ports/openmpi
ports/openmpi
1
1
Форкнуть 0
Код Релизы Активность
2 726 коммитов 3 Ветки 0 Теги 130 MiB
7f2b73a4e5
Граф коммитов

2726 Коммитов

This Branch
This Branch
Все ветки
Автор SHA1 Сообщение Дата
George Bosilca
6788ad6681 The new datatype engine. With this commit the engine is completly imported in OpenMPI. 
Better memory handling, faster it adapt it's behaviour to the low level driver and to the
datatype representation. However, only the TCP driver has been modified to fit the new engine.

For Fortran the MPI_DATATYPE_NULL is now registered at index 0 in the translation table.

This commit was SVN r2757.
 2004-09-19 07:20:24 +00:00
George Bosilca
c3b73e3b46 In the begining the parent is set to comm_null. If there are no communications a segfault is generated 
when ompi_mpi_finalize is called.

This commit was SVN r2756.
 2004-09-19 07:13:58 +00:00
George Bosilca
57b1659943 New datatype engine. 
This commit was SVN r2755.
 2004-09-19 07:12:29 +00:00
George Bosilca
33aa1c1e87 Adapt to the new datatype representation. 
This commit was SVN r2754.
 2004-09-19 07:06:46 +00:00
George Bosilca
eb5e5191fa Adapt to the new datatype engine and correct some minor bugs. 
This commit was SVN r2753.
 2004-09-19 07:05:58 +00:00
George Bosilca
3482ab96d6 Update the definition of the Fortran MPI_DATATYPE_NULL. 
This commit was SVN r2752.
 2004-09-19 03:50:30 +00:00
Jeff Squyres
c85f931ae5 - query the daemon in a persistent universe to find out how many app 
slots we have to fill in MPI_UNIVERSE_SIZE
- in a non-persistent universe, just use the size of COMM_WORLD for
  MPI_UNIVERSE_SIZE.  Note the comment in the file, however, I don't
  think that this is sufficient -- e.g., reserve 8 CPUs in a batch
  environment and then run "./master" -- UNIVERSE_SIZE should still be
  8, even though we don't have a persistent universe.  So more
  integration is necessary here.
- removed some debugging printfs and added a forgotten error condition

This commit was SVN r2751.
 2004-09-18 10:04:31 +00:00
Jeff Squyres
1f98abaced Make all the .o files in the ROMIO distribution (in the romio io 
compnent) adhere to the MCA name prefix rule through some symbolic
linking trickery.  See comments in romio-dist/mpi-io/Makefile.am or
any of the adion Makefile.am's for details.

This commit was SVN r2750.
 2004-09-18 10:00:34 +00:00
Brian Barrett
2660d42f90 * fix typo that broke things in the no signals in the event library case. 
This commit was SVN r2749.
 2004-09-17 17:27:08 +00:00
Brian Barrett
857fd5740f * Test enabling signal handling in the event library. Because this might 
have some nasty side effect we don't know about, make it a configure
  option for now.  Also add a harmless signal handler to the pcm open
  (since pcm_open will have a signal handler eventually for SIGCHLD, I
  think).  Use --enable-event-signals / --disable-event-signals to 
  control behavior.

This commit was SVN r2748.
 2004-09-17 17:04:05 +00:00
Sayantan Sur
62ea85332c Basic MPI level message passing is working. Fixed a connect 
ack bug in connection management. There might be a slight race
condition still.

This commit was SVN r2747.
 2004-09-17 16:52:42 +00:00
Edgar Gabriel
1d16996a48 make comm_join work 
This commit was SVN r2746.
 2004-09-17 16:30:40 +00:00
Edgar Gabriel
65e4b61ec2 fixes to make comm_join work 
This commit was SVN r2745.
 2004-09-17 16:28:58 +00:00
Edgar Gabriel
6f3ebd184a minor bug fixes 
This commit was SVN r2744.
 2004-09-17 13:16:47 +00:00
Edgar Gabriel
ad3702b96d fixing some minor issues in connect/accept. A connect/accept between two groups of processes (within the same MPI_COMM_WORLD) works already 
This commit was SVN r2743.
 2004-09-17 13:16:14 +00:00
Edgar Gabriel
9c64200871 adapting some functions to the new interface of comm_connect_accept. This takes now an additional argument (a tag) to be able to distnguish several ongoing connec/accept/join/spawn operations simultaniously. 
This commit was SVN r2742.
 2004-09-17 10:12:31 +00:00
Edgar Gabriel
e14bbf5fc4 adaping some routines to use proc_find_and_add 
activating name_publishing (also tested, testcode will be checked in soon).
adaping the port-stuff to the new format.

This commit was SVN r2741.
 2004-09-17 10:11:22 +00:00
Edgar Gabriel
54bdb20351 removing the second lock. By introducing a new function proc_find_and_add we can work with a single lock. Actually, this makes the stuff for the dynamic process management easier. 
This commit was SVN r2740.
 2004-09-17 10:10:24 +00:00
Tim Woodall
6653be59fd defer sends for second+ fragments to improve overlap by the scheduler 
This commit was SVN r2739.
 2004-09-17 08:07:13 +00:00
Vishal Sahay
55bdaca610 Win f77 bindings 
This commit was SVN r2738.
 2004-09-17 06:44:06 +00:00
Vishal Sahay
e1cb4cc751 More f77 bindings 
This commit was SVN r2737.
 2004-09-17 04:33:28 +00:00
Vishal Sahay
c666e04b6e More file f77 bindings and a small fix in the spawn bindings 
This commit was SVN r2736.
 2004-09-17 04:03:19 +00:00
Ralph Castain
6704bc9d39 Completing the cleanup of NULLing free'd pointers. All should be done now. 
This commit was SVN r2735.
 2004-09-17 01:23:04 +00:00
Ralph Castain
0d4e6482cd Continuing the cleanup process. Few minor fixes here and there - mostly just NULLing pointers that were free'd. Console now can connect to any universe, regardless of scope. 
This commit was SVN r2734.
 2004-09-17 00:59:14 +00:00
Jeff Squyres
8699fa86b2 - remove some printf's 
- when copyying all attributes, if the callback function returns
  flag==0, don't copy the attribute

This commit was SVN r2733.
 2004-09-16 23:54:01 +00:00
Jeff Squyres
71054a834e - minor format changes 
- remove some unnecessary functions
- use OBJ_* macros
- fix ompi_hase_table_get_[first|next]_key_uint[32|64] functions

This commit was SVN r2732.
 2004-09-16 23:51:30 +00:00
Ralph Castain
e7135621fe Okay, with this minor change, mpirun2 and mpi_init can now both rapidly determine that they don't want to join an existing universe and create their very own session directory tree - and even clean up correctly IF the program terminates normally. 
Still have to work on that abnormal termination stuff...

This commit was SVN r2731.
 2004-09-16 21:42:48 +00:00
Vishal Sahay
88ee7c2715 Fix the warnings 
This commit was SVN r2730.
 2004-09-16 19:54:13 +00:00
Ralph Castain
f6dc129754 Allow mpirun2 and mpi_init to cleanly detect and join an existing universe. Will continue testing to quickly move away from a non-responsive existing universe. 
This commit was SVN r2729.
 2004-09-16 19:45:32 +00:00
Jeff Squyres
31bacaee5a Fix a minor problem with one of the configure command line options, 
and add more description to its help message

This commit was SVN r2728.
 2004-09-16 19:42:52 +00:00
Ralph Castain
d0dffbc5ee Minor adjustment to latest "fix" - if a trigger was removed, the functions still attempted to adjust some of its fields, leading to potentially random failures. I've now included a check to protect against this behavior. 
This commit was SVN r2727.
 2004-09-16 16:49:47 +00:00
Jeff Squyres
998b9b47d2 Remove the keyval lookup because it drags in the rest of libmpi in 
places where we don't want it (e.g., mpirun).

This commit was SVN r2726.
 2004-09-16 16:17:36 +00:00
Sayantan Sur
8b03c59099 Added recv buffer re-cycling code. 
This commit was SVN r2725.
 2004-09-16 16:17:27 +00:00
Rich Graham
0b4a7aede6 start to implement the send side. 
This commit was SVN r2724.
 2004-09-16 15:40:24 +00:00
Graham Fagg
62f5ad9979 Changed c2f calls to not call an errorhandler but instead use NULL objects instead 
(file is special case, and a few like win still not implemented).

This commit was SVN r2723.
 2004-09-16 15:14:55 +00:00
Brian Barrett
4619f315ba * checkpoint some changes to make things compile 
This commit was SVN r2722.
 2004-09-16 15:07:46 +00:00
Brian Barrett
e76f53ba1e * add query to figure out if RMS is actually active on the current machine 
or not
* Add logic to RMS pcm's configure script for passing the correct LIBS
  around during static builds

This commit was SVN r2721.
 2004-09-16 14:49:47 +00:00
Graham Fagg
5760f3ef8c fixed truncation check (buf=size of data is not a truncation!) 
This commit was SVN r2720.
 2004-09-16 14:36:38 +00:00
Tim Woodall
838066ab12 changed list iteration code to allow for an item to be removed 
from the list while iterating through it

This commit was SVN r2719.
 2004-09-16 14:34:03 +00:00
Tim Woodall
c32cb44a37 correct mca_oob_tcp_ping to cleanup correctly in the case of a timeout 
This commit was SVN r2718.
 2004-09-16 14:12:22 +00:00
Brian Barrett
46f755c016 * Fix missing comma in argument list to CHECK_FUNC 
This commit was SVN r2717.
 2004-09-16 13:46:24 +00:00
Brian Barrett
0716403174 * make sure to strip off quotes from the lines in post_configure.sh so that 
they don't cause escaping issues when executing the last eval.

This commit was SVN r2716.
 2004-09-16 13:45:40 +00:00
Tim Woodall
8d0851253a cleanup cleanup 
This commit was SVN r2715.
 2004-09-16 13:01:32 +00:00
Tim Woodall
cb48562fa5 temporary fix for mpirun2 - note this may break openmpi 
This commit was SVN r2714.
 2004-09-16 12:59:27 +00:00
Tim Woodall
7317caf2ca changed debug output - added sanity check 
This commit was SVN r2713.
 2004-09-16 12:58:50 +00:00
Tim Woodall
513129e816 fixed a recursive lock 
This commit was SVN r2712.
 2004-09-16 12:56:45 +00:00
Edgar Gabriel
5f29fe41fa fixing a stupid bug 
This commit was SVN r2711.
 2004-09-16 12:16:21 +00:00
Tim Woodall
30195b17cb - no need to register/unregister process info 
- cleanup event processing thread before anything else

This commit was SVN r2710.
 2004-09-16 11:30:44 +00:00
Edgar Gabriel
ed964097f5 a sever bug fix in comm_compare: we always comm1 with comm1, and ignored comm2. Strangely, we passed the Intel comm_compare test with this bug... 
This commit was SVN r2709.
 2004-09-16 10:08:40 +00:00
Edgar Gabriel
29110f6bc6 adapting some routines to the new interface of ompi_proc_get_namebuf etc... 
This commit was SVN r2708.
 2004-09-16 10:07:42 +00:00