1
1
Граф коммитов

8724 Коммитов

Автор SHA1 Сообщение Дата
George Bosilca
882b429f64 ompi_mtl_datatype_pack is not a data-type function (really) so it still
need the free_after (which btw has a different meaning that the one removed
from the data-type engine few minutes ago).

This commit was SVN r12333.
2006-10-27 00:15:53 +00:00
George Bosilca
393657ee26 Initialize the sndbuf in all cases. Do not forget to initialize the
tree used in each of the broadcast functions.

This commit was SVN r12332.
2006-10-27 00:13:33 +00:00
George Bosilca
126a68dc9a Big datatype commit. Remove all unused features of the datatype engine. As the memory
allocation logic is completely done outside the data-type engine (in the PML) there is
no need for any special case inside the data-type engine. There is less arguments for
the ompi_convertor_pack and ompi_convertor_unpack as well (the last field free_after is
not required anymore as there is no memory allocated in the engine itself). This change
affect all components using datatypes. I test most of them, but it might happens that I
miss some ... If it's the case please let me know (don't shoot the pianist!!).

This commit was SVN r12331.
2006-10-26 23:11:26 +00:00
George Bosilca
a1a4f7c422 Reset the segment pointer once we release the self fragment.
This commit was SVN r12330.
2006-10-26 23:07:14 +00:00
George Bosilca
be8516e0d7 Anothers indentations.
This commit was SVN r12329.
2006-10-26 23:06:15 +00:00
George Bosilca
83dfd36c1f Indentations.
This commit was SVN r12328.
2006-10-26 23:05:41 +00:00
George Bosilca
91ab093e96 Cleanup. No extern required for the function prototypes.
This commit was SVN r12327.
2006-10-26 23:03:12 +00:00
George Bosilca
ba3c247f2a Big collective commit. I lightly test it, but I think it should be quite stable. Anyway,
the default decision functions (for broadcast, reduce and barrier) are based on a
high performance network (not TCP). It should give good performance (really good) for
any network having the following caracteristics: small latency (5 microseconds) and good
bandwidth (more than 1Gb/s).
+ Cleanup of the reduce algorithms, plus 2 new algorithms (binary and binomial). Now most
  of the reduce algorithms use a generic tree based function for completing the reduce.
+ Added macros for computing the trees (they are used for bcast and reduce right now).
+ Allow the usage of all 5 topologies.
+ Jelena's implementation of a binary tree that can be used for non commutative operations.
  Right now only the tree building function is there, it will get activated soon.
+ Some others minor cleanups.

This commit was SVN r12326.
2006-10-26 22:53:05 +00:00
Tim Prins
894b220fbb Fixes trac:487
Give a more intelligible error message when someone passes -nolocal and the only available node is the local node.

This commit was SVN r12325.

The following Trac tickets were found above:
  Ticket 487 --> https://svn.open-mpi.org/trac/ompi/ticket/487
2006-10-26 21:46:18 +00:00
Andrew Friedley
c752502dee Fix for a common race condition when running the Sandia mt_send_recv.cc test.
A segfault would occur in mca_pml_ob1_recv_request_progress() when trying to prepare the convertor for unpacking, because the request's req_proc field was NULL.

Turns out that we weren't setting the req_proc field in the MCA_PML_OB1_CHECK_SPECIFIC_AND_WILD_RECEIVES_FOR_MATCH macro.  Instead of just setting it there I removed the other place req_proc was being set correctly, and instead took care of all the cases at once in mca_pml_ob1_recv_frag_match().

This commit was SVN r12323.
2006-10-26 19:09:39 +00:00
Gleb Natapov
90be664b9f Some process_pending() functions get bml_btl on which resource was freed as a
parameter. For optimisation purpose only this BTL is used to send packet
through instead of trying to send packets through all BTLs. But actually the 
code was wrong. It simply used provided bml_btl and it may represent different
endpoint from packet's destination. The fixed code checks if packet's
destination is reachable through the BTL, finds appropriate bml_btl and only
then tries to send it through correct bml_btl.

This commit was SVN r12319.
2006-10-26 13:21:47 +00:00
Sven Stork
5861ed865d - Add parameter checking as required by the standard.
This commit was SVN r12318.
2006-10-26 09:18:21 +00:00
Sven Stork
9024c5be4b - Fix wrong error values.
This commit was SVN r12317.
2006-10-26 08:26:03 +00:00
Terry Dontje
7259d1b512 Adjust allocation size to be a quantity divisible by sizeof(size_t). This
is done to assure alignment so strictly aligned CPUs (like SPARC) do not
sigbus.  This also may benefit other platforms too.

This commit fixes trac:494.

This commit was SVN r12312.

The following Trac tickets were found above:
  Ticket 494 --> https://svn.open-mpi.org/trac/ompi/ticket/494
2006-10-25 18:22:38 +00:00
Ralph Castain
36d4511143 Bring the timing instrumentation to the trunk.
If you want to look at our launch and MPI process startup times, you can do so with two MCA params:

OMPI_MCA_orte_timing: set it to anything non-zero and you will get the launch time for different steps in the job launch procedure. The degree of detail depends on the launch environment. rsh will provide you with the average, min, and max launch time for the daemons. SLURM block launches the daemon, so you only get the time to launch the daemons and the total time to launch the job. Ditto for bproc. TM looks more like rsh. Only those four environments are currently supported - anyone interested in extending this capability to other environs is welcome to do so. In all cases, you also get the time to setup the job for launch.

OMPI_MCA_ompi_timing: set it to anything non-zero and you will get the time for mpi_init to reach the compound registry command, the time to execute that command, the time to go from our stage1 barrier to the stage2 barrier, and the time to go from the stage2 barrier to the end of mpi_init. This will be output for each process, so you'll have to compile any statistics on your own. Note: if someone develops a nice parser to do so, it would be really appreciated if you could/would share!

This commit was SVN r12302.
2006-10-25 15:27:47 +00:00
Brian Barrett
d6ff14ed61 Hand-pack the connection information for each peer rather than just
packing a sockaddr_in, as there are some endianness and padding issues
with sending a sockaddr_in.  Note that the sin_port and sin_addr are
already in network byte order, which is why we pack them as a byte
string.

Refs trac:493

This commit was SVN r12301.

The following Trac tickets were found above:
  Ticket 493 --> https://svn.open-mpi.org/trac/ompi/ticket/493
2006-10-25 15:09:30 +00:00
George Bosilca
2d17f0fa9d First step on supporting full external32 conversion on both operations
pack and unpack.

This commit was SVN r12299.
2006-10-25 14:33:06 +00:00
George Bosilca
ccbe5ee016 Remove unused code.
This commit was SVN r12297.
2006-10-25 13:59:57 +00:00
Jeff Squyres
126a8b1e22 If we pass an erroneous error code in, don't segv. Instead, return a
"this is bogus" kind of answer.  Passing in bad error codes should
only happen in erroneous sections of the OMPI code base, but still,
it's far more social to print a message saying, "hey, you messed up!"
rather than seg faulting.

Reviewed by Edgar.

This commit was SVN r12295.
2006-10-25 13:37:45 +00:00
Sven Stork
f3f39e003e - Increment the pipeline depth before we trigger the send function. As
mentioned in the comment the completion/callback of the triggered 
  send operation can happen before the call returns. If this happens and
  if the pipeline depth is 0 before we triggered the send operation and 
  this is the last send operation of the request then the completion detection
  code will decrement the pipeline depth and check it for equality to 0.
  Because (0-1) != 0 the pml completion function for this request will 
  *not* be called.
  This part 2 of the fix for ticket #246.

This commit was SVN r12292.
2006-10-25 08:52:39 +00:00
Sven Stork
3563f15fde - Fix a bug in descriptor handling code. The self BTL was mixing the different
kinds of descriptors (e.g. put rdma descriptor in the eager free-list). 
  This part 1 of the fix for ticket #246.

This commit was SVN r12291.
2006-10-25 08:45:29 +00:00
Ralph Castain
601443e690 Fix the other end of the attribute store/retrieve to correctly handle attributes with no value
This commit was SVN r12290.
2006-10-25 01:41:58 +00:00
Ralph Castain
46166a9c77 Fix a bug that would cause a segfault when someone specified an option that had no corresponding value
This commit was SVN r12288.
2006-10-24 22:06:18 +00:00
Brian Barrett
dec51d7cf7 When Apple converted to UNIX03 support (part of the feature list of 10.5), they
renamed the register fields in the thread state structures.  Support compiling
with either the old or new names, keying off the UNIX03 define (which is what
the 10.5 headers do).

Refs trac:450

This commit was SVN r12285.

The following Trac tickets were found above:
  Ticket 450 --> https://svn.open-mpi.org/trac/ompi/ticket/450
2006-10-24 18:30:09 +00:00
Ralph Castain
a7acd22e47 Fix a fix so we see the correct MCA param
This commit was SVN r12282.
2006-10-24 17:24:09 +00:00
George Bosilca
ed3caa9fc7 mca_base_param_reg_int require a pointer to a component not a name.
This commit was SVN r12281.
2006-10-24 16:41:51 +00:00
George Bosilca
7dec7731ce Instead of size_t use orte_std_cntr_t. Remove all warnings.
This commit was SVN r12280.
2006-10-24 16:40:49 +00:00
Ralph Castain
8636ac6a4d Fix ticket 353 - print out a nice message that the combination of debug-daemons and num_concurrent in the pls rsh launcher will cause deadlock and exit
This commit was SVN r12279.
2006-10-24 15:59:02 +00:00
Tim Prins
cb622db7c9 Fixes trac:352
Only close off stdout/stderr from the daemons if we are not debugging the slurm pls and --debug-daemons was not passed.

This commit was SVN r12276.

The following Trac tickets were found above:
  Ticket 352 --> https://svn.open-mpi.org/trac/ompi/ticket/352
2006-10-24 13:05:13 +00:00
Tim Prins
93d61d01fb Fix for a problem on SLURM we have neen having since r12243 where mpirun would hang after the process had finished. It turns out that we were always reporting the name of the daemon wrong, but we simply never noticed as we never used it, until r12243. This makes it so we report the name of the daemon correctly.
This commit was SVN r12274.

The following SVN revision numbers were found above:
  r12243 --> open-mpi/ompi@153e38ffc9
2006-10-24 01:41:28 +00:00
Jeff Squyres
26781a567d Fix contributed by Joe Landman to allow building on SuSE.
This commit was SVN r12273.
2006-10-24 01:26:13 +00:00
George Bosilca
99631ccf66 Cleanups.
This commit was SVN r12272.
2006-10-23 22:29:17 +00:00
George Bosilca
d7d3f9e486 Tuned collectives works only for at least 2 processes. We have the self module
for the other cases.

This commit was SVN r12271.
2006-10-23 22:28:56 +00:00
George Bosilca
b848a5ad06 Remove all ompi_coll_chain_t references.
This commit was SVN r12269.
2006-10-23 21:47:50 +00:00
George Bosilca
39cd8d3d17 One to rule them all. We only need one topology information: a tree. How we
build it it's hat make the difference.

This commit was SVN r12268.
2006-10-23 21:46:30 +00:00
Rolf vandeVaart
272f766c5f Fix for ticket #219 MPI::Grequest is missing from C++ API. I did the initial implementation and Jeff fixed it up. Passes a new test in trunk/simple/basic/cxx/grequest.cc.
This commit was SVN r12264.
2006-10-23 20:17:30 +00:00
George Bosilca
9cf3040e5f Allocate enough memory for the reduce operation when MPI_IN_PLACE is specified.
This commit was SVN r12260.
2006-10-23 17:51:36 +00:00
Edgar Gabriel
8b09bd181f just a reordering of the arguments in a comparison in order to comply with the
OMPI programming style...

This commit was SVN r12259.
2006-10-23 17:14:23 +00:00
Ralph Castain
7a77ef0ae3 Given the amount of pain singletons cause, one can't help but wonder if it REALLY was that much trouble for people to type "mpirun -n 1 foo"....sigh.
Get the ordering right so that a singleton can start.

Protect the rmgr copy app_context function from NULL fields

Tell the mapper it is okay for there not to be a pre-existing mapping plan for a parent when dynamically spawning processes

This commit was SVN r12257.
2006-10-23 15:15:45 +00:00
Ralph Castain
c5b59829aa Fix a long-lingering annoyance. Calling mpirun with a non-existent application would cause the system to hang on all environments. Reason was that the orted would exit, which it should never do without explicit orders to that affect.
This commit was SVN r12255.
2006-10-23 13:27:31 +00:00
George Bosilca
6745e1a7b7 Correctly detect the STDC_HEADERS, before adding flags to the CFLAGS. Now,
that we have the STDC_HEADERS we are able to detect when ptrdiff_t is defined.

This commit was SVN r12254.
2006-10-23 03:55:52 +00:00
George Bosilca
ee559e9947 Do not completely reset the orterun_globals. Keep the condition and the mutex,
but reset everything else. Once initialized the condition (and the attached
mutex) should be kept alive as long as possible if we want to be able to
retrieve all the informations.

This commit was SVN r12253.
2006-10-23 03:34:08 +00:00
Jeff Squyres
533b4db7b5 Fix a problem when $FC/$F77 were multiple tokens.
This commit was SVN r12246.
2006-10-21 14:24:30 +00:00
Jeff Squyres
7b6536e912 * Add a check for whether we're operating on environ; if so, don't
free anything.
 * Fix a small memory leak.

This commit was SVN r12245.
2006-10-21 13:46:04 +00:00
George Bosilca
2a863df0a5 Newline is required by some compilers at the end of a file.
This commit was SVN r12244.
2006-10-21 05:56:04 +00:00
Ralph Castain
153e38ffc9 Lesson to be learned: if you send an ack to a recv'd command, better not send it to the same tag it came from - at least, not if there is a persistent recv on that tag!
Fix the persistent daemon problem where it was exiting when a job completed. Problem was that the persistent daemon would order the job daemons to exit. They would then send an 'ack' back to the persistent daemon - but the ack consisted of an echo of the "exit" command, which was recv'd by the wrong listener who treated it as a properly sent cmd....and exited.

This commit was SVN r12243.
2006-10-21 02:53:19 +00:00
George Bosilca
6b697ad3dd If the operation is not commutative then force the basic reducve algorithm. The others
cannot be used for non commutative operations ... yet ...

This commit was SVN r12241.
2006-10-20 22:11:44 +00:00
Ralph Castain
ab7bbb80a5 Teach the mapper to correctly handle the unbalanced --host scenario. We now map in a more expected fashion.
This commit was SVN r12240.
2006-10-20 20:48:24 +00:00
Jeff Squyres
06d1c11efd Add MPI_GROUP_TRANSLATE_RANKS to the list.
This commit was SVN r12239.
2006-10-20 20:21:16 +00:00
George Bosilca
a7b6078b73 No more segfault. Still some wrong data around ...
This commit was SVN r12238.
2006-10-20 20:17:34 +00:00