPIT interface). Meaning: since supporting multiple tools in a single executable is an explicit goal, we should either allow MPIT profiling by multiple tools within a single process or we shouldn't allow it at all. Background: Two use cases for profiling MPIT cited on the call were: 1. Martin: Tools adding their own control / performance variables. My suggestion here is that maybe we can have a registration functionality for this (i.e., an MPIT function to register new control/performance variables). I have not thought this through deeply; my only point is that if we know tool writers want to do something (and Martin says that the tool writers do want to do it), then we should provide them a direct way to do it. Not an indirect way. 2. Marty: It is worthwhile to be able to profile the impact of MPIT itself -- e.g., see if MPIT itself is performing poorly. I extrapolate this to mean that it can be useful to help quantify the at least part of the MPIT's own overhead (as I typed that sentence, I'm not sure how much of MPIT's overhead will be measurable by simply timing how long its individual function calls...? I guess there's other metrics that can be measured, too...?). -- 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] Reminder: Next MPI-3 Tools WG TelCon: Monday 10/18
by Marc-André Hermanns 18 Oct '10

18 Oct '10

Hi Martin, I got feedback from Andreas (Kuntze), who is working on implementing the MPIT interface in MP-MPICH. Attached you find two documents he sent to me last week. Note that the file about his experience with MPIT refers to the pre-snowbird-comments document. On the weekend he wrote in an email that several of the issues he brings up have already been dealt with in the current document. I told him about today's telcon, however, I do not have feedback yet, whether he will be able to join. Cheers, Marc-Andre On 18.10.2010, at 06:13, Martin Schulz wrote: > Hi all, > > Just as a reminder: our next MPI-3 tools TelCon is scheduled > for tomorrow, Monday 10/18, at the usual time 8am PDT, 11am > CDT, 5pm MSZ. To login use: > > https://cisco.webex.com/ciscosales/j.php?ED=150743227&UID=0&PW=NZTRhZjI2NGF… > Password: tools > > We'll continue the discussion of the MPIT interface, which we started > last week at the MPI forum. The current draft of the MPIT proposal > is at: > > https://svn.mpi-forum.org/trac/mpi-forum-web/wiki/MPI3Tools/mpit > > Talk to you on Monday, > > Martin > ________________________________________________________________________ > Martin Schulz, schulzm(a)llnl.gov, http://people.llnl.gov/schulzm > CASC @ Lawrence Livermore National Laboratory, Livermore, USA > -- Marc-Andre Hermanns German Research School for Simulation Sciences GmbH c/o Laboratory for Parallel Programming 52056 Aachen | Germany Tel +49 241 80 99753 Fax +49 241 80 6 99753 Web www.grs-sim.de Members: Forschungszentrum Jülich GmbH | RWTH Aachen University Registered in the commercial register of the local court of Düren (Amtsgericht Düren) under registration number HRB 5268 Registered office: Jülich Executive board: Prof. Marek Behr Ph.D. | Dr. Norbert Drewes

1 0

[Mpi3-tools] Reminder: Next MPI-3 Tools WG TelCon: Monday 10/18
by Martin Schulz 17 Oct '10

17 Oct '10

Hi all, Just as a reminder: our next MPI-3 tools TelCon is scheduled for tomorrow, Monday 10/18, at the usual time 8am PDT, 11am CDT, 5pm MSZ. To login use: https://cisco.webex.com/ciscosales/j.php?ED=150743227&UID=0&PW=NZTRhZjI2NGF… Password: tools We'll continue the discussion of the MPIT interface, which we started last week at the MPI forum. The current draft of the MPIT proposal is at: https://svn.mpi-forum.org/trac/mpi-forum-web/wiki/MPI3Tools/mpit Talk to you on Monday, Martin ________________________________________________________________________ Martin Schulz, schulzm(a)llnl.gov, http://people.llnl.gov/schulzm CASC @ Lawrence Livermore National Laboratory, Livermore, USA

1 0

[Mpi3-tools] Notes 10/12 - Next meeting 10/18 8am PDT
by Martin Schulz 12 Oct '10

12 Oct '10

Hi all, During today's MPI-3 tools WG meeting during the MPI forum in Chicago we started going over the new MPIT draft. The notes from the discussion are on the WiKi and I will start integrating them into the document. We got to about page 10 and will continue the discussion during the next TelCons. The next one is scheduled for next Monday at 10/18 at the usual time at 8am PDT, 11am EDT, 5pm MSZ. Martin ________________________________________________________________________ Martin Schulz, schulzm(a)llnl.gov, http://people.llnl.gov/schulzm CASC @ Lawrence Livermore National Laboratory, Livermore, USA

1 0

Re: [Mpi3-tools] Fwd: [MPI Forum] #228: MPIR Specification - An Official Companion Document for the MPI Standard
by Jeff Squyres 11 Oct '10

11 Oct '10

On Oct 11, 2010, at 2:15 PM, Jeff Squyres wrote: > Martin and I talked about this in person: technically, this is a change in the document that was formally read in Stuttgart. I will ask the Forum if they're comfortable voting on this, or if we reset the clock and call this meeting the really-first-reading (and we have a first vote in December). Technically, we should do the latter, but it may be a small enough change to eek through. The Forum was ok voting on the document, even with the 1 word change. So we've passed 1 vote -- 1 more to go! -- 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] Fwd: [MPI Forum] #228: MPIR Specification - An Official Companion Document for the MPI Standard
by Jeff Squyres 11 Oct '10

11 Oct '10

On Oct 7, 2010, at 9:28 AM, Supalov, Alexander wrote: >>> Our definition: >>> volatile int MPIR_debug_state >>> >>> Proposal definition: >>> int MPIR_debug_state >>> >>> This may be material, as the MPIR_debug_state is set by the program and >>> inspected by the tool. We should probably add VOLATILE to the proposal. > > I think you're right, the document needs to be updated. I looked back at the original text document that described the interface, and it has: > > extern VOLATILE int MPIR_debug_state; I just updated the document to make MPIR_debug_state be VOLATILE. I'm up in a few minutes for the 1st vote here at the Forum in Chicago. Martin and I talked about this in person: technically, this is a change in the document that was formally read in Stuttgart. I will ask the Forum if they're comfortable voting on this, or if we reset the clock and call this meeting the really-first-reading (and we have a first vote in December). Technically, we should do the latter, but it may be a small enough change to eek through. I'll let you know how it goes. -- 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] New version of the MPIT Document
by Marc-André Hermanns 11 Oct '10

11 Oct '10

Hi Martin, thanks for the pointers. I couldn't quite relate the discussion from back then to the snowbird meeting. It's all coming back now ;) I won't be present at the TelCon tomorrow, but will try to work through and comment on your last document that you sent around. Cheers, Marc-Andre >> I seemed to have missed the discussion of the snowbird comments. >> Do you have some notes in electronic form that you could share. >> It would make it easier to read the new document to see how they >> relate to the comments. > > The workshop web page is: > > http://cscads.rice.edu/workshops/summer-2010/performance-tools > > It has all the slides that were presented, incl. the ones from the > MPIT breakout session. > > Also, the notes from the MPI-3 tools WG TelCon discussing the > workshop results are at: > > https://svn.mpi-forum.org/trac/mpi-forum-web/wiki/MPI3Tools/notes-2010-08-23 -- Marc-Andre Hermanns German Research School for Simulation Sciences GmbH c/o Laboratory for Parallel Programming 52056 Aachen | Germany Tel +49 241 80 99753 Fax +49 241 80 6 99753 Web www.grs-sim.de Members: Forschungszentrum Jülich GmbH | RWTH Aachen University Registered in the commercial register of the local court of Düren (Amtsgericht Düren) under registration number HRB 5268 Registered office: Jülich Executive board: Prof. Marek Behr Ph.D. | Dr. Norbert Drewes

1 0

Re: [Mpi3-tools] New version of the MPIT Document
by Martin Schulz 10 Oct '10

10 Oct '10

Hi all, Attached is a slightly revised version of the new MPIT document. Changes compared to the version on Friday are marked. I will upload this version to the Wiki and SVN later tonight once I am in Chicago. Martin On Oct 8, 2010, at 2:13 PM, Martin Schulz wrote: > Hi all, > > Attached is the new draft of the MPIT proposal for our discussion > next week in Chicago (during the Tuesday morning WG session). > This ended up being almost a complete rewrite and hence I didn't > use change bars this time - just assume that everything is new :). > > The text now includes the outcome of the discussions we had within > the WG as well as the extensive feedback I got from the tools > community at a workshop in the summer (and that we discussed > a few weeks back). > > A few items remain to be done and those are listed at the > beginning of the chapter. I would like to go over those during > the meeting as well and see if we can resolve them. > > Let me know if you have any questions or any immediate > feedback - otherwise we can address the rest next week. > > Thanks, > > Martin > > PS: I will also post this to the WiKi in a moment and check it > into the SVN repository. > > <mpi-report.pdf><ATT00001..txt> ________________________________________________________________________ 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 the MPIT Document
by Martin Schulz 10 Oct '10

10 Oct '10

Hi Marc-Andre, all, On Oct 10, 2010, at 2:47 AM, Marc-Andrï¿½ Hermanns wrote: > Hi Martin, > > I seemed to have missed the discussion of the snowbird comments. > Do you have some notes in electronic form that you could share. > It would make it easier to read the new document to see how they > relate to the comments. The workshop web page is: http://cscads.rice.edu/workshops/summer-2010/performance-tools It has all the slides that were presented, incl. the ones from the MPIT breakout session. Also, the notes from the MPI-3 tools WG TelCon discussing the workshop results are at: https://svn.mpi-forum.org/trac/mpi-forum-web/wiki/MPI3Tools/notes-2010-08-23 Martin > > Cheers, > Marc-Andre > > On 08.10.2010, at 23:13, Martin Schulz wrote: > >> Hi all, >> >> Attached is the new draft of the MPIT proposal for our discussion >> next week in Chicago (during the Tuesday morning WG session). >> This ended up being almost a complete rewrite and hence I didn't >> use change bars this time - just assume that everything is new :). >> >> The text now includes the outcome of the discussions we had within >> the WG as well as the extensive feedback I got from the tools >> community at a workshop in the summer (and that we discussed >> a few weeks back). >> >> A few items remain to be done and those are listed at the >> beginning of the chapter. I would like to go over those during >> the meeting as well and see if we can resolve them. >> >> Let me know if you have any questions or any immediate >> feedback - otherwise we can address the rest next week. >> >> Thanks, >> >> Martin >> >> PS: I will also post this to the WiKi in a moment and check it >> into the SVN repository. >> >> <mpi-report.pdf> >> ________________________________________________________________________ >> 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 > > -- > Marc-Andre Hermanns > German Research School for > Simulation Sciences GmbH > c/o Laboratory for Parallel Programming > 52056 Aachen | Germany > > Tel +49 241 80 99753 > Fax +49 241 80 6 99753 > Web www.grs-sim.de > > Members: Forschungszentrum Jï¿½lich GmbH | RWTH Aachen University > Registered in the commercial register of the local court of > Dï¿½ren (Amtsgericht Dï¿½ren) under registration number HRB 5268 > Registered office: Jï¿½lich > Executive board: Prof. Marek Behr Ph.D. | Dr. Norbert Drewes > > <smime.p7s><ATT00001..txt> ________________________________________________________________________ Martin Schulz, schulzm(a)llnl.gov, http://people.llnl.gov/schulzm CASC @ Lawrence Livermore National Laboratory, Livermore, USA

1 0