ports/openmpi
ports/openmpi
1
1
Форкнуть 0
Код Релизы Активность
3 825 коммитов 3 Ветки 0 Теги
Граф коммитов

3825 Коммитов

This Branch
This Branch
Все ветки
Автор SHA1 Сообщение Дата
Prabhanjan Kambadur
cc3a8ec302 Making topo component lazy load 
This commit was SVN r4659.
 2005-03-03 22:42:10 +00:00
Brian Barrett
4c03e33a3c nightly checkpoint of portals code. Might actually load, but that's about 
the only useful thing it would do.  Changes with this commit:

 - started writing component API functions (progress still really empty)
 - started writing the bootstrapping code that has to be done to set the
   environment variables correctly for the utcp nal in the reference
   implementation.

This commit was SVN r4631.
 2005-03-02 23:26:06 +00:00
George Bosilca
a6974cfc64 Use coherent types over all files. 
This commit was SVN r4630.
 2005-03-02 14:21:22 +00:00
George Bosilca
ccc11068ca Correct a type mismatch in prototypes between the generic atomic.h and the ia32/atomic.h 
This commit was SVN r4629.
 2005-03-02 13:35:40 +00:00
Edgar Gabriel
6579de7791 next step: the interface still needs discussion (especially with UT), its however pretty close to what I think it should be. Hopefully, the bcast_segmented routine will be removed again soon and used from the UT coll module. 
This commit was SVN r4628.
 2005-03-02 13:28:39 +00:00
George Bosilca
d92de0c662 Add a missing include. 
This commit was SVN r4627.
 2005-03-02 05:13:30 +00:00
Brian Barrett
b3ed4aca32 * yet another fix for my stupidity in shell scripting 
This commit was SVN r4625.
 2005-03-01 23:49:00 +00:00
Brian Barrett
545a91729a * Fix dumb test condition that was causing George pain 
This commit was SVN r4624.
 2005-03-01 23:31:00 +00:00
Jeff Squyres
8116e71c16 Update docs to be more clear. 
This commit was SVN r4620.
 2005-03-01 22:56:07 +00:00
Brian Barrett
864cca2996 * fix dumb configure mistake - need to include the portals header before 
calling functions :)

This commit was SVN r4608.
 2005-03-01 20:38:09 +00:00
Josh Hursey
0f319b0582 Whoops. Shouldn't include datatypes directory in the include path 
This commit was SVN r4598.
 2005-03-01 16:57:03 +00:00
Josh Hursey
5f2e45dca1 Added true path to included files 
This commit was SVN r4597.
 2005-03-01 16:55:46 +00:00
Josh Hursey
c079718379 Some small adjustments 
This commit was SVN r4596.
 2005-03-01 16:02:10 +00:00
Josh Hursey
7716230371 Added a couple of needed defines 
This commit was SVN r4594.
 2005-03-01 15:39:58 +00:00
Jeff Squyres
996f813a2a Use a portable macro 
This commit was SVN r4572.
 2005-02-28 17:39:17 +00:00
Josh Hursey
02934da2af Added err.h to dist for windows build. 
Built into src win_makefile the ability to exclude files from the build to allow
  dist to build properly on windows

This commit was SVN r4570.
 2005-02-28 16:00:55 +00:00
Rainer Keller
e0e8b82ff7 - just add the srcdir to allow compilation in a srcdir!=topsrcdir 
This commit was SVN r4569.
 2005-02-28 15:08:55 +00:00
Jeff Squyres
f36e255d16 Update to match tim branch. 
This commit was SVN r4565.
 2005-02-25 21:18:08 +00:00
Josh Hursey
3eb139f2d2 Added files and directories needed for building under Windows to the dist. 
This commit was SVN r4561.
 2005-02-25 20:29:36 +00:00
Jeff Squyres
f1b1bfb062 Update some svn:ignore's to handle C++ template kruft 
This commit was SVN r4545.
 2005-02-25 12:33:57 +00:00
George Bosilca
44c3dba63c Add the 
This commit was SVN r4536.
 2005-02-24 23:00:37 +00:00
Edgar Gabriel
dbedfcda69 next step: a first version of how a non-segmented hierarchical bcast might look like with the current data structures 
This commit was SVN r4531.
 2005-02-24 20:05:12 +00:00
Josh Hursey
140f937880 preventing some barfing on the part of ln by forcing it to relink every single time 
This commit was SVN r4530.
 2005-02-24 19:00:26 +00:00
Edgar Gabriel
72ad1cbcf0 minor fixes 
This commit was SVN r4528.
 2005-02-24 15:27:17 +00:00
Edgar Gabriel
0fef989e1f first cut on the framework, which detects the two layer (sm/non-sm) hierarchy. 
It compiles, is however still far, far away from being usable.

This commit was SVN r4527.
 2005-02-24 15:11:07 +00:00
George Bosilca
e5d4e90b56 One more param test (the status cannot be NULL). 
This commit was SVN r4523.
 2005-02-24 00:13:27 +00:00
George Bosilca
33f62c7f75 Add a new entry in the MPI_Status struct. We need a place to keep the cancelled flag which can 
be set using MPI_Status_set_cancelled.

This commit was SVN r4522.
 2005-02-24 00:12:26 +00:00
George Bosilca
52529660e2 Unlike test the get_status function will not destroy the request. But we still call the 
ompi_progress function.

This commit was SVN r4521.
 2005-02-23 23:50:52 +00:00
George Bosilca
62aee24fd2 Do not unlock the mutex if it wasn't locked. 
This commit was SVN r4513.
 2005-02-23 08:21:02 +00:00
George Bosilca
3e437bbd7b Normally from the MPI point of view the count == 0 is LEGAL only for MPI_UB and MPI_LB. AND the MPI level 
is responsible for checking this condition. At the datatype engine, I accept it just as a nice way to set
the soft UB for a data (without using a real UB marker). This approach can be used to create
the subarray and darray datatype. However from the MPI level this function should never be
called directly with a count set to 0.

This commit was SVN r4512.
 2005-02-23 08:20:01 +00:00
George Bosilca
a02e269d9d Do not skip the current iteration if the length is zero. 
This commit was SVN r4511.
 2005-02-23 08:17:19 +00:00
George Bosilca
422eb0fa11 Add the 3 typeclass parameters MPI_TYPECLASS_INTEGER, MPI_TYPECLASS_REAL and MPI_TYPECLASS_COMPLEX. 
This commit was SVN r4510.
 2005-02-23 08:16:09 +00:00
George Bosilca
2dd9a4ab08 First cut of the darray type. I add the arg checking and the combiner creation. Next step ... the most 
complex MPI type creation !!!

This commit was SVN r4509.
 2005-02-23 08:10:23 +00:00
George Bosilca
9ff9af2ab6 The subarray type should now be completed. Even the combiner is correctly created ... 
This commit was SVN r4508.
 2005-02-23 08:08:49 +00:00
George Bosilca
4dd4de0c0b Correctly detect the predefined datatype for MPI_Type_match_size 
And add the F90 INTEGER type (for MPI_TYPE_CREATE_F90_INTEGER).

This commit was SVN r4507.
 2005-02-23 05:48:02 +00:00
Josh Hursey
0b41524cc6 Small adjustment... 
This commit was SVN r4506.
 2005-02-23 02:49:19 +00:00
George Bosilca
d4aacdccca Just to be 100% ANSI compliant the files using malloc and NULL require stdlib.h respectivly stdio.h. 
This commit was SVN r4505.
 2005-02-23 01:58:01 +00:00
George Bosilca
9de8c7db54 Remove the stdio.h and stdlib.h, they are useless. 
This commit was SVN r4499.
 2005-02-22 23:30:23 +00:00
George Bosilca
1c290b8c3a Update the Fortran match size MPI function. We have a different function here as we are not supposed to return 
the same predefined datatype as in the C version. Each language is responsible to return datatypes matching
the current language.

This commit was SVN r4497.
 2005-02-22 22:55:54 +00:00
George Bosilca
29ff7bfbb3 Correctly initialize the external32 convertor. 
Add a match_size function. This function match depending on the language, the kind of data required and
the size. Special care should be taken for C++ if we are looking for integer or real types as C++ can
return C datatypes. We use MPI_DATATYPE_NULL as a return error if there is no matching datatype.

This commit was SVN r4496.
 2005-02-22 22:34:27 +00:00
Jeff Squyres
e1a70f871d Add a little error checking to ensure that we got an mpool allocation. 
This commit was SVN r4482.
 2005-02-22 14:30:26 +00:00
Jeff Squyres
8834bd9e08 Temporarily turn this off -- active work is occurring elsewhere. 
This commit was SVN r4481.
 2005-02-22 14:29:10 +00:00
George Bosilca
0452af1cfd Add the types to the predefined data definitions. Correct the dumping function. 
This commit was SVN r4480.
 2005-02-22 08:15:39 +00:00
George Bosilca
c9c6ce1945 Starting to support the language and type flags for the data. They will be used to detect the 
correct size in type_match_size function. I update the dump datatype function. However these flags are
only available for predefined datatypes, it's seems to be useless to use them for the derived datatypes ...

This commit was SVN r4479.
 2005-02-22 07:40:42 +00:00
George Bosilca
40d42f38d4 Usefull to allow the tree to compile ... The definition of the external32 convertor and the initialization function. 
This commit was SVN r4478.
 2005-02-22 07:37:44 +00:00
George Bosilca
6640e334dc Add a bunch of unimplemented functions. We have now the external functions. 
This commit was SVN r4477.
 2005-02-22 00:17:31 +00:00
Josh Hursey
bf44a03b69 Fixed comments to reflect the notion that if a string parameter is read in from 
a file then it will always have a value, even if that value is the empty string.
(e.g. it will never be NULL)

This commit was SVN r4476.
 2005-02-21 19:09:14 +00:00
Josh Hursey
ba4bcc48db Added two new MCA Parameters: 
* mpi_show_mca_params
   If set to true, this turns on the dumping of all MCA parameters when MPI_INIT is called. 
   Only the 'rank 0' processes will print the parameters.

* mpi_show_mca_params_file
   (This value is only used if the first argument is set to true) If this value is non-NULL 
   it specifies the file to put the dump into. This file can then be used as input to mpirun 
   for debugging purposes. If this value is not set (and mpi_show_mca_params is set) then 
   the parameters are dumped to stdout.

I also changed the following parameters to internal=true:
  gpr_base_replica
  ns_base_replica
  pcmclient_env_cellid

This commit was SVN r4475.
 2005-02-21 18:56:30 +00:00
George Bosilca
d51fbdacd1 Add the copyright notice in the begining of all files. 
Start the external32 representation.

This commit was SVN r4474.
 2005-02-20 00:18:32 +00:00
George Bosilca
e9136c99f1 First cut of the architecture discovery files. Next step is defining external MPI type. 
This commit was SVN r4473.
 2005-02-19 23:53:26 +00:00