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

Re: [mpiwg-tools] Questions for the tools community
by Jeff Squyres (jsquyres) 29 Aug '19

29 Aug '19

Should I infer from the lack of reply that no tools properly handle either Fortran descriptors or the descriptor-based symbols? > On Aug 25, 2019, at 11:31 AM, Jeff Squyres (jsquyres) via mpiwg-tools <mpiwg-tools(a)lists.mpi-forum.org> wrote: > > In Open MPI, we are finally working towards supporting Fortran descriptors for choice buffers (i.e., "TYPE(*), DIMENSION(..)"). > > Per MPI-3.1 p613 table 17.1, this means ***changing all the back-end symbols*** that are used for PMPI-style interception. > > A concrete example: > > - in mpif.h / the mpi module, MPI_SEND will become MPI_SEND_FTS > - in the mpi_f08 module, MPI_SEND will become MPI_SEND_F08TS > > I.e., anywhere the application calls MPI_SEND, it will be transmogrified to either MPI_SEND_FTS or MPI_SEND_F08TS. And the choice buffer passed will be a descriptor (which may therefore be a subarray, not a contiguous buffer), not a naked pointer. > > Here's my questions: > > 1. Do any tools support Fortran descriptors as choice buffers? > 2. Do any tools intercept the symbols as described in MPI-3.1 table 17.1? > > -- > Jeff Squyres > jsquyres(a)cisco.com > > _______________________________________________ > mpiwg-tools mailing list > mpiwg-tools(a)lists.mpi-forum.org > https://lists.mpi-forum.org/mailman/listinfo/mpiwg-tools -- Jeff Squyres jsquyres(a)cisco.com

1 0

Next calls and face-to-face meeting
by Marc-André Hermanns 27 Aug '19

27 Aug '19

Dear all, I have yet another conflict this week and with the face-to-face meeting and EuroMPI comping up, we'll skip the next three regular calls on: - August 27, - September 5, and - September 12. There is a Tools WG session and some plenary time regarding readings relevant for the Tools WG at the face-to-face. WebEx details will be updated in due time. Please consult the meetings page [1] for details. Cheers, Marc-Andre [1] https://www.mpi-forum.org/meetings/2019/09/agenda -- Dr. rer. nat. Marc-André Hermanns IT Center Group: High Performance Computing Division: Computational Science and Engineering RWTH Aachen University Seffenter Weg 23 52074 Aachen Phone: +49 241 80-24381 hermanns(a)itc.rwth-aachen.de www.itc.rwth-aachen.de

1 0

Questions for the tools community
by Jeff Squyres (jsquyres) 25 Aug '19

25 Aug '19

In Open MPI, we are finally working towards supporting Fortran descriptors for choice buffers (i.e., "TYPE(*), DIMENSION(..)"). Per MPI-3.1 p613 table 17.1, this means ***changing all the back-end symbols*** that are used for PMPI-style interception. A concrete example: - in mpif.h / the mpi module, MPI_SEND will become MPI_SEND_FTS - in the mpi_f08 module, MPI_SEND will become MPI_SEND_F08TS I.e., anywhere the application calls MPI_SEND, it will be transmogrified to either MPI_SEND_FTS or MPI_SEND_F08TS. And the choice buffer passed will be a descriptor (which may therefore be a subarray, not a contiguous buffer), not a naked pointer. Here's my questions: 1. Do any tools support Fortran descriptors as choice buffers? 2. Do any tools intercept the symbols as described in MPI-3.1 table 17.1? -- Jeff Squyres jsquyres(a)cisco.com

1 0

No call today
by Hermanns, Marc-André 22 Aug '19

22 Aug '19

Dear all, as I am still on travel, we'll skip today's call again. Cheers, Marc-Andre -- Dr. rer. nat. Marc-André Hermanns IT Center Group: High Performance Computing Division: Computational Science and Engineering RWTH Aachen University Seffenter Weg 23 52074 Aachen Phone: +49 241 80-24381 hermanns(a)itc.rwth-aachen.de www.itc.rwth-aachen.de

1 0

No call today
by Hermanns, Marc-André 15 Aug '19

15 Aug '19

Dear all, as Kathryn, Martin, and I are out of office, we'll skip the call today. Cheers, Marc-André Sent from phone. -- Dr. rer. nat. Marc-André Hermanns IT Center Group: High Performance Computing Division: Computational Science and Engineering RWTH Aachen University Seffenter Weg 23 52074 Aachen Phone: +49 241 80-24381 hermanns(a)itc.rwth-aachen.de www.itc.rwth-aachen.de

1 0

Meeting reminder: Aug 8, 2019 (today)
by Marc-André Hermanns 08 Aug '19

08 Aug '19

Hi everyone, This is a reminder of our regularly scheduled call today August 8 at 8 am Pacific / 11 am Eastern / 5pm CET. Webex Info: https://github.com/mpiwg-tools/mpi-standard/wiki *NOTE*: The WebEx information moved to to a non-public wiki. You have to be logged into Github *and* have to have access to the private Tools WG repositories. If you need access contact Kathryn or me. Agenda: ------- - QMPI Prototype walk-through - MPI_T Events open issues Cheers, Marc-Andre -- Dr. rer. nat. Marc-André Hermanns IT Center Group: High Performance Computing Division: Computational Science and Engineering RWTH Aachen University Seffenter Weg 23 52074 Aachen Phone: +49 241 80-24381 hermanns(a)itc.rwth-aachen.de www.itc.rwth-aachen.de

1 0

Meeting reminder: Aug 1, 2019 (today)
by Marc-André Hermanns 01 Aug '19

01 Aug '19

Hi everyone, This is a reminder of our regularly scheduled call today August 1 at 8 am Pacific / 11 am Eastern / 5pm CET. Webex Info: https://github.com/mpiwg-tools/mpi-standard/wiki *NOTE*: The WebEx information moved to to a non-public wiki. You have to be logged into Github *and* have to have access to the private Tools WG repositories. If you need access contact Kathryn or me. Agenda: ------- - MPI_Info before Init (status) - MPI_T Events update (update, if any)? - MPI_T Universal IDs (discussion) Please note that Matt Legendre will *not* be able to join today, so we won't talk in detail on stack walking for QMPI. Also, we shifted the QMPI code walk-through to next week. Cheers, Marc-Andre -- Dr. Marc-Andre Hermanns Jülich Aachen Research Alliance, High Performance Computing (JARA-HPC) Jülich Supercomputing Centre (JSC) Wilhelm-Johnen-Str. 52425 Jülich Germany Phone: +49 2461 61 2509 | +49 241 80 24381 Fax: +49 2461 80 6 99753 www.jara.org/jara-hpc email: hermanns(a)jara.rwth-aachen.de -- Dr. rer. nat. Marc-André Hermanns IT Center Group: High Performance Computing Division: Computational Science and Engineering RWTH Aachen University Seffenter Weg 23 52074 Aachen Phone: +49 241 80-24381 hermanns(a)itc.rwth-aachen.de www.itc.rwth-aachen.de

1 0

Fwd: [mpiwg-sessions] notes from 7/29/19 webex
by Marc-André Hermanns 29 Jul '19

29 Jul '19

Dear all, attached you find Howard's notes on a call we just had among the people involved with PRs currently in need of MPI_Info before init (MPI_T Events, Session, FT). Cheers, Marc-Andre -------- Forwarded Message -------- Subject: [mpiwg-sessions] notes from 7/29/19 webex Date: Mon, 29 Jul 2019 18:29:45 +0000 From: Pritchard Jr., Howard via mpiwg-sessions <mpiwg-sessions(a)lists.mpi-forum.org> Reply-To: MPI Sessions working group <mpiwg-sessions(a)lists.mpi-forum.org> To: MPI Sessions working group <mpiwg-sessions(a)lists.mpi-forum.org> CC: Pritchard Jr., Howard <howardp(a)lanl.gov> Hello Folks Some notes from the 7/29/19 webex. Attending: Dan Holmes, Wesley Bland, Aurelien Bouteiller, Marc-Andre Hermanns, Howard Pritchard Dan turned on recording of the meeting. Dan pointed out a conflict about info keys – namely how they will be accessed before MPI initialization and after MPI finalization. Dan and Mark-Andre had discussed at ISC and were of the opinion that changes for MPI info keys should go into the sessions (PR 55) proposal. Sessions is being considered as a gating factor for 4.0. We need to figure out what is the difference between the info object changes in the. Aurelien says all the FT WG is wanting is to have MPI_INFO_ENV available before MPI initialization so the app can figure out what error handler has been set. Tools and Sessions don’t really care so much about this being available or not before MPI initialization. Dan discusses compromise that Sessions has taken. Wesley says FT needs MPI_Error_class and MPI_Error_string to be invocable prior to MPI initialization as well. Aurelien thinks this is in PR 50. Dan thinks the issue of argc/argv and how to fill in MPI_INFO_ENV is the key question. Aurelien thinks maybe add a new function to set MPI_INFO_ENV? Maybe a new ticket to handle this. FT group will work on this. Dan says we need to go through discrepancies between tickets. Sessions will need to pull in some of the FT adds to callable-prior-to-MPI Init to the table in Chapter 10 of PR 55. Mark-Andre raised some concerns about MPI_Info_get and error cases, but we think this will be covered by the setting-default-error handler before MPI initialization approach. Howard -- Howard Pritchard HPC-ENV Los Alamos National Laboratory -- Dr. rer. nat. Marc-André Hermanns IT Center Group: High Performance Computing Division: Computational Science and Engineering RWTH Aachen University Seffenter Weg 23 52074 Aachen Phone: +49 241 80-24381 hermanns(a)itc.rwth-aachen.de www.itc.rwth-aachen.de

1 0

Meeting reminder: July 25, 2019
by Marc-Andre Hermanns 25 Jul '19

25 Jul '19

Hi everyone, This is a reminder of our regularly scheduled call today July 25 at 8 am Pacific / 11 am Eastern / 5pm CET. Webex Info: https://github.com/mpiwg-tools/mpi-standard/wiki *NOTE*: The WebEx information moved to to a non-public wiki. You have to be logged into Github *and* have to have access to the private Tools WG repositories. If you need access contact Kathryn or me. Agenda: ------- - QMPI: Matt LeGendre will provide details on stack-walking techniques Cheers, Marc-Andre -- Dr. Marc-Andre Hermanns Jülich Aachen Research Alliance, High Performance Computing (JARA-HPC) Jülich Supercomputing Centre (JSC) Wilhelm-Johnen-Str. 52425 Jülich Germany Phone: +49 2461 61 2509 | +49 241 80 24381 Fax: +49 2461 80 6 99753 www.jara.org/jara-hpc email: hermanns(a)jara.rwth-aachen.de

1 0

Running late
by Hermanns, Marc-André 18 Jul '19

18 Jul '19

Hi all I got delayed on my way home. Will be a couple of minutes late. Cheers Marc-Andre Sent from phone.

1 0