1
1

7604 Коммитов

Автор SHA1 Сообщение Дата
Artem Polyakov
f2e586980b Fix timing framework:
1. Fixes according to (http://www.open-mpi.org/community/lists/devel/2014/09/15869.php)
2. Force mpisync:rank0 to gather results. Now sync info is written by rank0 to the output file.
3. Improve mpirun_prof: 1) adopt to the environment (SLURM/TORQUE); 2) recognize some noteset-related mpirun options.

This commit was SVN r32772.
2014-09-23 12:59:54 +00:00
Ralph Castain
9c20940190 Remove mpif-sizeof.h during distclean
This commit was SVN r32771.
2014-09-21 14:26:19 +00:00
Ralph Castain
70896550bf Per input from Artem, update the copyrights on these files, ensuring to include all the licensing info for the files broght over from the mpiperf project.
This commit was SVN r32770.
2014-09-20 14:54:24 +00:00
Jeff Squyres
7f419dc5b6 fortran: set CLEANFILES properly
CLEANFILES was previously set; we need to use += to add to it.

refs trac:4917

This commit was SVN r32769.

The following Trac tickets were found above:
  Ticket 4917 --> https://svn.open-mpi.org/trac/ompi/ticket/4917
2014-09-20 10:43:49 +00:00
Jeff Squyres
040611556f fortran: don't complain about script args if we're not building fortran
refs trac:4917

This commit was SVN r32766.

The following Trac tickets were found above:
  Ticket 4917 --> https://svn.open-mpi.org/trac/ompi/ticket/4917
2014-09-20 01:22:40 +00:00
Jeff Squyres
d7eaca83fa Fortran: Fix MPI_SIZEOF. What a disaster. :-(
What started as a simple ticket ended up reaching the way up to the
MPI Forum.
    
It turns out that we are supposed to have MPI_SIZEOF for all Fortran
interfaces: mpif.h, the mpi module, and the mpi_f08 module.
    
It further turns out that to properly support MPI_SIZEOF, your Fortran
compiler *has* support the INTERFACE keyword and ISO_FORTRAN_ENV.  We
can't use "ignore TKR" functionality, because the whole point of
MPI_SIZEOF is that the implementation knows what type was passed to it
("ignore TKR" functionality, by definition, throws that information
away).  Hence, we have to have an MPI_SIZEOF interface+implementation
for all intrinsic types, kinds, and ranks.

This commit therefore adds a perl script that generates both the
interfaces and implementations for MPI_SIZEOF in each of mpif.h, the
mpi module, and mpi_f08 module (yay consolidation!).

The perl script uses the results of some new configure tests:

* check if the Fortran compiler supports the INTERFACE keyword
* check if the Fortran compiler supports ISO_FORTRAN_ENV
* find the max array rank (i.e., dimension) that the compiler supports

If the Fortran compiler supports both INTERFACE and ISO_FORTRAN_ENV,
then we'll build the MPI_SIZEOF interfaces.  If not, we'll skip
MPI_SIZEOF in mpif.h and the mpi module.  Note that we won't build the
mpi_f08 module -- to include the MPI_SIZEOF interfaces -- if the
Fortran compiler doesn't support INTERFACE, ISO_FORTRAN_ENV, and a
whole bunch of ther modern Fortran stuff.

Since MPI_SIZEOF interfaces are now generated by the perl script, this
commit also removes all the old MPI_SIZEOF implementations (which were
laden with a zillion #if blocks).

cmr=v1.8.3

This commit was SVN r32764.
2014-09-19 13:44:52 +00:00
Rolf vandeVaart
5c73101a72 Fix typo.
This commit was SVN r32755.
2014-09-18 13:58:54 +00:00
Vasily Filipov
c7c63fe73e COLL/TUNED: alltoall - return previous default values of algorithm choosing decision thresholds (were changed by r32735)
reviewed by miked
    cmr=v1.8.3:reviewer=ompi-rm1.8

This commit was SVN r32753.

The following SVN revision numbers were found above:
  r32735 --> open-mpi/ompi@5fecf65daf
2014-09-18 08:07:51 +00:00
Jeff Squyres
0f29f222f2 fortran: remove 2 unused files
As noted in the comments of these files, they aren't used.  Instead,
the Fortran interfaces for WTICK/WTIME just BIND(C) invoke the
back-end C functions (yay BIND(C)!).  Hence, there's no need to keep
these old wrapper files around any more.

cmr=v1.8.3

This commit was SVN r32751.
2014-09-17 21:49:24 +00:00
Rolf vandeVaart
8db1f89dd1 Small change to allow CUDA-aware to work with non-reduction nonblocking collectives.
Only used when CUDA-aware feature compiled in.

This commit was SVN r32750.
2014-09-17 16:55:01 +00:00
Ralph Castain
d50c8ba65f Per patch from Gilles, cleanup some errors that surface when building with PGI. Verified by Tetsuya, reviewed okay by Jeff.
RM-approved

cmr=v1.8.3:reviewer=ompi-gk1.8

This commit was SVN r32745.
2014-09-16 19:07:02 +00:00
Vasily Filipov
ff10b25e7d warnings (caused by commit r32735) fix.
reviewed by miked
    cmr=v1.8.3:reviewer=ompi-rm1.8 

This commit was SVN r32740.

The following SVN revision numbers were found above:
  r32735 --> open-mpi/ompi@5fecf65daf
2014-09-16 06:33:49 +00:00
Ralph Castain
dfb952fa78 [Contribution from Artem - moved it to svn from git for him]
Replace our old, clunky timing setup with a much nicer one that is only available if configured with --enable-timing. Add a tool for profiling clock differences between the nodes so you can get more precise timing measurements. I'll ask Artem to update the Github wiki with full instructions on how to use this setup.

This commit was SVN r32738.
2014-09-15 18:00:46 +00:00
Vasily Filipov
5fecf65daf OMPI/COLL/Tuned: add command line params for thresholds to decide if small/intermediate MSGs alltoall algorithm will be used.
cmr=v1.8.3:reviewer=miked

This commit was SVN r32735.
2014-09-15 12:34:21 +00:00
Mangala Jyothi Bhaskar
dc05b709a7 it is ok to not have a sharedfp component selected, as long as no
sharedfp functionality is being used. Return an error however if no
sharedfp component is selected and the applications calls a
file_read/write_shared function.

This commit was SVN r32718.
2014-09-12 21:15:58 +00:00
Edgar Gabriel
597177cd8b silence a warning regarding the return value of the fbtl's.
This commit was SVN r32717.
2014-09-12 18:01:30 +00:00
Mangala Jyothi Bhaskar
cd78a3a026 Fixed offset data type in communication
This commit was SVN r32710.
2014-09-11 14:51:30 +00:00
Mangala Jyothi Bhaskar
4ff21d6178 Fixed offset data type in communication
This commit was SVN r32709.
2014-09-11 14:51:07 +00:00
Mangala Jyothi Bhaskar
6e5f2c8ae8 Fixed offset data type in communication
This commit was SVN r32708.
2014-09-11 14:50:30 +00:00
Edgar Gabriel
4ccc0f5ea2 the length of the iov array should be limited to IOV_MAX, which is defined in limits.h
This commit was SVN r32706.
2014-09-10 21:59:45 +00:00
Edgar Gabriel
cc46b65a5e the fbtl interfaces should really be an ssize_t not a size_t, since the return
value could be negative, which is allowed for ssize_t, but not for size_t.

This commit was SVN r32700.
2014-09-10 15:01:54 +00:00
Edgar Gabriel
599cb7b351 update the pvfs2 fbtl to return the number of bytes generated.
This commit was SVN r32699.
2014-09-10 13:32:06 +00:00
Gilles Gouaillardet
e71452d73a Revert r32696
This commit was SVN r32697.

The following SVN revision numbers were found above:
  r32696 --> open-mpi/ompi@e4c3500166
2014-09-10 04:35:47 +00:00
Gilles Gouaillardet
e4c3500166 Fix MPI_Status_set_elements[_x] for non predefined datatypes
Fixes trac:4896

cmr=v1.8.3:reviewer=bosilca

This commit was SVN r32696.

The following Trac tickets were found above:
  Ticket 4896 --> https://svn.open-mpi.org/trac/ompi/ticket/4896
2014-09-10 02:41:29 +00:00
Edgar Gabriel
3a5f4f72da make the zero byte read/write scenarios work without the contiguous flag.
This commit was SVN r32690.
2014-09-09 16:26:14 +00:00
Edgar Gabriel
6a607caed8 fix some zero byte allocation scenarios.
This commit was SVN r32689.
2014-09-09 16:25:44 +00:00
Edgar Gabriel
ed02927767 - do not set the contiguous memory option in the collective operations. It
should not be stored on the file handle anyway, since it is not a property of
the file.
- protect a realloc for zero byte scenarios.

This commit was SVN r32678.
2014-09-07 18:09:43 +00:00
Edgar Gabriel
0d425e2f74 resetting the counter for the iov array has to happen outside of the if statement.
This commit was SVN r32677.
2014-09-07 16:30:56 +00:00
Edgar Gabriel
0f59ce6591 use the fbtl return value as originally intended, namely to retrieve the
number of bytes written and read. Status contains now the actual number of
bytes written for individual operations. For collective operations, this is
unfortunately not possible.

This commit was SVN r32674.
2014-09-07 15:14:57 +00:00
Ralph Castain
41c6058153 Bring over changes to MXM from pmix branch:
MTL MXM: establish endpoint connection on the first communication when direct_modex used

This commit was SVN r32668.
2014-09-03 18:22:11 +00:00
Gilles Gouaillardet
edfbeba7bf coll/ml: better error handling
when CHECK_AND_RECYCLE detects an error, a message is displayed
if the error occurs on an intrinsic communicator, then abort
the program (instead of trying to free the communicator)

cmr=v1.8.3:reviewer=hjelmn

This commit was SVN r32659.
2014-09-01 10:00:49 +00:00
Ralph Castain
aae1bb4f44 Silence warning
This commit was SVN r32657.
2014-08-31 08:10:35 +00:00
Jeff Squyres
f4238d65a5 fortran: also provide PMPI variants for MPI_Alloc_mem_cptr
r32622 was the first half of the fix -- we need the PMPI variants as well.

Refs trac:4882

This commit was SVN r32627.

The following SVN revision numbers were found above:
  r32622 --> open-mpi/ompi@cf0f734a98

The following Trac tickets were found above:
  Ticket 4882 --> https://svn.open-mpi.org/trac/ompi/ticket/4882
2014-08-28 23:47:38 +00:00
Gilles Gouaillardet
cf0f734a98 Fortran: add mpi_alloc_mem_cptr like bindings when configured with --without-weak-symbols
cmr=v1.8.3:reviewer=jsquyres

This commit was SVN r32622.
2014-08-28 09:34:54 +00:00
Ralph Castain
b554cd7d86 Turn off the coll/ml component if --without-hwloc was given
cmr=v1.8.3:reviewer=jsquyres

This commit was SVN r32621.
2014-08-27 20:25:39 +00:00
Jeff Squyres
d85527701a Fix MPI_COMM_SPLIT_TYPE with MPI_UNDEFINED
Thanks to Lisandro Dalcin for identifying the problem.

Fixes trac:4876

Submitted by George Boscila, reviewed by Jeff Squyres.

cmr=v1.8.3:reviewer=ompi-rm1.8

This commit was SVN r32615.

The following Trac tickets were found above:
  Ticket 4876 --> https://svn.open-mpi.org/trac/ompi/ticket/4876
2014-08-27 12:17:33 +00:00
Gilles Gouaillardet
7e3784e0b7 MPI_Type_create_indexed_block.3: fix a typo in the man page
cmr=v1.8.3:reviewer=jsquyres

This commit was SVN r32614.
2014-08-27 03:48:03 +00:00
George Bosilca
8de93982d5 Correctly build the args for the hindexed_block datatype.
cmr=v1.8.3:reviewer=jsquyres

This commit was SVN r32613.
2014-08-27 03:45:07 +00:00
Edgar Gabriel
46de730059 fix a typo
This commit was SVN r32603.
2014-08-25 20:53:19 +00:00
Edgar Gabriel
52eac0146d cleanup of the fbtl interfaces: remove the *sorted optimization flag, since it
was not used anyway in the last two years. Simplifies the code significantly.

This commit was SVN r32602.
2014-08-25 18:04:24 +00:00
Todd Kordenbrock
6a3225d800 Fix invalid symbols left by the PMIx merge.
This commit was SVN r32597.
2014-08-25 16:30:26 +00:00
Jeff Squyres
e8eb07ad87 ompi_common_dll.c: the topo mtc union offset must be saved
Since the union contains pointers -- not instances -- we need to save
the mtc offset to get to the pointers later.

This commit was SVN r32591.
2014-08-23 15:42:44 +00:00
Ralph Castain
ac0c584eb7 Add missing file
This commit was SVN r32588.
2014-08-23 04:31:35 +00:00
Ralph Castain
b1a7375192 Fix the "unreachable" message so it outputs the correct hostname for the remote proc. Cleanup some of the pmix stuff when running corner cases of errors
This commit was SVN r32584.
2014-08-22 19:20:45 +00:00
Vishwanath Venkatesan
b176787d0f Remove unwanted spaces + Test commit
This commit was SVN r32576.
2014-08-22 05:11:17 +00:00
Edgar Gabriel
9987135da0 add initial support for non-blocking read and write operations.
This commit was SVN r32571.
2014-08-22 01:34:19 +00:00
Ralph Castain
aec5cd08bd Per the PMIx RFC:
WHAT:    Merge the PMIx branch into the devel repo, creating a new
               OPAL “lmix” framework to abstract PMI support for all RTEs.
               Replace the ORTE daemon-level collectives with a new PMIx
               server and update the ORTE grpcomm framework to support
               server-to-server collectives

WHY:      We’ve had problems dealing with variations in PMI implementations,
               and need to extend the existing PMI definitions to meet exascale
               requirements.

WHEN:   Mon, Aug 25

WHERE:  https://github.com/rhc54/ompi-svn-mirror.git

Several community members have been working on a refactoring of the current PMI support within OMPI. Although the APIs are common, Slurm and Cray implement a different range of capabilities, and package them differently. For example, Cray provides an integrated PMI-1/2 library, while Slurm separates the two and requires the user to specify the one to be used at runtime. In addition, several bugs in the Slurm implementations have caused problems requiring extra coding.

All this has led to a slew of #if’s in the PMI code and bugs when the corner-case logic for one implementation accidentally traps the other. Extending this support to other implementations would have increased this complexity to an unacceptable level.

Accordingly, we have:

* created a new OPAL “pmix” framework to abstract the PMI support, with separate components for Cray, Slurm PMI-1, and Slurm PMI-2 implementations.

* Replaced the current ORTE grpcomm daemon-based collective operation with an integrated PMIx server, and updated the grpcomm APIs to provide more flexible, multi-algorithm support for collective operations. At this time, only the xcast and allgather operations are supported.

* Replaced the current global collective id with a signature based on the names of the participating procs. The allows an unlimited number of collectives to be executed by any group of processes, subject to the requirement that only one collective can be active at a time for a unique combination of procs. Note that a proc can be involved in any number of simultaneous collectives - it is the specific combination of procs that is subject to the constraint

* removed the prior OMPI/OPAL modex code

* added new macros for executing modex send/recv to simplify use of the new APIs. The send macros allow the caller to specify whether or not the BTL supports async modex operations - if so, then the non-blocking “fence” operation is used, if the active PMIx component supports it. Otherwise, the default is a full blocking modex exchange as we currently perform.

* retained the current flag that directs us to use a blocking fence operation, but only to retrieve data upon demand

This commit was SVN r32570.
2014-08-21 18:56:47 +00:00
Mangala Jyothi Bhaskar
a5973c3f8c revamp of the aggregator selection logic, part 1.
This commit was SVN r32557.
2014-08-20 19:28:04 +00:00
Rolf vandeVaart
8709071819 Fix missing help file.
This commit was SVN r32550.
2014-08-18 21:52:31 +00:00
Jeff Squyres
0a398c155f opal MCA params: Move (and adapt) help message to opal help file
This commit was SVN r32547.
2014-08-16 11:54:41 +00:00