- mpi-comments - mailman3.cels.anl.gov

Re: [Mpi-comments] One comment on MPI-3.0 Draft 2, August 2012
by Richard Graham 15 Aug '12

15 Aug '12

Sï¿½bastien, Thanks you for sending in this request, this is truly appreciated. The propose change is a large change to the standard, even though it may not seem so at a first glance. Today the standard is explicit about lack of fairness, and we would have to think through in detail on the broad impact of such a change. So, this does not the sort of item we can address for MPI 3.0. However, if you would like to push a change like this into MPI, this can be addressed for the following version of the standard. What I would suggest, if this is something that you really think belongs in the standard, that you either get involved directly in the process (the standardization effort is open to all), or find someone the is involved to actively push such a request. As I said, this is a non-trivial change, so that both API changes (if any) and implementation issues need to be addressed. Please let me know if you need help with figuring out how to get involved. Thanks again, Rich -----Original Message----- From: Sï¿½bastien Boisvert [mailto:[email protected]] Sent: Tuesday, August 14, 2012 9:47 AM To: mpi-comments(a)mpi-forum.org Subject: [Mpi-comments] One comment on MPI-3.0 Draft 2, August 2012 Dear MPI Forum committee members, I would like to submit a comment on the MPI-3.0 Draft 2, August 2012 for your consideration. Version: MPI-3.0 Draft 2, August 2012. The URL of the version of the MPI standard: http://meetings.mpi-forum.org/draft_standard/mpi3.0_draft_2.pdf Page: 65 Line number: 28 Section: 3.8.1 In: 3. Point-to-Point Communication 3.8 Probe and Cancel 3.8.1 Probe Comment: It says that the source argument of MPI_Iprobe can be MPI_ANY_SOURCE, but it does say anything about fairness. Therefore MPI_ANY_SOURCE can lead to resource starvation. I think it would be better if probing would be done in a round-robin fashion when the source is MPI_ANY_SOURCE so that any MPI rank has an equal chance of having its message probed and received. Presently, the MPI standard contains nothing about which source should be probed when MPI_ANY_SOURCE is provided. I hope you will consider my comment. Sincerely, Sï¿½bastien Boisvert PhD student Universitï¿½ Laval _______________________________________________ mpi-comments mailing list mpi-comments(a)lists.mpi-forum.org http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments

1 0

Re: [Mpi-comments] MPI_Is_thread_main: usage allowed for any thread in MPI_THREAD_FUNNELED mode?
by Richard Graham 15 Aug '12

15 Aug '12

Philippe, Thank you very much for taking the time to send in this comment - this is greatly appreciated. The request is a good request, and the working group in charge of this particular chapter has agreed to look into this. However, this will not make it into MPI 3.0, but is targeted for the following MPI. We have to carefully check the full list of affected routines, and do not have the time to do this correctly before MPI 3.0 is slated to be voted on. Thanks again, Rich -----Original Message----- From: Philippe Wautelet [mailto:[email protected]] Sent: Tuesday, August 14, 2012 10:12 AM To: mpi-comments(a)mpi-forum.org Subject: [Mpi-comments] MPI_Is_thread_main: usage allowed for any thread in MPI_THREAD_FUNNELED mode? Dear all, It is not clear to me if I am allowed to make calls to MPI_Is_thread_main from any thread if I initialized MPI with the MPI_THREAD_FUNNELED mode. For this mode, the norm says: "the application must ensure that only the main thread makes MPI calls" (page 384, lines 46-47) Therefore, a call to MPI_Is_thread_main should be incorrect. But in this case, how can I determine which thread is the main one? Or, maybe, some MPI calls could be explicitly allowed to be made by any thread? Best regards, Philippe Wautelet -------------------------------------------------------------------- Philippe Wautelet (Research Engineer) wautelet(a)idris.fr Tel. : 01 69 35 85 76 Fax : 01 69 85 37 75 IDRIS, Batiment 506, BP 167, 91403 ORSAY Cedex, France Prof. website : http://www.idris.fr Private website : http://www.fractalzone.be -------------------------------------------------------------------- _______________________________________________ mpi-comments mailing list mpi-comments(a)lists.mpi-forum.org http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments

1 0

Re: [Mpi-comments] More on "MPI-3.0 public draft and call for comments"
by Richard Graham 14 Aug '12

14 Aug '12

Fernando, First, thanks a lot for responding to the request for information - we really appreciate this. Comments in line. Thanks, Rich -----Original Message----- From: mpi-comments-bounces(a)lists.mpi-forum.org [mailto:[email protected]] On Behalf Of Fernando G. Tinetti Sent: Friday, August 10, 2012 9:53 AM To: mpi-comments(a)mpi-forum.org Subject: [Mpi-comments] More on "MPI-3.0 public draft and call for comments" Hi, I'm commenting/suggesting some details about http://meetings.mpi-forum.org/mpi3.0_draft_2.pdf because I received the e-mail from " Rolf Rabenseifner <rabenseifner(a)hlrs.de> to: fernando(a)info.unlp.edu.ar subject: MPI-3.0 public draft and call for comments " However, I did not receive any reply to my previous comment/question. Anyway, I'll make some extra comments, including the previous one: 1) I read at the end of page ii: " c 1993, 1994, 1995, 1996, 1997, 2008, 2009, 2012 University of Tennessee, Knoxville, Tennessee. Permission to copy without fee all or part of this material is granted, provided the University of Tennessee copyright notice and the title of this document appear, and notice is given that copying is by permission of the University of Tennessee. " Is it right? If so, could you give a short explanation (or a link to) about the reason/s? [rich] For historical reasons. The first copyright was owned by the University of Tennessee, as it helped start the forum in the first place, with Professor Dongarra being co-chair of the MPI-1 effort. 2) Given that of the "MPI: A Message-Passing Interface Standard" is from the "Message Passing Interface Forum" (page i in the .pdf) a minimum explanation of what the forum is and how it has evolved is in order. There are some explanations in pages iii-iv, references to MPI Forum, MPI-2 Forum (pp. iii, xxi, xxiv), and MPI-3 Forum (p. xxix) and I think it is not clear how those forums are made up e.g. the meeting itself is the Forum? The text at page 2 "These meetings and the email discussion together constituted the MPI Forum, membership of which has been open to all members of the high performance computing community." is clear about the 1993 decision, but I did not see a clear reference to the current Forum definition, Forum (committee/s) members, etc. I would like to see an informal MPI Forum evolution such as that given for MPI in pp. iii-iv. [rich] Look at meetings.mpi-forum.org - the current structure is described there. Membership and voting rights ebb and flow, as this is a voluntary organization, with organizational voting rights determined solely by participation at the Forum meetings. 3) I suggest a minor change in the text in p. 1 from " The goal of the Message-Passing Interface simply stated is to develop a widely used standard for writing message-passing programs. As such the interface should establish a practical, portable, e fficient, and flexible standard for message passing. A complete list of goals follows. " to " The goal of the Message-Passing Interface simply stated is to develop a widely used standard for writing message-passing programs. As such the interface should establish a practical, portable, e fficient, and flexible standard for message passing. A complete list of goals follows. " since I think the complete list is part of the paragraph defining goals of MPI. [rich] Thanks for the suggestion. In general, we try to avoid such sweeping statements, unless they add things to the standard that are not present in other parts of the document. We have found that with such a large document, when the standard moves forward it is too easy to miss such broad statements when the standard is being changed, producing inconsistencies in the document. Best regards, Fernando. --- La parte gris de las zapatillas es bien resbalosa, con ï¿½sas voy como Flash por la baranda de la escalera. (Santi, 7 aï¿½os) ---------------------------------------------------------------- This message was sent using IMP, the Internet Messaging Program. _______________________________________________ mpi-comments mailing list mpi-comments(a)lists.mpi-forum.org http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments

1 0

More on "MPI-3.0 public draft and call for comments"
by Fernando G. Tinetti 10 Aug '12

10 Aug '12

Hi, I'm commenting/suggesting some details about http://meetings.mpi-forum.org/mpi3.0_draft_2.pdf because I received the e-mail from " Rolf Rabenseifner <rabenseifner(a)hlrs.de> to: fernando(a)info.unlp.edu.ar subject: MPI-3.0 public draft and call for comments " However, I did not receive any reply to my previous comment/question. Anyway, I'll make some extra comments, including the previous one: 1) I read at the end of page ii: " c 1993, 1994, 1995, 1996, 1997, 2008, 2009, 2012 University of Tennessee, Knoxville, Tennessee. Permission to copy without fee all or part of this material is granted, provided the University of Tennessee copyright notice and the title of this document appear, and notice is given that copying is by permission of the University of Tennessee. " Is it right? If so, could you give a short explanation (or a link to) about the reason/s? 2) Given that of the "MPI: A Message-Passing Interface Standard" is from the "Message Passing Interface Forum" (page i in the .pdf) a minimum explanation of what the forum is and how it has evolved is in order. There are some explanations in pages iii-iv, references to MPI Forum, MPI-2 Forum (pp. iii, xxi, xxiv), and MPI-3 Forum (p. xxix) and I think it is not clear how those forums are made up e.g. the meeting itself is the Forum? The text at page 2 "These meetings and the email discussion together constituted the MPI Forum, membership of which has been open to all members of the high performance computing community." is clear about the 1993 decision, but I did not see a clear reference to the current Forum definition, Forum (committee/s) members, etc. I would like to see an informal MPI Forum evolution such as that given for MPI in pp. iii-iv. 3) I suggest a minor change in the text in p. 1 from " The goal of the Message-Passing Interface simply stated is to develop a widely used standard for writing message-passing programs. As such the interface should establish a practical, portable, efficient, and flexible standard for message passing. A complete list of goals follows. " to " The goal of the Message-Passing Interface simply stated is to develop a widely used standard for writing message-passing programs. As such the interface should establish a practical, portable, efficient, and flexible standard for message passing. A complete list of goals follows. " since I think the complete list is part of the paragraph defining goals of MPI. Best regards, Fernando. --- La parte gris de las zapatillas es bien resbalosa, con ï¿½sas voy como Flash por la baranda de la escalera. (Santi, 7 aï¿½os) ---------------------------------------------------------------- This message was sent using IMP, the Internet Messaging Program.

1 0

Comment on RMA: MPIX_ERR_RMA_WRONG_FLAVOR
by Jim Dinan 07 Aug '12

07 Aug '12

Hi All, [ moving posting from mpi3-rma list to mpi-comments ] In the RMA chapter, we define an error class, MPIX_ERR_RMA_WRONG_FLAVOR. The use of "WRONG" in the error class name is inconsistent with MPI naming conventions and the new error class should be renamed to MPIX_ERR_RMA_FLAVOR. ~Jim.

1 0

Re: [Mpi-comments] MPI-3.0 public draft and call for comments
by Rolf Rabenseifner 05 Aug '12

05 Aug '12

Bill, we should do this. Especially when looking for 16.2.2 or any other subsection number, it is hard to find it in the pdf. With Jurgen Rusch's idea, it should work. Jurgen, (sorry, my previous mail had wrong destination) thank you for your proposal! Whether minor or not, we pipilene all helpful comments through this email. Therefore I forwarded it. Best regards Rolf ----- Forwarded Message ----- From: "Jurgen Rusch" <jurgen.rusch(a)philips.com> To: "Rolf Rabenseifner" <rabenseifner(a)hlrs.de> Sent: Friday, August 3, 2012 3:43:26 PM Subject: RE: MPI-3.0 public draft and call for comments Dear Mister Rabenseifner, I'm responding to you via this informal e-mail channel and not via the recommended way to subscribe to the MPI-forum and send a comment to mpi-comments(a)mpi-forum.org. The reason is that I have a very minor, more PDF related, remark which I explain next. When I opened http://meetings.mpi-forum.org/mpi3.0_draft_2.pdf I saw that the bookmarks as shown in the left part of Adobe Reader do not contain the chapter/section numbers: they merely contain the text of the respective chapter, section, annex etc. Having seen that this document has been written in LaTeX (File --> Properties in Adobe Reader shows that the application "LaTeX with hyperref package" was used) I would like to tell to you and the other MPI-authors, that the hyperref package has an interesting option called "bookmarksnumbered" (default: false) which, when true, causes the bookmarks to include the section numbers. This is an option I often use which makes the bookmarks nicely numbered. For more information please see http://www.tug.org/applications/hyperref/manual.html (section 3.6, "PDF-specific display options"). With this e-mail I hope to have contributed a little bit to the MPI standard (although only PDF related). With kind regards, Jurgen Rusch (a former student of you) ________________________________ The information contained in this message may be confidential and legally protected under applicable law. The message is intended solely for the addressee(s). If you are not the intended recipient, you are hereby notified that any use, forwarding, dissemination, or reproduction of this message is strictly prohibited and may be unlawful. If you are not the intended recipient, please contact the sender by return e-mail and destroy all copies of the original message. -- Dr. Rolf Rabenseifner . . . . . . . . . .. email rabenseifner(a)hlrs.de High Performance Computing Center (HLRS) . phone ++49(0)711/685-65530 University of Stuttgart . . . . . . . . .. fax ++49(0)711 / 685-65832 Head of Dpmt Parallel Computing . . . www.hlrs.de/people/rabenseifner Nobelstr. 19, D-70550 Stuttgart, Germany . (Office: Allmandring 30)

1 0

Fwd: MPI-3.0 public draft and call for comments
by Rolf Rabenseifner 05 Aug '12

05 Aug '12

Felix, thank you! Forwarded to mpi-comments(a)mpi-forum.org Bets regards Rolf ----- Forwarded Message ----- From: "Felix Wolf" <f.wolf(a)fz-juelich.de> To: "Rolf Rabenseifner" <rabenseifner(a)hlrs.de> Sent: Friday, August 3, 2012 3:55:52 PM Subject: Re: MPI-3.0 public draft and call for comments Hallo Rolf, 2 stilistische Vorschläge (modulo Native-Speaker fragen): p. 563, l.14: "insight into internal performance information" -> "insight into internal performance behavior" or "insight into internal performance " p. 563, l. 44: "that number of variables" -> "that the number of variables" Viele Grüße, Felix On 8/3/12 2:32 PM, Rolf Rabenseifner wrote: > [Our apologies if you receive multiple copies or if you are not related to MPI] > > Dear MPI user, > > The MPI-Forum is about to ratify the MPI-3.0, a new version of the MPI standard. > As part of this process, we are soliciting feedback on the current draft standard. > The draft document can be found here: > > http://meetings.mpi-forum.org/mpi3.0_draft_2.pdf > > We are seeking the following feedback in particular: > > - Small requests/corrections that can be resolved before finishing 3.0. > - Requests for clarification of unclear text or undefined behavior. > - Undetected severe inconsistencies or bugs that may delay the > standard publication. > - Wishes for additional enhancements or new functionality > to be taken into account after 3.0 is published. > > Please comment before Sep. 6, 2012 as follows: > > 1. Subscribe to: http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments > 2. Then send your comments to: mpi-comments(a)mpi-forum.org , > together with the URL of the version of the MPI standard and > the page and line numbers on which you are commenting. > > Messages sent from an unsubscribed e-mail address will not be considered. > > Thank you in advance for your help. > > Best regards, > > The MPI Forum > > > ( www.mpi-forum.org ) -- Dr. Rolf Rabenseifner . . . . . . . . . .. email rabenseifner(a)hlrs.de High Performance Computing Center (HLRS) . phone ++49(0)711/685-65530 University of Stuttgart . . . . . . . . .. fax ++49(0)711 / 685-65832 Head of Dpmt Parallel Computing . . . www.hlrs.de/people/rabenseifner Nobelstr. 19, D-70550 Stuttgart, Germany . (Office: Allmandring 30)

1 0

One comment on MPI-3.0 Draft 2, August 2012
by Sébastien Boisvert 04 Aug '12

04 Aug '12

Dear MPI Forum committee members, I would like to submit a comment on the MPI-3.0 Draft 2, August 2012 for your consideration. Version: MPI-3.0 Draft 2, August 2012. The URL of the version of the MPI standard: http://meetings.mpi-forum.org/draft_standard/mpi3.0_draft_2.pdf Page: 65 Line number: 28 Section: 3.8.1 In: 3. Point-to-Point Communication 3.8 Probe and Cancel 3.8.1 Probe Comment: It says that the source argument of MPI_Iprobe can be MPI_ANY_SOURCE, but it does say anything about fairness. Therefore MPI_ANY_SOURCE can lead to resource starvation. I think it would be better if probing would be done in a round-robin fashion when the source is MPI_ANY_SOURCE so that any MPI rank has an equal chance of having its message probed and received. Presently, the MPI standard contains nothing about which source should be probed when MPI_ANY_SOURCE is provided. I hope you will consider my comment. Sincerely, Sï¿½bastien Boisvert PhD student Universitï¿½ Laval

1 0

Question
by Fernando G. Tinetti 03 Aug '12

03 Aug '12

Hi, In a quick look to http://meetings.mpi-forum.org/mpi3.0_draft_2.pdf I read at the end of page ii: " c 1993, 1994, 1995, 1996, 1997, 2008, 2009, 2012 University of Tennessee, Knoxville, Tennessee. Permission to copy without fee all or part of this material is granted, provided the University of Tennessee copyright notice and the title of this document appear, and notice is given that copying is by permission of the University of Tennessee. " Is it right? If so, could you give a short explanation (or a link to) about the reason/s? Thanks in advance, Fernando. --- La parte gris de las zapatillas es bien resbalosa, con ï¿½sas voy como Flash por la baranda de la escalera. (Santi, 7 aï¿½os) ---------------------------------------------------------------- This message was sent using IMP, the Internet Messaging Program.

1 0

AUTO: Rolf Riesen is on vacation (returning 07/08/2012)
by Rolf Riesen 03 Aug '12

03 Aug '12

I am out of the office until 07/08/2012. I'm on vacation (mostly) without email access. I'll respond to your email when I get back. Note: This is an automated response to your message "[Mpi-comments] Test Mail" sent on 3/8/2012 13:07:46. This is the only notification you will receive while this person is away.

1 0