From Edric.Ellis at [hidden] Fri Mar 7 07:56:32 2008 From: Edric.Ellis at [hidden] (Edric Ellis) Date: Fri, 7 Mar 2008 13:56:32 -0000 Subject: [Mpi3-abi] Slides for the meeting next week In-Reply-To: <6.2.3.4.2.20080222095907.02bea5b0@ccs-mail.lanl.gov> Message-ID: Hi All, We need to present something to the group at the start of the meeting on Monday. I hadn't seen anything about this, so I've taken the liberty of knocking something together from the proposal on the wiki. What do people think? I've also uploaded this to the wiki, so perhaps edits should be done there. I look forward to seeing you all at the meeting. I'm not going to be there right from the start (couldn't get an early enough flight), but I should be there towards the end of Monday. Cheers, Edric. * -------------- next part -------------- A non-text attachment was scrubbed... Name: ABI_slides_for_10th_March_2008.ppt Type: application/vnd.ms-powerpoint Size: 34816 bytes Desc: ABI slides for 10th March 2008.ppt URL: From jeffb at [hidden] Fri Mar 7 09:11:46 2008 From: jeffb at [hidden] (Jeff Brown) Date: Fri, 07 Mar 2008 08:11:46 -0700 Subject: [Mpi3-abi] Slides for the meeting next week In-Reply-To: Message-ID: <6.2.3.4.2.20080307081036.02b12498@ccs-mail.lanl.gov> I was going to put some slides together today - this is a good starting point. I'll do some edits and send out. Thanks for getting the ball rolling. Jeff At 06:56 AM 3/7/2008, you wrote: >Hi All, > >We need to present something to the group at the start of the meeting on >Monday. I hadn't seen anything about this, so I've taken the liberty of >knocking something together from the proposal on the wiki. What do >people think? I've also uploaded this to the wiki, so perhaps edits >should be done there. > >I look forward to seeing you all at the meeting. I'm not going to be >there right from the start (couldn't get an early enough flight), but I >should be there towards the end of Monday. > >Cheers, > >Edric. > >_______________________________________________ >mpi3-abi mailing list >mpi3-abi_at_[hidden] >http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi From jeffb at [hidden] Fri Mar 7 10:52:30 2008 From: jeffb at [hidden] (Jeff Brown) Date: Fri, 07 Mar 2008 09:52:30 -0700 Subject: [Mpi3-abi] Slides for the meeting next week In-Reply-To: Message-ID: <6.2.3.4.2.20080307094003.02d25a40@ccs-mail.lanl.gov> here's my cut at the slides (also posted to the wiki) I used material from Edric's slides and added thoughts of my own. I'll do the presentation Monday. We only have 5 minutes so I'll focus on the key points. Feel free to clarify during the briefing, but that's not the time for discussion. Jeff At 06:56 AM 3/7/2008, Edric Ellis wrote: >Hi All, > >We need to present something to the group at the start of the meeting on >Monday. I hadn't seen anything about this, so I've taken the liberty of >knocking something together from the proposal on the wiki. What do >people think? I've also uploaded this to the wiki, so perhaps edits >should be done there. > >I look forward to seeing you all at the meeting. I'm not going to be >there right from the start (couldn't get an early enough flight), but I >should be there towards the end of Monday. > >Cheers, > >Edric. > >_______________________________________________ >mpi3-abi mailing list >mpi3-abi_at_[hidden] >http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi * -------------- next part -------------- A non-text attachment was scrubbed... Name: MPI_ABI_Working_Group_status.ppt Type: application/octet-stream Size: 201728 bytes Desc: MPI_ABI_Working_Group_status.ppt URL: From alexander.supalov at [hidden] Fri Mar 7 15:17:55 2008 From: alexander.supalov at [hidden] (Supalov, Alexander) Date: Fri, 7 Mar 2008 21:17:55 -0000 Subject: [Mpi3-abi] Slides for the meeting next week In-Reply-To: <6.2.3.4.2.20080307094003.02d25a40@ccs-mail.lanl.gov> Message-ID: <5ECAB1304A8B5B4CB3F9D6C01E4E21A2011F4E0F@swsmsx413.ger.corp.intel.com> Dear Jeff, Thanks. I believe we should reference Bill Gropp's paper in the last slide (details). All material was shamelessly lifted from there by me, which is mentioned in the draft proposal. Also, to facilitate further member enrollment, it may make sense to explicitly mention the subscription link and invite people to join the Monday evening ABI F2F meeting. Best regards. Alexander -----Original Message----- From: mpi3-abi-bounces_at_[hidden] [mailto:mpi3-abi-bounces_at_[hidden]] On Behalf Of Jeff Brown Sent: Friday, March 07, 2008 5:52 PM To: MPI 3.0 ABI working group; mpi3-abi_at_[hidden] Subject: Re: [Mpi3-abi] Slides for the meeting next week here's my cut at the slides (also posted to the wiki) I used material from Edric's slides and added thoughts of my own. I'll do the presentation Monday. We only have 5 minutes so I'll focus on the key points. Feel free to clarify during the briefing, but that's not the time for discussion. Jeff At 06:56 AM 3/7/2008, Edric Ellis wrote: >Hi All, > >We need to present something to the group at the start of the meeting on >Monday. I hadn't seen anything about this, so I've taken the liberty of >knocking something together from the proposal on the wiki. What do >people think? I've also uploaded this to the wiki, so perhaps edits >should be done there. > >I look forward to seeing you all at the meeting. I'm not going to be >there right from the start (couldn't get an early enough flight), but I >should be there towards the end of Monday. > >Cheers, > >Edric. > >_______________________________________________ >mpi3-abi mailing list >mpi3-abi_at_[hidden] >http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi --------------------------------------------------------------------- Intel GmbH Dornacher Strasse 1 85622 Feldkirchen/Muenchen Germany Sitz der Gesellschaft: Feldkirchen bei Muenchen Geschaeftsfuehrer: Douglas Lusk, Peter Gleissner, Hannes Schwaderer Registergericht: Muenchen HRB 47456 Ust.-IdNr. VAT Registration No.: DE129385895 Citibank Frankfurt (BLZ 502 109 00) 600119052 This e-mail and any attachments may contain confidential material for the sole use of the intended recipient(s). Any review or distribution by others is strictly prohibited. If you are not the intended recipient, please contact the sender and delete all copies. From jeffb at [hidden] Fri Mar 7 16:04:20 2008 From: jeffb at [hidden] (Jeff Brown) Date: Fri, 07 Mar 2008 15:04:20 -0700 Subject: [Mpi3-abi] Slides for the meeting next week In-Reply-To: <5ECAB1304A8B5B4CB3F9D6C01E4E21A2011F4E0F@swsmsx413.ger.cor p.intel.com> Message-ID: <6.2.3.4.2.20080307150308.0a840028@ccs-mail.lanl.gov> At 02:17 PM 3/7/2008, you wrote: >Dear Jeff, > >Thanks. I believe we should reference Bill Gropp's paper in the last >slide (details). All material was shamelessly lifted from there by me, >which is mentioned in the draft proposal. OK >Also, to facilitate further member enrollment, it may make sense to >explicitly mention the subscription link and invite people to join the >Monday evening ABI F2F meeting. sure - I'll add that to the final slide where I list the members of the WG >Best regards. > >Alexander > >-----Original Message----- >From: mpi3-abi-bounces_at_[hidden] >[mailto:mpi3-abi-bounces_at_[hidden]] On Behalf Of Jeff Brown >Sent: Friday, March 07, 2008 5:52 PM >To: MPI 3.0 ABI working group; mpi3-abi_at_[hidden] >Subject: Re: [Mpi3-abi] Slides for the meeting next week > >here's my cut at the slides (also posted to the wiki) > >I used material from Edric's slides and added thoughts of my >own. I'll do the presentation Monday. We only have 5 minutes so >I'll focus on the key points. Feel free to clarify during the >briefing, but that's not the time for discussion. > >Jeff > >At 06:56 AM 3/7/2008, Edric Ellis wrote: > >Hi All, > > > >We need to present something to the group at the start of the meeting >on > >Monday. I hadn't seen anything about this, so I've taken the liberty of > >knocking something together from the proposal on the wiki. What do > >people think? I've also uploaded this to the wiki, so perhaps edits > >should be done there. > > > >I look forward to seeing you all at the meeting. I'm not going to be > >there right from the start (couldn't get an early enough flight), but I > >should be there towards the end of Monday. > > > >Cheers, > > > >Edric. > > > >_______________________________________________ > >mpi3-abi mailing list > >mpi3-abi_at_[hidden] > >http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi >--------------------------------------------------------------------- >Intel GmbH >Dornacher Strasse 1 >85622 Feldkirchen/Muenchen Germany >Sitz der Gesellschaft: Feldkirchen bei Muenchen >Geschaeftsfuehrer: Douglas Lusk, Peter Gleissner, Hannes Schwaderer >Registergericht: Muenchen HRB 47456 Ust.-IdNr. >VAT Registration No.: DE129385895 >Citibank Frankfurt (BLZ 502 109 00) 600119052 > >This e-mail and any attachments may contain confidential material for >the sole use of the intended recipient(s). Any review or distribution >by others is strictly prohibited. If you are not the intended >recipient, please contact the sender and delete all copies. > > >_______________________________________________ >mpi3-abi mailing list >mpi3-abi_at_[hidden] >http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi From jeffb at [hidden] Fri Mar 7 18:01:45 2008 From: jeffb at [hidden] (Jeff Brown) Date: Fri, 07 Mar 2008 17:01:45 -0700 Subject: [Mpi3-abi] Slides for the meeting next week In-Reply-To: <6.2.3.4.2.20080307081036.02b12498@ccs-mail.lanl.gov> Message-ID: <6.2.3.4.2.20080307170035.02d04e98@ccs-mail.lanl.gov> Current set of slides attached (and posted) I'll likely go with these. Jeff At 08:11 AM 3/7/2008, Jeff Brown wrote: >I was going to put some slides together today - this is a good >starting point. I'll do some edits and send out. > >Thanks for getting the ball rolling. > >Jeff > >At 06:56 AM 3/7/2008, you wrote: > >Hi All, > > > >We need to present something to the group at the start of the meeting on > >Monday. I hadn't seen anything about this, so I've taken the liberty of > >knocking something together from the proposal on the wiki. What do > >people think? I've also uploaded this to the wiki, so perhaps edits > >should be done there. > > > >I look forward to seeing you all at the meeting. I'm not going to be > >there right from the start (couldn't get an early enough flight), but I > >should be there towards the end of Monday. > > > >Cheers, > > > >Edric. > > > >_______________________________________________ > >mpi3-abi mailing list > >mpi3-abi_at_[hidden] > >http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi > > >_______________________________________________ >mpi3-abi mailing list >mpi3-abi_at_[hidden] >http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi * -------------- next part -------------- A non-text attachment was scrubbed... Name: MPI_ABI_Working_Group_status.ppt Type: application/octet-stream Size: 227328 bytes Desc: MPI_ABI_Working_Group_status.ppt URL: From Edric.Ellis at [hidden] Thu Mar 13 10:22:13 2008 From: Edric.Ellis at [hidden] (Edric Ellis) Date: Thu, 13 Mar 2008 15:22:13 -0000 Subject: [Mpi3-abi] Meeting notes from 10th March Message-ID: My notes from the ABI-WG meeting are now on the wiki at: http://svn.mpi-forum.org/trac/mpi-forum-web/wiki/ABI-WG%20Meeting%20note s%2020080310 Best regards, Edric. * -------------- next part -------------- An HTML attachment was scrubbed... URL: From jeffb at [hidden] Thu Mar 13 17:10:42 2008 From: jeffb at [hidden] (Jeff Brown) Date: Thu, 13 Mar 2008 16:10:42 -0600 Subject: [Mpi3-abi] Meeting notes from 10th March In-Reply-To: Message-ID: <6.2.3.4.2.20080313160548.02c2f358@ccs-mail.lanl.gov> I propose a way we can make progress ... Let's start populating a matrix (excel spreadsheet) with a column for each MPI implementation, and rows for the various MPI datatypes, constants, etc. where the internal implementations varys. I'll kick it off for OpenMPI and send out. The last column of the matrix can be "ABI" where we propose a common approach across the implementations. A couple of driving principles: 1. the ABI solution shouldn't negatively impact quality of implementation 2. minimize platform specific solutions I'd like to see if we can produce a single ABI that spans platforms. comments? Jeff From erezh at [hidden] Fri Mar 14 11:21:53 2008 From: erezh at [hidden] (Erez Haba) Date: Fri, 14 Mar 2008 09:21:53 -0700 Subject: [Mpi3-abi] Meeting notes from 10th March In-Reply-To: <6.2.3.4.2.20080313160548.02c2f358@ccs-mail.lanl.gov> Message-ID: <6B68D01C00C9994A8E150183E62A119E719263BC3B@NA-EXMSG-C105.redmond.corp.microsoft.com> I will upload the various mpi.h files and start the table on the ABI wiki page. -----Original Message----- From: mpi3-abi-bounces_at_[hidden] [mailto:mpi3-abi-bounces_at_[hidden]] On Behalf Of Jeff Brown Sent: Thursday, March 13, 2008 3:11 PM To: MPI 3.0 ABI working group; mpi3-abi_at_[hidden] Subject: Re: [Mpi3-abi] Meeting notes from 10th March I propose a way we can make progress ... Let's start populating a matrix (excel spreadsheet) with a column for each MPI implementation, and rows for the various MPI datatypes, constants, etc. where the internal implementations varys. I'll kick it off for OpenMPI and send out. The last column of the matrix can be "ABI" where we propose a common approach across the implementations. A couple of driving principles: 1. the ABI solution shouldn't negatively impact quality of implementation 2. minimize platform specific solutions I'd like to see if we can produce a single ABI that spans platforms. comments? Jeff _______________________________________________ mpi3-abi mailing list mpi3-abi_at_[hidden] http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi From jsquyres at [hidden] Fri Mar 14 11:35:59 2008 From: jsquyres at [hidden] (Jeff Squyres) Date: Fri, 14 Mar 2008 12:35:59 -0400 Subject: [Mpi3-abi] Meeting notes from 10th March In-Reply-To: <6B68D01C00C9994A8E150183E62A119E719263BC3B@NA-EXMSG-C105.redmond.corp.microsoft.com> Message-ID: <89726CBA-0161-4908-A927-4D8CC1926173@cisco.com> Perhaps you could just link to OMPI's header file -- it's a live copy of what's in our SVN repository: https://svn.open-mpi.org/trac/ompi/browser/trunk/ompi/include/mpi.h.in On Mar 14, 2008, at 12:21 PM, Erez Haba wrote: > I will upload the various mpi.h files and start the table on the ABI > wiki page. > > -----Original Message----- > From: mpi3-abi-bounces_at_[hidden] [mailto:mpi3-abi-bounces_at_[hidden] > ] On Behalf Of Jeff Brown > Sent: Thursday, March 13, 2008 3:11 PM > To: MPI 3.0 ABI working group; mpi3-abi_at_[hidden] > Subject: Re: [Mpi3-abi] Meeting notes from 10th March > > I propose a way we can make progress ... > > Let's start populating a matrix (excel spreadsheet) with a column for > each MPI implementation, and rows for the various MPI datatypes, > constants, etc. where the internal implementations varys. I'll kick > it off for OpenMPI and send out. > > The last column of the matrix can be "ABI" where we propose a common > approach across the implementations. > > A couple of driving principles: > 1. the ABI solution shouldn't negatively impact quality of > implementation > 2. minimize platform specific solutions > > I'd like to see if we can produce a single ABI that spans platforms. > > comments? > > Jeff > > > _______________________________________________ > mpi3-abi mailing list > mpi3-abi_at_[hidden] > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi > > _______________________________________________ > mpi3-abi mailing list > mpi3-abi_at_[hidden] > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi -- Jeff Squyres Cisco Systems From alexander.supalov at [hidden] Fri Mar 14 11:44:51 2008 From: alexander.supalov at [hidden] (Supalov, Alexander) Date: Fri, 14 Mar 2008 16:44:51 -0000 Subject: [Mpi3-abi] Meeting notes from 10th March In-Reply-To: <89726CBA-0161-4908-A927-4D8CC1926173@cisco.com> Message-ID: <5ECAB1304A8B5B4CB3F9D6C01E4E21A201250ED9@swsmsx413.ger.corp.intel.com> Hi, I'd propose to have hard copies in place for all implementations involved, and complement them by links if necessary. This would eliminate quiet header drift that certain implementations might experience once in a while. Best regards. Alexander -----Original Message----- From: mpi3-abi-bounces_at_[hidden] [mailto:mpi3-abi-bounces_at_[hidden]] On Behalf Of Jeff Squyres Sent: Friday, March 14, 2008 5:36 PM To: MPI 3.0 ABI working group Subject: Re: [Mpi3-abi] Meeting notes from 10th March Perhaps you could just link to OMPI's header file -- it's a live copy of what's in our SVN repository: https://svn.open-mpi.org/trac/ompi/browser/trunk/ompi/include/mpi.h.in On Mar 14, 2008, at 12:21 PM, Erez Haba wrote: > I will upload the various mpi.h files and start the table on the ABI > wiki page. > > -----Original Message----- > From: mpi3-abi-bounces_at_[hidden] [mailto:mpi3-abi-bounces_at_[hidden] > ] On Behalf Of Jeff Brown > Sent: Thursday, March 13, 2008 3:11 PM > To: MPI 3.0 ABI working group; mpi3-abi_at_[hidden] > Subject: Re: [Mpi3-abi] Meeting notes from 10th March > > I propose a way we can make progress ... > > Let's start populating a matrix (excel spreadsheet) with a column for > each MPI implementation, and rows for the various MPI datatypes, > constants, etc. where the internal implementations varys. I'll kick > it off for OpenMPI and send out. > > The last column of the matrix can be "ABI" where we propose a common > approach across the implementations. > > A couple of driving principles: > 1. the ABI solution shouldn't negatively impact quality of > implementation > 2. minimize platform specific solutions > > I'd like to see if we can produce a single ABI that spans platforms. > > comments? > > Jeff > > > _______________________________________________ > mpi3-abi mailing list > mpi3-abi_at_[hidden] > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi > > _______________________________________________ > mpi3-abi mailing list > mpi3-abi_at_[hidden] > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi -- Jeff Squyres Cisco Systems _______________________________________________ mpi3-abi mailing list mpi3-abi_at_[hidden] http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi --------------------------------------------------------------------- Intel GmbH Dornacher Strasse 1 85622 Feldkirchen/Muenchen Germany Sitz der Gesellschaft: Feldkirchen bei Muenchen Geschaeftsfuehrer: Douglas Lusk, Peter Gleissner, Hannes Schwaderer Registergericht: Muenchen HRB 47456 Ust.-IdNr. VAT Registration No.: DE129385895 Citibank Frankfurt (BLZ 502 109 00) 600119052 This e-mail and any attachments may contain confidential material for the sole use of the intended recipient(s). Any review or distribution by others is strictly prohibited. If you are not the intended recipient, please contact the sender and delete all copies. From erezh at [hidden] Fri Mar 14 11:51:40 2008 From: erezh at [hidden] (Erez Haba) Date: Fri, 14 Mar 2008 09:51:40 -0700 Subject: [Mpi3-abi] Meeting notes from 10th March In-Reply-To: <5ECAB1304A8B5B4CB3F9D6C01E4E21A201250ED9@swsmsx413.ger.corp.intel.com> Message-ID: <6B68D01C00C9994A8E150183E62A119E719263BC71@NA-EXMSG-C105.redmond.corp.microsoft.com> Yes, that was my intention. :) -----Original Message----- From: mpi3-abi-bounces_at_[hidden] [mailto:mpi3-abi-bounces_at_[hidden]] On Behalf Of Supalov, Alexander Sent: Friday, March 14, 2008 9:45 AM To: MPI 3.0 ABI working group Subject: Re: [Mpi3-abi] Meeting notes from 10th March Hi, I'd propose to have hard copies in place for all implementations involved, and complement them by links if necessary. This would eliminate quiet header drift that certain implementations might experience once in a while. Best regards. Alexander -----Original Message----- From: mpi3-abi-bounces_at_[hidden] [mailto:mpi3-abi-bounces_at_[hidden]] On Behalf Of Jeff Squyres Sent: Friday, March 14, 2008 5:36 PM To: MPI 3.0 ABI working group Subject: Re: [Mpi3-abi] Meeting notes from 10th March Perhaps you could just link to OMPI's header file -- it's a live copy of what's in our SVN repository: https://svn.open-mpi.org/trac/ompi/browser/trunk/ompi/include/mpi.h.in On Mar 14, 2008, at 12:21 PM, Erez Haba wrote: > I will upload the various mpi.h files and start the table on the ABI > wiki page. > > -----Original Message----- > From: mpi3-abi-bounces_at_[hidden] [mailto:mpi3-abi-bounces_at_[hidden] > ] On Behalf Of Jeff Brown > Sent: Thursday, March 13, 2008 3:11 PM > To: MPI 3.0 ABI working group; mpi3-abi_at_[hidden] > Subject: Re: [Mpi3-abi] Meeting notes from 10th March > > I propose a way we can make progress ... > > Let's start populating a matrix (excel spreadsheet) with a column for > each MPI implementation, and rows for the various MPI datatypes, > constants, etc. where the internal implementations varys. I'll kick > it off for OpenMPI and send out. > > The last column of the matrix can be "ABI" where we propose a common > approach across the implementations. > > A couple of driving principles: > 1. the ABI solution shouldn't negatively impact quality of > implementation > 2. minimize platform specific solutions > > I'd like to see if we can produce a single ABI that spans platforms. > > comments? > > Jeff > > > _______________________________________________ > mpi3-abi mailing list > mpi3-abi_at_[hidden] > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi > > _______________________________________________ > mpi3-abi mailing list > mpi3-abi_at_[hidden] > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi -- Jeff Squyres Cisco Systems _______________________________________________ mpi3-abi mailing list mpi3-abi_at_[hidden] http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi --------------------------------------------------------------------- Intel GmbH Dornacher Strasse 1 85622 Feldkirchen/Muenchen Germany Sitz der Gesellschaft: Feldkirchen bei Muenchen Geschaeftsfuehrer: Douglas Lusk, Peter Gleissner, Hannes Schwaderer Registergericht: Muenchen HRB 47456 Ust.-IdNr. VAT Registration No.: DE129385895 Citibank Frankfurt (BLZ 502 109 00) 600119052 This e-mail and any attachments may contain confidential material for the sole use of the intended recipient(s). Any review or distribution by others is strictly prohibited. If you are not the intended recipient, please contact the sender and delete all copies. _______________________________________________ mpi3-abi mailing list mpi3-abi_at_[hidden] http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi From david.solt at [hidden] Fri Mar 14 13:08:17 2008 From: david.solt at [hidden] (Solt, David George) Date: Fri, 14 Mar 2008 18:08:17 +0000 Subject: [Mpi3-abi] Meeting notes from 10th March In-Reply-To: <89726CBA-0161-4908-A927-4D8CC1926173@cisco.com> Message-ID: I have to wonder if MPI_BOTTOM will cause us headaches. MPI_BOTTOM is meant to be used directly as an address. Our ABI would really have to specify exactly what that address is using either a constant or the address of a variable in the library. We have found that many non-compliant codes exist which assume that MPI_BOTTOM is zero. Assuming MPI_BOTTOM lives on to MPI-3, the ABI effort may want to dictate its value on a per-target architecture/OS basis, with ((void*) 0) the most obvious setting for nearly all targets. Dave -----Original Message----- From: mpi3-abi-bounces_at_[hidden] [mailto:mpi3-abi-bounces_at_[hidden]] On Behalf Of Jeff Squyres Sent: Friday, March 14, 2008 11:36 AM To: MPI 3.0 ABI working group Subject: Re: [Mpi3-abi] Meeting notes from 10th March Perhaps you could just link to OMPI's header file -- it's a live copy of what's in our SVN repository: https://svn.open-mpi.org/trac/ompi/browser/trunk/ompi/include/mpi.h.in On Mar 14, 2008, at 12:21 PM, Erez Haba wrote: > I will upload the various mpi.h files and start the table on the ABI > wiki page. > > -----Original Message----- > From: mpi3-abi-bounces_at_[hidden] > [mailto:mpi3-abi-bounces_at_[hidden] > ] On Behalf Of Jeff Brown > Sent: Thursday, March 13, 2008 3:11 PM > To: MPI 3.0 ABI working group; mpi3-abi_at_[hidden] > Subject: Re: [Mpi3-abi] Meeting notes from 10th March > > I propose a way we can make progress ... > > Let's start populating a matrix (excel spreadsheet) with a column for > each MPI implementation, and rows for the various MPI datatypes, > constants, etc. where the internal implementations varys. I'll kick > it off for OpenMPI and send out. > > The last column of the matrix can be "ABI" where we propose a common > approach across the implementations. > > A couple of driving principles: > 1. the ABI solution shouldn't negatively impact quality of > implementation 2. minimize platform specific solutions > > I'd like to see if we can produce a single ABI that spans platforms. > > comments? > > Jeff > > > _______________________________________________ > mpi3-abi mailing list > mpi3-abi_at_[hidden] > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi > > _______________________________________________ > mpi3-abi mailing list > mpi3-abi_at_[hidden] > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi -- Jeff Squyres Cisco Systems _______________________________________________ mpi3-abi mailing list mpi3-abi_at_[hidden] http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi From erezh at [hidden] Fri Mar 14 13:46:44 2008 From: erezh at [hidden] (Erez Haba) Date: Fri, 14 Mar 2008 11:46:44 -0700 Subject: [Mpi3-abi] Meeting notes from 10th March In-Reply-To: Message-ID: <6B68D01C00C9994A8E150183E62A119E719263BDC6@NA-EXMSG-C105.redmond.corp.microsoft.com> I have posted the mpi.h files to the wiki page and started a table comparing the various implementations. See http://svn.mpi-forum.org/trac/mpi-forum-web/wiki/Compare_mpi_h Thanks, .Erez -----Original Message----- From: mpi3-abi-bounces_at_[hidden] [mailto:mpi3-abi-bounces_at_[hidden]] On Behalf Of Solt, David George Sent: Friday, March 14, 2008 11:08 AM To: MPI 3.0 ABI working group Subject: Re: [Mpi3-abi] Meeting notes from 10th March I have to wonder if MPI_BOTTOM will cause us headaches. MPI_BOTTOM is meant to be used directly as an address. Our ABI would really have to specify exactly what that address is using either a constant or the address of a variable in the library. We have found that many non-compliant codes exist which assume that MPI_BOTTOM is zero. Assuming MPI_BOTTOM lives on to MPI-3, the ABI effort may want to dictate its value on a per-target architecture/OS basis, with ((void*) 0) the most obvious setting for nearly all targets. Dave -----Original Message----- From: mpi3-abi-bounces_at_[hidden] [mailto:mpi3-abi-bounces_at_[hidden]] On Behalf Of Jeff Squyres Sent: Friday, March 14, 2008 11:36 AM To: MPI 3.0 ABI working group Subject: Re: [Mpi3-abi] Meeting notes from 10th March Perhaps you could just link to OMPI's header file -- it's a live copy of what's in our SVN repository: https://svn.open-mpi.org/trac/ompi/browser/trunk/ompi/include/mpi.h.in On Mar 14, 2008, at 12:21 PM, Erez Haba wrote: > I will upload the various mpi.h files and start the table on the ABI > wiki page. > > -----Original Message----- > From: mpi3-abi-bounces_at_[hidden] > [mailto:mpi3-abi-bounces_at_[hidden] > ] On Behalf Of Jeff Brown > Sent: Thursday, March 13, 2008 3:11 PM > To: MPI 3.0 ABI working group; mpi3-abi_at_[hidden] > Subject: Re: [Mpi3-abi] Meeting notes from 10th March > > I propose a way we can make progress ... > > Let's start populating a matrix (excel spreadsheet) with a column for > each MPI implementation, and rows for the various MPI datatypes, > constants, etc. where the internal implementations varys. I'll kick > it off for OpenMPI and send out. > > The last column of the matrix can be "ABI" where we propose a common > approach across the implementations. > > A couple of driving principles: > 1. the ABI solution shouldn't negatively impact quality of > implementation 2. minimize platform specific solutions > > I'd like to see if we can produce a single ABI that spans platforms. > > comments? > > Jeff > > > _______________________________________________ > mpi3-abi mailing list > mpi3-abi_at_[hidden] > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi > > _______________________________________________ > mpi3-abi mailing list > mpi3-abi_at_[hidden] > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi -- Jeff Squyres Cisco Systems _______________________________________________ mpi3-abi mailing list mpi3-abi_at_[hidden] http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi _______________________________________________ mpi3-abi mailing list mpi3-abi_at_[hidden] http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi From alexander.supalov at [hidden] Fri Mar 14 18:45:22 2008 From: alexander.supalov at [hidden] (Supalov, Alexander) Date: Fri, 14 Mar 2008 23:45:22 -0000 Subject: [Mpi3-abi] Meeting notes from 10th March In-Reply-To: <6B68D01C00C9994A8E150183E62A119E719263BDC6@NA-EXMSG-C105.redmond.corp.microsoft.com> Message-ID: <5ECAB1304A8B5B4CB3F9D6C01E4E21A201250F7D@swsmsx413.ger.corp.intel.com> Dear Erez, Thank you. I'm afraid that the MPICH2 mpi.h has not been configured - it carries configure macros in the datatypes definition part and elsewhere. By the way, I'm going to upload MPICH2 1.0.3 mpi.h early next week as a placeholder for Intel MPI mpi.h, pending legal OK for the latter. I'll need to instantiate this file. According to my experience, the result may depend on the platform and compiler used. What platform/compiler should we use for reference? At least, what word size should we target? I seek to eliminate an accidental apple to pears comparison here. Best regards. Alexander -----Original Message----- From: mpi3-abi-bounces_at_[hidden] [mailto:mpi3-abi-bounces_at_[hidden]] On Behalf Of Erez Haba Sent: Friday, March 14, 2008 7:47 PM To: MPI 3.0 ABI working group Subject: Re: [Mpi3-abi] Meeting notes from 10th March I have posted the mpi.h files to the wiki page and started a table comparing the various implementations. See http://svn.mpi-forum.org/trac/mpi-forum-web/wiki/Compare_mpi_h Thanks, .Erez -----Original Message----- From: mpi3-abi-bounces_at_[hidden] [mailto:mpi3-abi-bounces_at_[hidden]] On Behalf Of Solt, David George Sent: Friday, March 14, 2008 11:08 AM To: MPI 3.0 ABI working group Subject: Re: [Mpi3-abi] Meeting notes from 10th March I have to wonder if MPI_BOTTOM will cause us headaches. MPI_BOTTOM is meant to be used directly as an address. Our ABI would really have to specify exactly what that address is using either a constant or the address of a variable in the library. We have found that many non-compliant codes exist which assume that MPI_BOTTOM is zero. Assuming MPI_BOTTOM lives on to MPI-3, the ABI effort may want to dictate its value on a per-target architecture/OS basis, with ((void*) 0) the most obvious setting for nearly all targets. Dave -----Original Message----- From: mpi3-abi-bounces_at_[hidden] [mailto:mpi3-abi-bounces_at_[hidden]] On Behalf Of Jeff Squyres Sent: Friday, March 14, 2008 11:36 AM To: MPI 3.0 ABI working group Subject: Re: [Mpi3-abi] Meeting notes from 10th March Perhaps you could just link to OMPI's header file -- it's a live copy of what's in our SVN repository: https://svn.open-mpi.org/trac/ompi/browser/trunk/ompi/include/mpi.h.in On Mar 14, 2008, at 12:21 PM, Erez Haba wrote: > I will upload the various mpi.h files and start the table on the ABI > wiki page. > > -----Original Message----- > From: mpi3-abi-bounces_at_[hidden] > [mailto:mpi3-abi-bounces_at_[hidden] > ] On Behalf Of Jeff Brown > Sent: Thursday, March 13, 2008 3:11 PM > To: MPI 3.0 ABI working group; mpi3-abi_at_[hidden] > Subject: Re: [Mpi3-abi] Meeting notes from 10th March > > I propose a way we can make progress ... > > Let's start populating a matrix (excel spreadsheet) with a column for > each MPI implementation, and rows for the various MPI datatypes, > constants, etc. where the internal implementations varys. I'll kick > it off for OpenMPI and send out. > > The last column of the matrix can be "ABI" where we propose a common > approach across the implementations. > > A couple of driving principles: > 1. the ABI solution shouldn't negatively impact quality of > implementation 2. minimize platform specific solutions > > I'd like to see if we can produce a single ABI that spans platforms. > > comments? > > Jeff > > > _______________________________________________ > mpi3-abi mailing list > mpi3-abi_at_[hidden] > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi > > _______________________________________________ > mpi3-abi mailing list > mpi3-abi_at_[hidden] > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi -- Jeff Squyres Cisco Systems _______________________________________________ mpi3-abi mailing list mpi3-abi_at_[hidden] http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi _______________________________________________ mpi3-abi mailing list mpi3-abi_at_[hidden] http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi _______________________________________________ mpi3-abi mailing list mpi3-abi_at_[hidden] http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi --------------------------------------------------------------------- Intel GmbH Dornacher Strasse 1 85622 Feldkirchen/Muenchen Germany Sitz der Gesellschaft: Feldkirchen bei Muenchen Geschaeftsfuehrer: Douglas Lusk, Peter Gleissner, Hannes Schwaderer Registergericht: Muenchen HRB 47456 Ust.-IdNr. VAT Registration No.: DE129385895 Citibank Frankfurt (BLZ 502 109 00) 600119052 This e-mail and any attachments may contain confidential material for the sole use of the intended recipient(s). Any review or distribution by others is strictly prohibited. If you are not the intended recipient, please contact the sender and delete all copies. From jsquyres at [hidden] Fri Mar 14 18:52:05 2008 From: jsquyres at [hidden] (Jeff Squyres) Date: Fri, 14 Mar 2008 19:52:05 -0400 Subject: [Mpi3-abi] Meeting notes from 10th March In-Reply-To: <5ECAB1304A8B5B4CB3F9D6C01E4E21A201250F7D@swsmsx413.ger.corp.intel.com> Message-ID: <811704E3-F16B-4B78-BA6C-78CA08502B14@cisco.com> FWIW: the same is true for the Open MPI header file (contents depend on platform / compiler / etc.). On Mar 14, 2008, at 7:45 PM, Supalov, Alexander wrote: > Dear Erez, > > Thank you. I'm afraid that the MPICH2 mpi.h has not been configured > - it > carries configure macros in the datatypes definition part and > elsewhere. > By the way, I'm going to upload MPICH2 1.0.3 mpi.h early next week > as a > placeholder for Intel MPI mpi.h, pending legal OK for the latter. > > I'll need to instantiate this file. According to my experience, the > result may depend on the platform and compiler used. What > platform/compiler should we use for reference? At least, what word > size > should we target? I seek to eliminate an accidental apple to pears > comparison here. > > Best regards. > > Alexander > > -----Original Message----- > From: mpi3-abi-bounces_at_[hidden] > [mailto:mpi3-abi-bounces_at_[hidden]] On Behalf Of Erez Haba > Sent: Friday, March 14, 2008 7:47 PM > To: MPI 3.0 ABI working group > Subject: Re: [Mpi3-abi] Meeting notes from 10th March > > I have posted the mpi.h files to the wiki page and started a table > comparing the various implementations. > > See http://svn.mpi-forum.org/trac/mpi-forum-web/wiki/Compare_mpi_h > > Thanks, > .Erez > > -----Original Message----- > From: mpi3-abi-bounces_at_[hidden] > [mailto:mpi3-abi-bounces_at_[hidden]] On Behalf Of Solt, David > George > Sent: Friday, March 14, 2008 11:08 AM > To: MPI 3.0 ABI working group > Subject: Re: [Mpi3-abi] Meeting notes from 10th March > > I have to wonder if MPI_BOTTOM will cause us headaches. MPI_BOTTOM is > meant to be used directly as an address. Our ABI would really have to > specify exactly what that address is using either a constant or the > address of a variable in the library. We have found that many > non-compliant codes exist which assume that MPI_BOTTOM is zero. > Assuming MPI_BOTTOM lives on to MPI-3, the ABI effort may want to > dictate its value on a per-target architecture/OS basis, with ((void*) > 0) the most obvious setting for nearly all targets. > > Dave > > > -----Original Message----- > From: mpi3-abi-bounces_at_[hidden] > [mailto:mpi3-abi-bounces_at_[hidden]] On Behalf Of Jeff > Squyres > Sent: Friday, March 14, 2008 11:36 AM > To: MPI 3.0 ABI working group > Subject: Re: [Mpi3-abi] Meeting notes from 10th March > > Perhaps you could just link to OMPI's header file -- it's a live > copy of > what's in our SVN repository: > > https://svn.open-mpi.org/trac/ompi/browser/trunk/ompi/include/mpi.h.in > > > On Mar 14, 2008, at 12:21 PM, Erez Haba wrote: > >> I will upload the various mpi.h files and start the table on the ABI >> wiki page. >> >> -----Original Message----- >> From: mpi3-abi-bounces_at_[hidden] >> [mailto:mpi3-abi-bounces_at_[hidden] >> ] On Behalf Of Jeff Brown >> Sent: Thursday, March 13, 2008 3:11 PM >> To: MPI 3.0 ABI working group; mpi3-abi_at_[hidden] >> Subject: Re: [Mpi3-abi] Meeting notes from 10th March >> >> I propose a way we can make progress ... >> >> Let's start populating a matrix (excel spreadsheet) with a column for >> each MPI implementation, and rows for the various MPI datatypes, >> constants, etc. where the internal implementations varys. I'll kick >> it off for OpenMPI and send out. >> >> The last column of the matrix can be "ABI" where we propose a common >> approach across the implementations. >> >> A couple of driving principles: >> 1. the ABI solution shouldn't negatively impact quality of >> implementation 2. minimize platform specific solutions >> >> I'd like to see if we can produce a single ABI that spans platforms. >> >> comments? >> >> Jeff >> >> >> _______________________________________________ >> mpi3-abi mailing list >> mpi3-abi_at_[hidden] >> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi >> >> _______________________________________________ >> mpi3-abi mailing list >> mpi3-abi_at_[hidden] >> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi > > > -- > Jeff Squyres > Cisco Systems > > _______________________________________________ > mpi3-abi mailing list > mpi3-abi_at_[hidden] > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi > > _______________________________________________ > mpi3-abi mailing list > mpi3-abi_at_[hidden] > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi > > _______________________________________________ > mpi3-abi mailing list > mpi3-abi_at_[hidden] > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi > --------------------------------------------------------------------- > Intel GmbH > Dornacher Strasse 1 > 85622 Feldkirchen/Muenchen Germany > Sitz der Gesellschaft: Feldkirchen bei Muenchen > Geschaeftsfuehrer: Douglas Lusk, Peter Gleissner, Hannes Schwaderer > Registergericht: Muenchen HRB 47456 Ust.-IdNr. > VAT Registration No.: DE129385895 > Citibank Frankfurt (BLZ 502 109 00) 600119052 > > This e-mail and any attachments may contain confidential material for > the sole use of the intended recipient(s). Any review or distribution > by others is strictly prohibited. If you are not the intended > recipient, please contact the sender and delete all copies. > > > _______________________________________________ > mpi3-abi mailing list > mpi3-abi_at_[hidden] > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi -- Jeff Squyres Cisco Systems From erezh at [hidden] Fri Mar 14 18:59:41 2008 From: erezh at [hidden] (Erez Haba) Date: Fri, 14 Mar 2008 16:59:41 -0700 Subject: [Mpi3-abi] Meeting notes from 10th March In-Reply-To: <5ECAB1304A8B5B4CB3F9D6C01E4E21A201250F7D@swsmsx413.ger.corp.intel.com> Message-ID: <6B68D01C00C9994A8E150183E62A119E719263BFD5@NA-EXMSG-C105.redmond.corp.microsoft.com> Thank you Alexander, I did not configure mpich2.mpi.h on purpose. The reason was to help us understand that mpich2 ABI varies base on configuration. For msmpi we took the approach to include the 64 vs 32 bit ifdef in the file rather than require configuring for the word width. I think that we'll need to understand the various configure issue. I'd like us to focus on 64 bit as it reveals more issue than 32 bit. (e.g., sizeof(void*) size is different than sizeof(int) in most cases) Thanks, .Erez -----Original Message----- From: mpi3-abi-bounces_at_[hidden] [mailto:mpi3-abi-bounces_at_[hidden]] On Behalf Of Supalov, Alexander Sent: Friday, March 14, 2008 4:45 PM To: MPI 3.0 ABI working group Subject: Re: [Mpi3-abi] Meeting notes from 10th March Dear Erez, Thank you. I'm afraid that the MPICH2 mpi.h has not been configured - it carries configure macros in the datatypes definition part and elsewhere. By the way, I'm going to upload MPICH2 1.0.3 mpi.h early next week as a placeholder for Intel MPI mpi.h, pending legal OK for the latter. I'll need to instantiate this file. According to my experience, the result may depend on the platform and compiler used. What platform/compiler should we use for reference? At least, what word size should we target? I seek to eliminate an accidental apple to pears comparison here. Best regards. Alexander -----Original Message----- From: mpi3-abi-bounces_at_[hidden] [mailto:mpi3-abi-bounces_at_[hidden]] On Behalf Of Erez Haba Sent: Friday, March 14, 2008 7:47 PM To: MPI 3.0 ABI working group Subject: Re: [Mpi3-abi] Meeting notes from 10th March I have posted the mpi.h files to the wiki page and started a table comparing the various implementations. See http://svn.mpi-forum.org/trac/mpi-forum-web/wiki/Compare_mpi_h Thanks, .Erez -----Original Message----- From: mpi3-abi-bounces_at_[hidden] [mailto:mpi3-abi-bounces_at_[hidden]] On Behalf Of Solt, David George Sent: Friday, March 14, 2008 11:08 AM To: MPI 3.0 ABI working group Subject: Re: [Mpi3-abi] Meeting notes from 10th March I have to wonder if MPI_BOTTOM will cause us headaches. MPI_BOTTOM is meant to be used directly as an address. Our ABI would really have to specify exactly what that address is using either a constant or the address of a variable in the library. We have found that many non-compliant codes exist which assume that MPI_BOTTOM is zero. Assuming MPI_BOTTOM lives on to MPI-3, the ABI effort may want to dictate its value on a per-target architecture/OS basis, with ((void*) 0) the most obvious setting for nearly all targets. Dave -----Original Message----- From: mpi3-abi-bounces_at_[hidden] [mailto:mpi3-abi-bounces_at_[hidden]] On Behalf Of Jeff Squyres Sent: Friday, March 14, 2008 11:36 AM To: MPI 3.0 ABI working group Subject: Re: [Mpi3-abi] Meeting notes from 10th March Perhaps you could just link to OMPI's header file -- it's a live copy of what's in our SVN repository: https://svn.open-mpi.org/trac/ompi/browser/trunk/ompi/include/mpi.h.in On Mar 14, 2008, at 12:21 PM, Erez Haba wrote: > I will upload the various mpi.h files and start the table on the ABI > wiki page. > > -----Original Message----- > From: mpi3-abi-bounces_at_[hidden] > [mailto:mpi3-abi-bounces_at_[hidden] > ] On Behalf Of Jeff Brown > Sent: Thursday, March 13, 2008 3:11 PM > To: MPI 3.0 ABI working group; mpi3-abi_at_[hidden] > Subject: Re: [Mpi3-abi] Meeting notes from 10th March > > I propose a way we can make progress ... > > Let's start populating a matrix (excel spreadsheet) with a column for > each MPI implementation, and rows for the various MPI datatypes, > constants, etc. where the internal implementations varys. I'll kick > it off for OpenMPI and send out. > > The last column of the matrix can be "ABI" where we propose a common > approach across the implementations. > > A couple of driving principles: > 1. the ABI solution shouldn't negatively impact quality of > implementation 2. minimize platform specific solutions > > I'd like to see if we can produce a single ABI that spans platforms. > > comments? > > Jeff > > > _______________________________________________ > mpi3-abi mailing list > mpi3-abi_at_[hidden] > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi > > _______________________________________________ > mpi3-abi mailing list > mpi3-abi_at_[hidden] > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi -- Jeff Squyres Cisco Systems _______________________________________________ mpi3-abi mailing list mpi3-abi_at_[hidden] http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi _______________________________________________ mpi3-abi mailing list mpi3-abi_at_[hidden] http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi _______________________________________________ mpi3-abi mailing list mpi3-abi_at_[hidden] http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi --------------------------------------------------------------------- Intel GmbH Dornacher Strasse 1 85622 Feldkirchen/Muenchen Germany Sitz der Gesellschaft: Feldkirchen bei Muenchen Geschaeftsfuehrer: Douglas Lusk, Peter Gleissner, Hannes Schwaderer Registergericht: Muenchen HRB 47456 Ust.-IdNr. VAT Registration No.: DE129385895 Citibank Frankfurt (BLZ 502 109 00) 600119052 This e-mail and any attachments may contain confidential material for the sole use of the intended recipient(s). Any review or distribution by others is strictly prohibited. If you are not the intended recipient, please contact the sender and delete all copies. _______________________________________________ mpi3-abi mailing list mpi3-abi_at_[hidden] http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi From alexander.supalov at [hidden] Fri Mar 14 19:28:43 2008 From: alexander.supalov at [hidden] (Supalov, Alexander) Date: Sat, 15 Mar 2008 00:28:43 -0000 Subject: [Mpi3-abi] Meeting notes from 10th March In-Reply-To: <6B68D01C00C9994A8E150183E62A119E719263BFD5@NA-EXMSG-C105.redmond.corp.microsoft.com> Message-ID: <5ECAB1304A8B5B4CB3F9D6C01E4E21A201250F88@swsmsx413.ger.corp.intel.com> Thanks. I think to have a complete picture we need to cover both unconfigured mpi.h.in files (if applicable), as well as instantiated 32- and 64-bit mpi.h files. -----Original Message----- From: mpi3-abi-bounces_at_[hidden] [mailto:mpi3-abi-bounces_at_[hidden]] On Behalf Of Erez Haba Sent: Saturday, March 15, 2008 1:00 AM To: MPI 3.0 ABI working group Subject: Re: [Mpi3-abi] Meeting notes from 10th March Thank you Alexander, I did not configure mpich2.mpi.h on purpose. The reason was to help us understand that mpich2 ABI varies base on configuration. For msmpi we took the approach to include the 64 vs 32 bit ifdef in the file rather than require configuring for the word width. I think that we'll need to understand the various configure issue. I'd like us to focus on 64 bit as it reveals more issue than 32 bit. (e.g., sizeof(void*) size is different than sizeof(int) in most cases) Thanks, .Erez -----Original Message----- From: mpi3-abi-bounces_at_[hidden] [mailto:mpi3-abi-bounces_at_[hidden]] On Behalf Of Supalov, Alexander Sent: Friday, March 14, 2008 4:45 PM To: MPI 3.0 ABI working group Subject: Re: [Mpi3-abi] Meeting notes from 10th March Dear Erez, Thank you. I'm afraid that the MPICH2 mpi.h has not been configured - it carries configure macros in the datatypes definition part and elsewhere. By the way, I'm going to upload MPICH2 1.0.3 mpi.h early next week as a placeholder for Intel MPI mpi.h, pending legal OK for the latter. I'll need to instantiate this file. According to my experience, the result may depend on the platform and compiler used. What platform/compiler should we use for reference? At least, what word size should we target? I seek to eliminate an accidental apple to pears comparison here. Best regards. Alexander -----Original Message----- From: mpi3-abi-bounces_at_[hidden] [mailto:mpi3-abi-bounces_at_[hidden]] On Behalf Of Erez Haba Sent: Friday, March 14, 2008 7:47 PM To: MPI 3.0 ABI working group Subject: Re: [Mpi3-abi] Meeting notes from 10th March I have posted the mpi.h files to the wiki page and started a table comparing the various implementations. See http://svn.mpi-forum.org/trac/mpi-forum-web/wiki/Compare_mpi_h Thanks, .Erez -----Original Message----- From: mpi3-abi-bounces_at_[hidden] [mailto:mpi3-abi-bounces_at_[hidden]] On Behalf Of Solt, David George Sent: Friday, March 14, 2008 11:08 AM To: MPI 3.0 ABI working group Subject: Re: [Mpi3-abi] Meeting notes from 10th March I have to wonder if MPI_BOTTOM will cause us headaches. MPI_BOTTOM is meant to be used directly as an address. Our ABI would really have to specify exactly what that address is using either a constant or the address of a variable in the library. We have found that many non-compliant codes exist which assume that MPI_BOTTOM is zero. Assuming MPI_BOTTOM lives on to MPI-3, the ABI effort may want to dictate its value on a per-target architecture/OS basis, with ((void*) 0) the most obvious setting for nearly all targets. Dave -----Original Message----- From: mpi3-abi-bounces_at_[hidden] [mailto:mpi3-abi-bounces_at_[hidden]] On Behalf Of Jeff Squyres Sent: Friday, March 14, 2008 11:36 AM To: MPI 3.0 ABI working group Subject: Re: [Mpi3-abi] Meeting notes from 10th March Perhaps you could just link to OMPI's header file -- it's a live copy of what's in our SVN repository: https://svn.open-mpi.org/trac/ompi/browser/trunk/ompi/include/mpi.h.in On Mar 14, 2008, at 12:21 PM, Erez Haba wrote: > I will upload the various mpi.h files and start the table on the ABI > wiki page. > > -----Original Message----- > From: mpi3-abi-bounces_at_[hidden] > [mailto:mpi3-abi-bounces_at_[hidden] > ] On Behalf Of Jeff Brown > Sent: Thursday, March 13, 2008 3:11 PM > To: MPI 3.0 ABI working group; mpi3-abi_at_[hidden] > Subject: Re: [Mpi3-abi] Meeting notes from 10th March > > I propose a way we can make progress ... > > Let's start populating a matrix (excel spreadsheet) with a column for > each MPI implementation, and rows for the various MPI datatypes, > constants, etc. where the internal implementations varys. I'll kick > it off for OpenMPI and send out. > > The last column of the matrix can be "ABI" where we propose a common > approach across the implementations. > > A couple of driving principles: > 1. the ABI solution shouldn't negatively impact quality of > implementation 2. minimize platform specific solutions > > I'd like to see if we can produce a single ABI that spans platforms. > > comments? > > Jeff > > > _______________________________________________ > mpi3-abi mailing list > mpi3-abi_at_[hidden] > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi > > _______________________________________________ > mpi3-abi mailing list > mpi3-abi_at_[hidden] > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi -- Jeff Squyres Cisco Systems _______________________________________________ mpi3-abi mailing list mpi3-abi_at_[hidden] http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi _______________________________________________ mpi3-abi mailing list mpi3-abi_at_[hidden] http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi _______________________________________________ mpi3-abi mailing list mpi3-abi_at_[hidden] http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi --------------------------------------------------------------------- Intel GmbH Dornacher Strasse 1 85622 Feldkirchen/Muenchen Germany Sitz der Gesellschaft: Feldkirchen bei Muenchen Geschaeftsfuehrer: Douglas Lusk, Peter Gleissner, Hannes Schwaderer Registergericht: Muenchen HRB 47456 Ust.-IdNr. VAT Registration No.: DE129385895 Citibank Frankfurt (BLZ 502 109 00) 600119052 This e-mail and any attachments may contain confidential material for the sole use of the intended recipient(s). Any review or distribution by others is strictly prohibited. If you are not the intended recipient, please contact the sender and delete all copies. _______________________________________________ mpi3-abi mailing list mpi3-abi_at_[hidden] http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi _______________________________________________ mpi3-abi mailing list mpi3-abi_at_[hidden] http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi --------------------------------------------------------------------- Intel GmbH Dornacher Strasse 1 85622 Feldkirchen/Muenchen Germany Sitz der Gesellschaft: Feldkirchen bei Muenchen Geschaeftsfuehrer: Douglas Lusk, Peter Gleissner, Hannes Schwaderer Registergericht: Muenchen HRB 47456 Ust.-IdNr. VAT Registration No.: DE129385895 Citibank Frankfurt (BLZ 502 109 00) 600119052 This e-mail and any attachments may contain confidential material for the sole use of the intended recipient(s). Any review or distribution by others is strictly prohibited. If you are not the intended recipient, please contact the sender and delete all copies. From Edric.Ellis at [hidden] Mon Mar 17 04:18:17 2008 From: Edric.Ellis at [hidden] (Edric Ellis) Date: Mon, 17 Mar 2008 09:18:17 -0000 Subject: [Mpi3-abi] Meeting notes from 10th March In-Reply-To: <6.2.3.4.2.20080313160548.02c2f358@ccs-mail.lanl.gov> Message-ID: I'm not sure how best to express this, but there are a couple of things that occur to me that might be important: 1. The size of the handle types (cf. size of a pointer perhaps?) 2. should we add some sort of table describing the current situation as to how applications pick up the value of e.g. MPI_COMM_WORLD? E.g. MPICH2 uses "#define MPI_COMM_WORLD 0x44000000", so that value is burned into the binary; whereas OpenMPI uses extern pointers - i.e. ompi_mpi_comm_world is in the initialized data section of libmpi.so, and the value resolved at (dynamic) link time. Cheers, Edric. > -----Original Message----- > From: mpi3-abi-bounces_at_[hidden] [mailto:mpi3-abi- > bounces_at_[hidden]] On Behalf Of Jeff Brown > Sent: Thursday, March 13, 2008 10:11 PM > To: MPI 3.0 ABI working group; mpi3-abi_at_[hidden] > Subject: Re: [Mpi3-abi] Meeting notes from 10th March > > I propose a way we can make progress ... > > Let's start populating a matrix (excel spreadsheet) with a column for > each MPI implementation, and rows for the various MPI datatypes, > constants, etc. where the internal implementations varys. I'll kick > it off for OpenMPI and send out. > > The last column of the matrix can be "ABI" where we propose a common > approach across the implementations. > > A couple of driving principles: > 1. the ABI solution shouldn't negatively impact quality of implementation > 2. minimize platform specific solutions > > I'd like to see if we can produce a single ABI that spans platforms. > > comments? > > Jeff > > > _______________________________________________ > mpi3-abi mailing list > mpi3-abi_at_[hidden] > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi From alexander.supalov at [hidden] Tue Mar 25 13:31:37 2008 From: alexander.supalov at [hidden] (Supalov, Alexander) Date: Tue, 25 Mar 2008 18:31:37 -0000 Subject: [Mpi3-abi] Meeting notes from 10th March In-Reply-To: <6B68D01C00C9994A8E150183E62A119E719263BDC6@NA-EXMSG-C105.redmond.corp.microsoft.com> Message-ID: <5ECAB1304A8B5B4CB3F9D6C01E4E21A2012DE212@swsmsx413.ger.corp.intel.com> Hi everybody, I've uploaded two instantiated MPICH2 1.0.3 files for IA32 and Intel(R) 64 platforms. Best regards. Alexander -----Original Message----- From: mpi3-abi-bounces_at_[hidden] [mailto:mpi3-abi-bounces_at_[hidden]] On Behalf Of Erez Haba Sent: Friday, March 14, 2008 7:47 PM To: MPI 3.0 ABI working group Subject: Re: [Mpi3-abi] Meeting notes from 10th March I have posted the mpi.h files to the wiki page and started a table comparing the various implementations. See http://svn.mpi-forum.org/trac/mpi-forum-web/wiki/Compare_mpi_h Thanks, .Erez -----Original Message----- From: mpi3-abi-bounces_at_[hidden] [mailto:mpi3-abi-bounces_at_[hidden]] On Behalf Of Solt, David George Sent: Friday, March 14, 2008 11:08 AM To: MPI 3.0 ABI working group Subject: Re: [Mpi3-abi] Meeting notes from 10th March I have to wonder if MPI_BOTTOM will cause us headaches. MPI_BOTTOM is meant to be used directly as an address. Our ABI would really have to specify exactly what that address is using either a constant or the address of a variable in the library. We have found that many non-compliant codes exist which assume that MPI_BOTTOM is zero. Assuming MPI_BOTTOM lives on to MPI-3, the ABI effort may want to dictate its value on a per-target architecture/OS basis, with ((void*) 0) the most obvious setting for nearly all targets. Dave -----Original Message----- From: mpi3-abi-bounces_at_[hidden] [mailto:mpi3-abi-bounces_at_[hidden]] On Behalf Of Jeff Squyres Sent: Friday, March 14, 2008 11:36 AM To: MPI 3.0 ABI working group Subject: Re: [Mpi3-abi] Meeting notes from 10th March Perhaps you could just link to OMPI's header file -- it's a live copy of what's in our SVN repository: https://svn.open-mpi.org/trac/ompi/browser/trunk/ompi/include/mpi.h.in On Mar 14, 2008, at 12:21 PM, Erez Haba wrote: > I will upload the various mpi.h files and start the table on the ABI > wiki page. > > -----Original Message----- > From: mpi3-abi-bounces_at_[hidden] > [mailto:mpi3-abi-bounces_at_[hidden] > ] On Behalf Of Jeff Brown > Sent: Thursday, March 13, 2008 3:11 PM > To: MPI 3.0 ABI working group; mpi3-abi_at_[hidden] > Subject: Re: [Mpi3-abi] Meeting notes from 10th March > > I propose a way we can make progress ... > > Let's start populating a matrix (excel spreadsheet) with a column for > each MPI implementation, and rows for the various MPI datatypes, > constants, etc. where the internal implementations varys. I'll kick > it off for OpenMPI and send out. > > The last column of the matrix can be "ABI" where we propose a common > approach across the implementations. > > A couple of driving principles: > 1. the ABI solution shouldn't negatively impact quality of > implementation 2. minimize platform specific solutions > > I'd like to see if we can produce a single ABI that spans platforms. > > comments? > > Jeff > > > _______________________________________________ > mpi3-abi mailing list > mpi3-abi_at_[hidden] > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi > > _______________________________________________ > mpi3-abi mailing list > mpi3-abi_at_[hidden] > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi -- Jeff Squyres Cisco Systems _______________________________________________ mpi3-abi mailing list mpi3-abi_at_[hidden] http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi _______________________________________________ mpi3-abi mailing list mpi3-abi_at_[hidden] http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi _______________________________________________ mpi3-abi mailing list mpi3-abi_at_[hidden] http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-abi --------------------------------------------------------------------- Intel GmbH Dornacher Strasse 1 85622 Feldkirchen/Muenchen Germany Sitz der Gesellschaft: Feldkirchen bei Muenchen Geschaeftsfuehrer: Douglas Lusk, Peter Gleissner, Hannes Schwaderer Registergericht: Muenchen HRB 47456 Ust.-IdNr. VAT Registration No.: DE129385895 Citibank Frankfurt (BLZ 502 109 00) 600119052 This e-mail and any attachments may contain confidential material for the sole use of the intended recipient(s). Any review or distribution by others is strictly prohibited. If you are not the intended recipient, please contact the sender and delete all copies.