April 2008 - mpiwg-coll - mailman3.cels.anl.gov

[mpi3-coll] Draft of a proposal for persistent, sparse, nonblocking collectives
by Jesper Larsson Traeff 25 Apr '08

25 Apr '08

Dear all, attached is a first, very rough draft of a proposal for sparse, persistent, nonblocking collectives that Torsten and I have been discussing during IPDPS last week (with some local discussion here afterwards). It may serve to start discussion - either Torsten or I can present the gist of it during the group slot wednesday (not much time there, I just realize) Sorry for the rough state of it all... (note: it is *not* (yet) LaTeX) Jesper

1 0

Re: [mpi3-coll] April Telecon Notes
by Torsten Hoefler 11 Apr '08

11 Apr '08

On Fri, Apr 11, 2008 at 07:54:18AM +0100, Supalov, Alexander wrote: > Hi, > > Thanks for the minutes. I'm afraid I'm not on the list even though I > attended the meeting. you're right - I fixed it in the Wiki. Sorry & Best, Torsten -- bash$ :(){ :|:&};: --------------------- http://www.unixer.de/ ----- Indiana University | http://www.indiana.edu Open Systems Lab | http://osl.iu.edu/ 150 S. Woodlawn Ave. | Bloomington, IN, 474045-7104 | USA Lindley Hall Room 135 | +01 (812) 855-3608

1 0

Re: [mpi3-coll] April Telecon Notes
by Supalov, Alexander 11 Apr '08

11 Apr '08

Hi, Thanks for the minutes. I'm afraid I'm not on the list even though I attended the meeting. Best regards. Alexander -----Original Message----- From: mpi3-coll-bounces(a)lists.mpi-forum.org [mailto:[email protected]] On Behalf Of Torsten Hoefler Sent: Friday, April 11, 2008 12:30 AM To: mpi3-coll(a)lists.mpi-forum.org Subject: [mpi3-coll] April Telecon Notes Hi all, here are my notes from today's telecon. They are also in the wiki (http://svn.mpi-forum.org/trac/mpi-forum-web/wiki/Telecon041008) Please feel free to post additions or corrections and edit the wiki! Attendees (alphabetically): * Greg Bronevetsky (LLNL) * Torsten Hoefler (IU) * Bin Jia (IBM) * Andrew Lumsdaine (IU) * Amith Mamidala (OSU) * Adam Moody (LLNL) * Hans-Joachim Plum (Intel) * Christian Siebert (NEC) * Jesper Larsson Traff (NEC) * Rolf VandeVaart (Sun) Agenda 1. Schedule regular meeting time and interval 2. Updates on Non-blocking collectives - Torsten 1. revisiting tags 2. the interface (naming) 3. using MPI_Requests? 4. matching (rescue it?) 5. Free/Cancel? 3. Updates on Topological/Sparse Collectives 1. more than one neighbor per direction 2. Neighbor_allreduce 3. proposing changed graph interface officially 4. Dynamically-sized collective operations 1. malloc vs. upper bounded memory 2. new interface functions vs. changed semantics 5. Persistent Collectives 6. New Collective Ops 7. Modularity Discussion Notes 1. regular meetings will be on Thursdays 12pm once between two face to face meetings, might be more often later 2. topological collectives, we don't want to distinguish between graph and Cartesian communicators (Torsten) 1. do the more than one neighbor per direction via a conversion from Cartesian communicators to graph communicators (Torsten) 2. weather codes use "more than one neighbor", also non-symmetric (Adam) 3. but we don't want to double the function interfaces (Torsten) 4. do better analysis of codes (Adam) 5. consensus on Neighbor_allreduce 6. how do we optimize it? (Jesper) 7. changes to graph interface? We decided to write a proposal down, add Info object (Jesper and Torsten go off) 8. Allgather will be added to the proposal later 3. Non-blocking collectives 1. nobody spoke up for tags, so they're dropped officially now 2. naming conventions, "I" might be confusing but we can just define it clearly (Torsten) 3. Jesper proposes "N" as new name 4. using MPI_Requests would be helpful/more consistent (Torsten) 5. problems with Cancel/Free will occur (Adam, Alexander) 6. we can just stick to the current definition of Cancel which only defines send/recv clearly (Alexander) 7. Free would be useful too, could stick to the current definition (Alexander, Torsten) 8. we all agree to not allow matching 9. Cancel might also have problems with hardware support 10. Cancel should not be enforced but still be defined for implementations that can do it (might fail) 4. dynamic collectives might limit performance if used at places where one could use normal collectives (Adam) 1. we need an advice to users there 2. we need to add a new interface (max data-size is not in the old interface), old interface might limit scalability (huge arrays on huge communicators) 3. add also non-blocking versions of those operations 5. persistent collectives are useful 1. defined similarly to send/recv 2. matching persistent and non-persistent could limit optimization -> no matching 3. Alexander, Christian and Torsten go off and write it down 4. we need ordering in Startall now (changes semantics of other calls) 5. should tehy be part of non-blocking ops or not (subsetting)? -> we should have a separate module for persistent ops All the Best, Torsten -- bash$ :(){ :|:&};: --------------------- http://www.unixer.de/ ----- Indiana University | http://www.indiana.edu Open Systems Lab | http://osl.iu.edu/ 150 S. Woodlawn Ave. | Bloomington, IN, 474045-7104 | USA Lindley Hall Room 135 | +01 (812) 855-3608 _______________________________________________ mpi3-coll mailing list mpi3-coll(a)lists.mpi-forum.org http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-coll --------------------------------------------------------------------- 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.

1 0

[mpi3-coll] April Telecon Notes
by Torsten Hoefler 10 Apr '08

10 Apr '08

Hi all, here are my notes from today's telecon. They are also in the wiki (http://svn.mpi-forum.org/trac/mpi-forum-web/wiki/Telecon041008) Please feel free to post additions or corrections and edit the wiki! Attendees (alphabetically): * Greg Bronevetsky (LLNL) * Torsten Hoefler (IU) * Bin Jia (IBM) * Andrew Lumsdaine (IU) * Amith Mamidala (OSU) * Adam Moody (LLNL) * Hans-Joachim Plum (Intel) * Christian Siebert (NEC) * Jesper Larsson Traff (NEC) * Rolf VandeVaart (Sun) Agenda 1. Schedule regular meeting time and interval 2. Updates on Non-blocking collectives - Torsten 1. revisiting tags 2. the interface (naming) 3. using MPI_Requests? 4. matching (rescue it?) 5. Free/Cancel? 3. Updates on Topological/Sparse Collectives 1. more than one neighbor per direction 2. Neighbor_allreduce 3. proposing changed graph interface officially 4. Dynamically-sized collective operations 1. malloc vs. upper bounded memory 2. new interface functions vs. changed semantics 5. Persistent Collectives 6. New Collective Ops 7. Modularity Discussion Notes 1. regular meetings will be on Thursdays 12pm once between two face to face meetings, might be more often later 2. topological collectives, we don't want to distinguish between graph and Cartesian communicators (Torsten) 1. do the more than one neighbor per direction via a conversion from Cartesian communicators to graph communicators (Torsten) 2. weather codes use "more than one neighbor", also non-symmetric (Adam) 3. but we don't want to double the function interfaces (Torsten) 4. do better analysis of codes (Adam) 5. consensus on Neighbor_allreduce 6. how do we optimize it? (Jesper) 7. changes to graph interface? We decided to write a proposal down, add Info object (Jesper and Torsten go off) 8. Allgather will be added to the proposal later 3. Non-blocking collectives 1. nobody spoke up for tags, so they're dropped officially now 2. naming conventions, "I" might be confusing but we can just define it clearly (Torsten) 3. Jesper proposes "N" as new name 4. using MPI_Requests would be helpful/more consistent (Torsten) 5. problems with Cancel/Free will occur (Adam, Alexander) 6. we can just stick to the current definition of Cancel which only defines send/recv clearly (Alexander) 7. Free would be useful too, could stick to the current definition (Alexander, Torsten) 8. we all agree to not allow matching 9. Cancel might also have problems with hardware support 10. Cancel should not be enforced but still be defined for implementations that can do it (might fail) 4. dynamic collectives might limit performance if used at places where one could use normal collectives (Adam) 1. we need an advice to users there 2. we need to add a new interface (max data-size is not in the old interface), old interface might limit scalability (huge arrays on huge communicators) 3. add also non-blocking versions of those operations 5. persistent collectives are useful 1. defined similarly to send/recv 2. matching persistent and non-persistent could limit optimization -> no matching 3. Alexander, Christian and Torsten go off and write it down 4. we need ordering in Startall now (changes semantics of other calls) 5. should tehy be part of non-blocking ops or not (subsetting)? -> we should have a separate module for persistent ops All the Best, Torsten -- bash$ :(){ :|:&};: --------------------- http://www.unixer.de/ ----- Indiana University | http://www.indiana.edu Open Systems Lab | http://osl.iu.edu/ 150 S. Woodlawn Ave. | Bloomington, IN, 474045-7104 | USA Lindley Hall Room 135 | +01 (812) 855-3608

1 0

Re: [mpi3-coll] April teleconference - call-in information & tentative agenda
by Jesper Larsson Traeff 10 Apr '08

10 Apr '08

Hi all, since I'll be leaving for IPDPS, I'm not sure I can make the teleconf this afternoon. I will try my best regards Jesper On Tue, Apr 08, 2008 at 05:57:47PM -0400, Torsten Hoefler wrote: > Hello, > here is the information for our next teleconference: > > Date: 4/10/2008 > Begin Time: 12:00:00 PM Eastern Time (US & Canada) > Phone Number: +1 812 856-3600 > Participant PIN: 002118# > > Tentative Agenda: > 1. Schedule regular meeting time and interval - Andrew > > 2. Updates on Non-blocking collectives - Torsten > a) revisiting tags > b) the interface (naming) > c) using MPI_Requests? > d) matching (rescue it?) > e) Free/Cancel? > > 3. Updates on Topological/Sparse Collectives - Torsten, Jesper > a) more than one neighbor > b) Neighbor_allreduce > c) proposing changed graph interface officially > > 4. Dynamically-sized collective operations - Hans-Joachim, Torsten > a) malloc vs. upper bounded memory > b) new interface functions vs. changed semantics > > 5. Persistent Collectives - ? > > 6. New Collective Ops - Jesper > a) anything to discuss? MPI-2.2? > > 7) Modularity - Torsten > a) defining modules (blocking colls, non-blocking colls, ...?) > > Please feel free to add/bring up additional agenda items! > > Best, > Andrew Lumsdaine & Torsten Hoefler > > > -- > bash$ :(){ :|:&};: --------------------- http://www.unixer.de/ ----- > Indiana University | http://www.indiana.edu > Open Systems Lab | http://osl.iu.edu/ > 150 S. Woodlawn Ave. | Bloomington, IN, 474045-7104 | USA > Lindley Hall Room 135 | +01 (812) 855-3608 > _______________________________________________ > mpi3-coll mailing list > mpi3-coll(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-coll

1 0

[mpi3-coll] April teleconference - call-in information & tentative agenda
by Torsten Hoefler 08 Apr '08

08 Apr '08

Hello, here is the information for our next teleconference: Date: 4/10/2008 Begin Time: 12:00:00 PM Eastern Time (US & Canada) Phone Number: +1 812 856-3600 Participant PIN: 002118# Tentative Agenda: 1. Schedule regular meeting time and interval - Andrew 2. Updates on Non-blocking collectives - Torsten a) revisiting tags b) the interface (naming) c) using MPI_Requests? d) matching (rescue it?) e) Free/Cancel? 3. Updates on Topological/Sparse Collectives - Torsten, Jesper a) more than one neighbor b) Neighbor_allreduce c) proposing changed graph interface officially 4. Dynamically-sized collective operations - Hans-Joachim, Torsten a) malloc vs. upper bounded memory b) new interface functions vs. changed semantics 5. Persistent Collectives - ? 6. New Collective Ops - Jesper a) anything to discuss? MPI-2.2? 7) Modularity - Torsten a) defining modules (blocking colls, non-blocking colls, ...?) Please feel free to add/bring up additional agenda items! Best, Andrew Lumsdaine & Torsten Hoefler -- bash$ :(){ :|:&};: --------------------- http://www.unixer.de/ ----- Indiana University | http://www.indiana.edu Open Systems Lab | http://osl.iu.edu/ 150 S. Woodlawn Ave. | Bloomington, IN, 474045-7104 | USA Lindley Hall Room 135 | +01 (812) 855-3608

1 0

Re: [mpi3-coll] April teleconference date
by Torsten Hoefler 02 Apr '08

02 Apr '08

Hi Adam, > I think it may help to adopt a regular schedule like Rich has done with > the FT working group, where they call in at the same time on Friday > every two weeks. If we can find a reasonable time to start with, and > then we make the meetings at some regular time, I think folks will be > able to adapt their schedule to accommodate the call. This would > simplify the scheduling. yes, I agree. This also makes the setup of the teleconferences much easier for me. I will add the discussion of a time/periodicity for a regular telecon to the agenda of the next one. Thanks, Torsten -- bash$ :(){ :|:&};: --------------------- http://www.unixer.de/ ----- Indiana University | http://www.indiana.edu Open Systems Lab | http://osl.iu.edu/ 150 S. Woodlawn Ave. | Bloomington, IN, 474045-7104 | USA Lindley Hall Room 135 | +01 (812) 855-3608

1 0

Re: [mpi3-coll] April teleconference date
by Adam Moody 02 Apr '08

02 Apr '08

Hi Torsten, I think it may help to adopt a regular schedule like Rich has done with the FT working group, where they call in at the same time on Friday every two weeks. If we can find a reasonable time to start with, and then we make the meetings at some regular time, I think folks will be able to adapt their schedule to accommodate the call. This would simplify the scheduling. -Adam Torsten Hoefler wrote: >Hello, >there is only one date where not at least two people can't make it. >Thus, the teleconference will be at > > April 10, 12:00pm EDT > >Please let us know if there are any strong objections. An agenda and a number >to call will be anounced later. > >Thanks & Best, > Andrew Lumsdaine and Torsten Hoefler > > >

1 0

[mpi3-coll] April teleconference date
by Torsten Hoefler 02 Apr '08

02 Apr '08

Hello, there is only one date where not at least two people can't make it. Thus, the teleconference will be at April 10, 12:00pm EDT Please let us know if there are any strong objections. An agenda and a number to call will be anounced later. Thanks & Best, Andrew Lumsdaine and Torsten Hoefler -- bash$ :(){ :|:&};: --------------------- http://www.unixer.de/ ----- Indiana University | http://www.indiana.edu Open Systems Lab | http://osl.iu.edu/ 150 S. Woodlawn Ave. | Bloomington, IN, 474045-7104 | USA Lindley Hall Room 135 | +01 (812) 855-3608

1 0