- Mpi-21 - mailman3.cels.anl.gov

Re: [Mpi-21] MPI-2.1 Version June 13, 2008 is ready
by unknown＠example.com 18 Jun '08

18 Jun '08

Rolf, A few small typos in mpi-report-2.1draft-2008-06-13-colored.pdf: * pg 229, ln 1: "the the location" should be "the location" * pg 229, ln 4: should be MPI_Comm_get_attr instead of set_attr. set_attr is covered on line 3. * pg 415, end of ln 19: "If the they" should be "If they" Rajeev > -----Original Message----- > From: mpi-21-bounces_at_[hidden] > [mailto:mpi-21-bounces_at_[hidden]] On Behalf Of Rolf > Rabenseifner > Sent: Friday, June 13, 2008 4:40 PM > To: MPI 2.1 Mailing List > Subject: [Mpi-21] MPI-2.1 Version June 13, 2008 is ready > > Hi all - and dear reviewer, > > it is Friday the 13th -- > and > MPI-2.1 Version June 13, 2008 is ready, see > > http://www.hlrs.de/organization/par/services/models/mpi/mpi21/doc/ > > user and pw = mpi21 > > Hopefully all bugs are fixed, see bug list on > http://www.hlrs.de/organization/par/services/models/mpi/mpi21/ > doc/MPI-2.1-reviews.pdf > > Dear reviewer, > please check whether your items are fixed and all is fine. > Only items that cannot wait until MPI 2.2 have a chance. > > !!! Latest deadline is June 20, 2008. Last chance. !!! > > Then I'll make the final MPI-2.1 for voting. > > It's still Friday the 13th. > > Best regards > Rolf > > > Dr. Rolf Rabenseifner . . . . . . . . . .. email rabenseifner_at_[hidden] > 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) > _______________________________________________ > mpi-21 mailing list > mpi-21_at_[hidden] > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-21 >

1 0

MPI-2.1 Version June 13, 2008 is ready
by unknown＠example.com 13 Jun '08

13 Jun '08

Hi all - and dear reviewer, it is Friday the 13th -- and MPI-2.1 Version June 13, 2008 is ready, see http://www.hlrs.de/organization/par/services/models/mpi/mpi21/doc/ user and pw = mpi21 Hopefully all bugs are fixed, see bug list on http://www.hlrs.de/organization/par/services/models/mpi/mpi21/doc/MPI-2.1-r… Dear reviewer, please check whether your items are fixed and all is fine. Only items that cannot wait until MPI 2.2 have a chance. !!! Latest deadline is June 20, 2008. Last chance. !!! Then I'll make the final MPI-2.1 for voting. It's still Friday the 13th. Best regards Rolf Dr. Rolf Rabenseifner . . . . . . . . . .. email rabenseifner_at_[hidden] 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

Re: [Mpi-21] running MPI only in a single function
by unknown＠example.com 03 Jun '08

03 Jun '08

Greetings Vishal. This list is not really for user-level questions about MPI; it is for the standards committee itself to discuss issues related to the upcoming MPI-2.1 standard. You might want to google around for various MPI tutorials and/or textbooks. Many of them address exactly this kind of topic. Good luck! On Jun 3, 2008, at 5:59 PM, Vishal Agarwal wrote: > Dear All > > I am trying to parallelize my code using MPI. My problem is i want > to run several processes only in one of the functions of my C++ code > (not in main). > Basically my code is divided into several functions. And in one of > the functions i want to parallelize. > > How can I do that using MPI............. > > -- > Regards > > *********************************************** > Vishal Agarwal > Research Scholar > *********************************************** > > _______________________________________________ > mpi-21 mailing list > mpi-21_at_[hidden] > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-21 -- Jeff Squyres Cisco Systems

1 0

running MPI only in a single function
by unknown＠example.com 03 Jun '08

03 Jun '08

Dear All I am trying to parallelize my code using MPI. My problem is i want to run several processes only in one of the functions of my C++ code (not in main). Basically my code is divided into several functions. And in one of the functions i want to parallelize. How can I do that using MPI............. -- Regards *********************************************** Vishal Agarwal Research Scholar *********************************************** *

1 0

Official reading for MPI-1.3 and MPI-2.1 is finished
by unknown＠example.com 03 May '08

03 May '08

At the meeting April 28-30, 2008, the MPI Forum has finished the review process for MPI-1.3 and MPI-2.1. The official reading is hereby finished. The final bug-fixes (discussed and voted at the meeting) will be done in the next weeks. All protocol files can be found at http://www.hlrs.de/mpi/mpi21/doc/ (username and pw is mpi21) and soon at the official protocol page at http://meetings.mpi-forum.org/secretary/2008/04/ All further enhancements will be part of MPI-2.2 (except obvious typos detected before final document assembly) I want to thank all who contributed to the ballots and the combined documents. Best regards Rolf Dr. Rolf Rabenseifner . . . . . . . . . .. email rabenseifner_at_[hidden] 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

MPI 2.1 doc changes
by unknown＠example.com 01 May '08

01 May '08

Rolf - Per the April meeting, I have made the following changes in mpi-forum- docs/MPI-2.1: 1. You likely have these noted somewhere already -- this is just to confirm to you that I performed these changes in the latex already (sections / line numbers based on the April 19 doc): - 5.9.5: p168.21: add missing C++ binding for User_function - 16.1.1: p449.25-26: delete sentence "See Annex ... for a function cross reference" because we deleted that table from the annex. --> I did not remove this table from Annex A; I think you were going to do it 2. Made the changes with regards to the beginning of Annex A: - Added the missing MPI-2 types - Added missing C++ types (all) - Updated to use the MPI-2 names and typedefs for attribute and errhandler callbacks - Added the missing C++ typedefs for attribute and errhandler callbacks - Split list of handle/opaque types and attribute/errhandler callback typedefs into two subsections - Created new subsection for the deprecated MPI-1 attribute and errhandler typedef names 3. Added many examples to the Examples Index. I'm sure we can always add more here. 4. Added all C++ types to the Declaration Index. 5. Added missing types from the Callback Function Prototype Index --> Do we want the deprecated typedefs listed here? Currently, they *are*. 6. Committed a change from Rainer to the MAKE-APPLANG script that enables building the doc properly on both Linux and OS X. 7. Committed new text introductions to many of the new indexes. 8. Fixed most (all?) of the non-functions listed in the MPI function index. There's at least one left that I'm aware of -- MPI_THREAD_INIT -- but I know that that item is already on your list to fix, so I left it for you. 9. I checked: all the pre-defined attribute callback functions (both old and new) show up properly in the main function index. I then stole a LaTeX trick from the bindings chapter to have functions show up in the C/Fortran indexes, but not have them show up in the chapter itself. That means that I didn't have to modify the index-generating scripts at all to make the 3 functions show up in the C/Fortran indexes properly (MPI_NULL_COPY_FN, MPI_DUP_FN, MPI_NULL_DELETE_FN). Woo hoo! -------------------------------------------------------- You can see a summary of all my changes this way: svn log -v -r119:132 To see the details of exactly what I changed in any given SVN commit, use this: svn diff -r N-1:N where N is an individual r number. For example, to see exactly what I did in r124: svn diff -r 123:124 With this mail, I am giving the SVN write token back to you. Please let me know if you have any questions, comments, etc. If I did something that you think is "bad", please let me know and I'll back it out. -- Jeff Squyres Cisco Systems

1 0

Chapter 3-4 minor comments and suggestions
by unknown＠example.com 29 Apr '08

29 Apr '08

dear all, ome final, minor comments to chapters 3 and 4 jesper *

1 0

Chapter 5 comments/suggestions
by unknown＠example.com 29 Apr '08

29 Apr '08

hi all uggestions/changes for Chapter 5 jesper (and adam) *

1 0

Chapter 1 && 2 minor comments
by unknown＠example.com 29 Apr '08

29 Apr '08

Dear all, here are some minor typos and sugestiosn for chaps 1&2 Jesper --- General: "message-passing" or "message passing"? Chapter 1: p.4, line 3: electonically -> electronically p.4, line 38: "hybrid architectures," - add space p.5 (ff): Update "Miscellany chapter" to something with info object p.6, line 17: drop "more convenient methods of constructing intercommunicators" (not in that chapter) p.6, line 22: correct chapter name (add "Caching") Chapter 2: p.94).", line 14: MPI-2 -> MPI p.9, line 15-19: drop (or reformulate) p.9, line 40-45: reformulate: "In many cases MPI names for C functions are of the form Class_action_subset. This convention originated with MPI-1. Since MPI-2 an attempt has been made to standardize the names of MPI functions according to the following rules. The C++ bindings in particular follow these rules (see Sectoion 2.6)." p.10, line 15: MPI-1 functions -> some MPI functions (that were defined during the MPI-1 process) p.10, line 26-30: bullet unintelligible. Change: An argument whose value may be used by a call but may not be updated is marked as IN An argument that may be updated by a call but whose value will not be used is marked as OUT An argument whose value may be used and that may be updated is marked as INOUT p.10, line 47: too much blank LaTeX space p.11, line 38: add: ".it is freed." -> ".it is freed, and becomes inactive if it was persistent." p.16, line 11: since MPI-2. -> since better solutions were provided with MPI-2. p.16, line 14: replace with: "With MPI-2, these functions were given new names with new bindings for the address arguments." p.16, line 19: "error-prone, ." -> "error-prone. The MPI-2 ." p.16, line 28: Change to "Originally MPI-1.1 provided bindings for Fortran 77. These bindings are retained, but they are now." p.16, line 45: Give some reason or excuse: "Unfortunately but unavoidably, the MPI Fortran bindings." p.18, line 6: MPI-2 -> MPI p.18, line 10: can't -> cannot p.19, line 45: MPI-2 -> MPI p.19, line 46: "the new MPI-2." -> "an MPI function is related to a function defined already for MPI-1 p.19, line 48: MPI-2 -> MPI p.20, line 2: drop "for MPI-1" p.20, line 3: "MPI-1 name is" -> "MPI name may be" p.23, line 3: drop latter part of sentence, "are described in Chapter ". p.23, line 32: MPI-2 -> MPI

1 0

MPI-2.1: Review of Chapter 1 and buglets in Chapter 15
by unknown＠example.com 29 Apr '08

29 Apr '08

Dear Rolf, please find enclosed the changes to Chapter 1 in the file below. Also, please note, that in Chapter 15, starting with p441, l38 the word language is misspelled several times: p441, l38 -- laguage --> language. With best regards, Rainer -- ---------------------------------------------------------------- Dipl.-Inf. Rainer Keller http://www.hlrs.de/people/keller HLRS Tel: ++49 (0)711-685 6 5858 Nobelstrasse 19 Fax: ++49 (0)711-685 6 5832 70550 Stuttgart email: keller_at_[hidden] Germany AIM/Skype:rusraink *

1 0