ports/openmpi
ports/openmpi
1
1
Форкнуть 0
Код Релизы Активность
17 001 коммит 3 Ветки 0 Теги 130 MiB
019857b616
Граф коммитов

3801 Коммитов

Автор SHA1 Сообщение Дата
Mike Dubman
a45898ea9c fix support for fca 2.2, warning fixes on rhel 6.x 
This commit was SVN r26166.
 2012-03-20 10:00:52 +00:00
Nathan Hjelm
135ac32b64 ugni: use hash table to keep track of smsg frag completion 
This commit was SVN r26154.
 2012-03-15 20:15:59 +00:00
Nathan Hjelm
fca42347e3 ugni: use hash table to keep track of smsg frag completion 
This commit was SVN r26153.
 2012-03-15 20:13:32 +00:00
Nathan Hjelm
deddf0b33e ugni: fix frag leak in sendi 
This commit was SVN r26152.
 2012-03-15 20:13:20 +00:00
Nathan Hjelm
99f05d56e3 ugni: updated parameters and code cleanup 
This commit was SVN r26151.
 2012-03-15 20:13:11 +00:00
Nathan Hjelm
921176745d vader: remove lock based fifos 
This commit was SVN r26150.
 2012-03-15 20:12:59 +00:00
Nathan Hjelm
4e01440b05 vader: clean frag alloc/return 
This commit was SVN r26149.
 2012-03-15 20:12:46 +00:00
Terry Dontje
e73df369e4 Update bfo pml with code from ob1 to support mprobe, improbe, mrecv, imrecv and cuda. 
This commit was SVN r26145.
 2012-03-15 10:20:46 +00:00
Christopher Yeoh
524de80eaa Adds support for Cross Memory Attach in the sm btl. 
This feature can be enabled at compile time with --with-cma passed
to configure.

At runtime it is also necessary to add "--mca btl btl_sm_use_cma 1"
to the mpirun command.

If both CMA and KNEM are compiled in and enabled at runtime then 
KNEM will take precedence and CMA will disable itself

This commit was SVN r26134.
 2012-03-14 06:29:09 +00:00
Mike Dubman
bd7abd72a9 in mca_mtl_mxm, don't allow negative tags for MPI_ANY_TAG 
This commit was SVN r26128.
 2012-03-09 22:11:14 +00:00
Rolf vandeVaart
41870ce6ee Mostly fix some of the verbose output. Also fix issue 
where memory handle was blocking other registration.

This commit was SVN r26124.
 2012-03-09 21:28:56 +00:00
George Bosilca
de1078a71b Thanks to Alex Margolin for pointing out this relique. 
This commit was SVN r26121.
 2012-03-09 14:01:45 +00:00
Pavel Shamis
102da281c4 OPENIB BTL - use orte_show_help instead of BTL_ERROR print in case ibv_reg_mr failed. 
This commit was SVN r26111.
 2012-03-08 09:04:03 +00:00
Mike Dubman
4e7e7d7c3f print error which is ignored on upper layer 
This commit was SVN r26106.
 2012-03-06 14:25:56 +00:00
George Bosilca
a78a7bd8e8 The tuned collectives can now deal with more than 2Gb of data. 
This commit was SVN r26103.
 2012-03-05 22:23:44 +00:00
George Bosilca
762b3e13a9 Use the correct name for the datatype destruction function. 
This commit was SVN r26100.
 2012-03-05 15:54:53 +00:00
George Bosilca
7d523a8852 Avoid calling the bcast with counts larger than INT_MAX. 
This commit was SVN r26098.
 2012-03-05 14:30:30 +00:00
George Bosilca
e8c358c188 Allow Open MPI to deal with size_t internally. 
This commit was SVN r26097.
 2012-03-05 14:10:26 +00:00
Abhishek Kulkarni
08ca0f80bc Fix a C/R bug where the restart hung due to 
dangling fds in the openib btl.

This commit was SVN r26094.
 2012-03-04 06:57:33 +00:00
George Bosilca
f83670211e Allow the user to define dynamic rules for messages larger than 2GB. 
This commit was SVN r26084.
 2012-03-02 21:16:23 +00:00
George Bosilca
8791ade293 Help he selection of the right algorithm for large data (> 2Gb). Thanks 
to Fujitsu for the patch.

This commit was SVN r26080.
 2012-03-02 19:12:22 +00:00
Terry Dontje
3e70cad203 Correct a few alignment problems to address the issue brought up in ticket #2964 
This commit was SVN r26078.
 2012-03-01 17:29:40 +00:00
Nathan Hjelm
f1525bdbff ob1: fix two fragment leaks 
- MAJOR! get src descriptor leaks if mca_bml_base_send fails
 - minor. descriptor leaked in mca_pml_send_request_start_copy if the btl returns OMPI_ERR_RESOURCE_BUSY.

This commit was SVN r26077.
 2012-03-01 15:53:39 +00:00
Mike Dubman
540b3c0c25 update mxm mtl to changes in mxm api 
This commit was SVN r26073.
 2012-02-29 22:02:34 +00:00
Nathan Hjelm
a7209e309a ugni: opps, sendi was missing from Makefile.am 
This commit was SVN r26067.
 2012-02-28 16:10:35 +00:00
Edgar Gabriel
636cf786be local_heap_sort should really be a static function. 
This commit was SVN r26065.
 2012-02-28 14:42:56 +00:00
Vishwanath Venkatesan
7c9c3ede61 Modified implementation for the static segmentation read algorithm with improved 
performance and bug fixes.

This commit was SVN r26056.
 2012-02-24 20:55:33 +00:00
Vishwanath Venkatesan
d5a9223a9a Removed a variable which was allocated but never used. 
This commit was SVN r26055.
 2012-02-24 20:48:52 +00:00
Vishwanath Venkatesan
326bc69df4 Modified implementation for static file write all algorithm which fixes all the previous 
bugs and provides improved performance.

This commit was SVN r26054.
 2012-02-24 20:47:09 +00:00
Rolf vandeVaart
b0a84b0a7d New btl that extends sm btl to support GPU transfers within a node. 
Uses new CUDA IPC support.  Also, a few minor changes in PML to take
advantage of it.

This code has no effect unless user asks for it explicitly via 
configure arguments.  Otherwise, it is either #ifdef'ed out or
not compiled.

This commit was SVN r26039.
 2012-02-24 02:13:33 +00:00
Nathan Hjelm
8217c46666 ompi_free_list: allocate payload if payload size > 0 in the fl_mpool = NULL case 
This commit was SVN r26027.
 2012-02-23 16:47:28 +00:00
Nathan Hjelm
9843cd0466 ugni: missed one more merge typo 
This commit was SVN r26026.
 2012-02-23 16:39:15 +00:00
Nathan Hjelm
d7cd95c802 vader: fix typo 
This commit was SVN r26025.
 2012-02-23 16:29:45 +00:00
Nathan Hjelm
c83fe003a0 ugni: fix branch merge typo 
This commit was SVN r26024.
 2012-02-23 16:16:21 +00:00
Rolf vandeVaart
c7a0ce2755 Two new mpools. They are not used now (and by default, not compiled) 
but they will be soon.  Provide support for GPU buffer transfers within a node.

This commit was SVN r26008.
 2012-02-22 23:32:36 +00:00
Nathan Hjelm
dc94b6a3fb vader: minor fast box optimization 
This commit was SVN r26002.
 2012-02-22 20:53:49 +00:00
Nathan Hjelm
4c7b7c675a vader: minor code cleanup. move xpmem_create to component_init 
This commit was SVN r25999.
 2012-02-22 18:32:40 +00:00
Mike Dubman
81bd5eee8d in mxm, use sender_len field and not actual_len when returning result from probe 
This commit was SVN r25993.
 2012-02-21 19:55:16 +00:00
Jeff Squyres
b6a90434e4 Fix some include file header ordering issues for some BSDs, suggested 
by Paul Hargrove.

This commit was SVN r25984.
 2012-02-21 13:32:14 +00:00
Jeff Squyres
7fb1144d27 Add missing AC_CONFIG_FILES 
This commit was SVN r25974.
 2012-02-20 22:59:03 +00:00
Jeff Squyres
6a109c8e15 Ensure that we have aio.h before trying to compile this component. 
This commit was SVN r25966.
 2012-02-20 15:53:20 +00:00
George Bosilca
bd9402fd8d Fix call to error_cb. 
This commit was SVN r25946.
 2012-02-17 03:18:51 +00:00
Jeff Squyres
4c0d24ff9a Improve the help messages for tcp_btl_if_[in|ex]clude 
This commit was SVN r25940.
 2012-02-16 15:59:18 +00:00
Nathan Hjelm
1ee6d5d21a ugni: fix typos in mca_btl_ugni_put 
This commit was SVN r25937.
 2012-02-15 23:26:03 +00:00
Mike Dubman
6ec768f0c6 fix #2971 
This commit was SVN r25908.
 2012-02-12 09:28:42 +00:00
Nathan Hjelm
8c9dc990a9 ugni: fixed typo 
This commit was SVN r25899.
 2012-02-10 01:27:31 +00:00
Nathan Hjelm
994127cb6b ugni: move mca_btl_ugni_smsg_mbox_t back to btl_ugni_endpoint.h 
This commit was SVN r25898.
 2012-02-10 01:11:54 +00:00
Nathan Hjelm
0ccfd3e6db ugni btl update. changes: 
- re-enable sendi
 - move smsg common code into btl_ugni_smsg.h
 - added new parameters for smsg/eager frags
 - use get for frags larger than the smsg_limit
 - bug fixes
 - code cleanup

This commit was SVN r25897.
 2012-02-10 00:47:29 +00:00
Mike Dubman
b18a1611c3 - if everything is ok set return value to OMPI_SUCCESS in mtl/mxm 
This commit was SVN r25879.
 2012-02-08 14:19:58 +00:00
Christopher Yeoh
bc26adcc32 Fixes trac:2998 
Adds a lock to protect the sm pending_sends list from concurrent access
Fixes bug where btl_sm_process_pending_sends would return an item to
the free list and then continue to use it for a little while

cmr:v1.6

This commit was SVN r25878.

The following Trac tickets were found above:
  Ticket 2998 --> https://svn.open-mpi.org/trac/ompi/ticket/2998
 2012-02-08 01:32:36 +00:00
Brian Barrett
25d48e22fa Implementation of the MPI-3 Matched Probe functionality. Currently only 
implemented in the OB1 PML, will return NOT_SUPPORTED in other PMLs.

This commit was SVN r25865.
 2012-02-06 17:35:21 +00:00
Mike Dubman
6188ab7317 * ep init refactoring 
* split ep_info into fragments to fit PMI limit

This commit was SVN r25857.
 2012-02-02 15:00:47 +00:00
Vishwanath Venkatesan
15ebe838e9 Modified implementation of two_phase read all similar to the changes for the write all 
incorporating romio style partitioning.

This commit was SVN r25853.
 2012-02-01 18:30:13 +00:00
Vishwanath Venkatesan
158374bdd0 Dynamic file write all algorithm optimized by using derived datatype for receiving actual data 
thereby avoiding the merging step in the fbtl.

This commit was SVN r25852.
 2012-02-01 18:20:44 +00:00
Vishwanath Venkatesan
b9026ccbd0 Fix for two-phase generating flattened datatype using decoded iovec for 
handling non-contiguous memory and contiguous file cases.

This commit was SVN r25850.
 2012-02-01 17:23:51 +00:00
Jeff Squyres
4f35b62154 Components should not be linking to top-level libraries -- and 
definitely should not be linking to more than libmpi.la!  (remember
that libmpi.la now wholly contains libopen-rte.la, which wholly
contains libopen-pal.la).

This commit was SVN r25843.
 2012-01-31 20:43:27 +00:00
Mike Dubman
92873872f5 revert r25813 
This commit was SVN r25816.

The following SVN revision numbers were found above:
  r25813 --> open-mpi/ompi@8ed781d7e9
 2012-01-30 13:22:38 +00:00
Mike Dubman
8ed781d7e9 add mca param to enable/disable mxm 
This commit was SVN r25813.
 2012-01-30 11:14:20 +00:00
Mike Dubman
9f0ca9dfc0 fix: extract source from imm request fields instead from depricated api 
This commit was SVN r25812.
 2012-01-30 10:37:37 +00:00
Ralph Castain
a0edae52f2 Ensure the wrapper flags get entered in the right order, with -lpmi coming before the alps util libs 
This commit was SVN r25809.
 2012-01-27 20:56:21 +00:00
Nathan Hjelm
97dad0ac49 ugni: don't release eager fragments until we get local smsg completion 
This commit was SVN r25796.
 2012-01-27 00:32:43 +00:00
Nathan Hjelm
669f0afd14 ugni: poll smsg mailbox until it is empty 
This commit was SVN r25794.
 2012-01-26 20:50:09 +00:00
Nathan Hjelm
2a83297f96 silence vader warnings 
This commit was SVN r25793.
 2012-01-26 20:07:33 +00:00
Mike Dubman
6c954ad43f set mxm to call opal_progress in tight loops 
This commit was SVN r25788.
 2012-01-26 18:33:43 +00:00
Nathan Hjelm
3d9bc68435 reorder udcm init/finalize code. fixes trac:2973 
This commit was SVN r25787.

The following Trac tickets were found above:
  Ticket 2973 --> https://svn.open-mpi.org/trac/ompi/ticket/2973
 2012-01-26 16:28:55 +00:00
Shiqing Fan
2c9a4beffd Add and remove a few components for windows build. 
This commit was SVN r25775.
 2012-01-25 09:01:27 +00:00
Nathan Hjelm
7b9bf6fe41 ugni: remove another erroneous error message 
This commit was SVN r25768.
 2012-01-23 21:23:01 +00:00
Nathan Hjelm
f3b60062cb ugni: remove erroneous error message 
This commit was SVN r25767.
 2012-01-23 21:05:24 +00:00
Nathan Hjelm
521546aaa3 bug fix: ugni: pack only as many bytes as the pml requested 
This commit was SVN r25766.
 2012-01-23 17:21:45 +00:00
Ralph Castain
f03b82ab0a Don't fiddle with the port_name memory as, per standard, this is an input-only parameter 
This commit was SVN r25756.
 2012-01-20 13:15:41 +00:00
Ralph Castain
be3dfb6a1a Ensure that we only add -lpmi once to the wrapper compilers, no matter how many components might use it. 
This commit was SVN r25753.
 2012-01-20 04:56:38 +00:00
Ralph Castain
f5c43e8d60 Get the header files into the tarball 
This commit was SVN r25746.
 2012-01-19 21:02:20 +00:00
Nathan Hjelm
804a494036 zero out ugni fragments in constructor 
This commit was SVN r25731.
 2012-01-17 19:52:26 +00:00
Vishwanath Venkatesan
1e95d8b1e2 remove the MPI functions used in these files by the OMPI internal corresponding functionality and also add error checking in these for functions which did not have them' 
This commit was SVN r25723.
 2012-01-13 17:21:51 +00:00
Rolf vandeVaart
16d676aa5b Fix minor issue with CUDA. Cannot register overlappiing regions. 
This commit was SVN r25716.
 2012-01-12 13:00:42 +00:00
Samuel Gutierrez
63869c431b init seg_num_procs_inited to zero before the atomic add. 
This commit was SVN r25710.
 2012-01-11 03:37:23 +00:00
Nathan Hjelm
96c1df8d90 clean up vader registration code 
This commit was SVN r25704.
 2012-01-10 22:33:22 +00:00
Edgar Gabriel
fb4d1a7099 remove the MPI functions used in this file by the OMPI internal corresponding functionality. 
This commit was SVN r25703.
 2012-01-10 19:55:05 +00:00
Nathan Hjelm
f65f6f5c39 bugfix: ugni: increase smsg mailbox size to a multiple of 4096 
This commit was SVN r25702.
 2012-01-10 19:50:25 +00:00
Mike Dubman
37dc53bbc9 mxm: return the MXM_REQ_SEND_SYNC flag to mxm_req_send 
This commit was SVN r25694.
 2012-01-06 18:56:28 +00:00
Mike Dubman
3b97d609a8 mtl_mxm: fix double free 
This commit was SVN r25693.
 2012-01-06 16:22:58 +00:00
Samuel Gutierrez
d1a44ecd34 send packed buffers instead of using iovecs in common sm rml. this commit will 
hopefully resolve the periodic bus errors that some mtt tests have been
encountering.

This commit was SVN r25692.
 2012-01-05 00:11:59 +00:00
Rolf vandeVaart
9441f33981 Improve an error message. Replace tabs with spaces. 
This commit was SVN r25688.
 2012-01-03 15:19:01 +00:00
Rolf vandeVaart
8073f5002a Some additional CUDA specific code. 
Adding a few more support functions that will be used in future development.

This commit was SVN r25684.
 2011-12-29 12:31:54 +00:00
Edgar Gabriel
e0139a2d7e provide descriptions about the functionality of these frameworks. 
This commit was SVN r25682.
 2011-12-22 19:42:00 +00:00
Vishwanath Venkatesan
0f928be1d5 Modifying selection logic back to select two-phase at the cases it should. 
This commit was SVN r25681.
 2011-12-22 01:01:32 +00:00
Vishwanath Venkatesan
37c8470e3d modified implementation for two-phase write_all incorporating romio style domain partitioning 
This commit was SVN r25680.
 2011-12-22 00:16:29 +00:00
Vishwanath Venkatesan
738a67b704 Removing duplicate code while setting default file view and using internal file-set-view for setting the default file view 
This commit was SVN r25679.
 2011-12-21 21:50:47 +00:00
Rolf vandeVaart
6ca186fb64 Delay some initialization until needed. This eliminates some 
warnings and removes need for CUDA init before MPI_Init.

This commit was SVN r25678.
 2011-12-21 15:21:57 +00:00
Samuel Gutierrez
519f71ab7e silences valgrind warning in common sm (Syscall param writev(vector[...]) points 
to uninitialised byte(s)).  probably also silences a large stack allocation
warning in coverity.

This commit was SVN r25666.
 2011-12-16 23:17:48 +00:00
Samuel Gutierrez
0ca6603fa0 remove some unused cruft in shmem. minor common sm cleanup. 
This commit was SVN r25665.
 2011-12-16 22:43:55 +00:00
Nathan Hjelm
71527c8058 minor ugni btl code cleanup 
This commit was SVN r25618.
 2011-12-10 08:20:46 +00:00
Nathan Hjelm
c8a4687402 don't set SIGSEGV to default 
This commit was SVN r25610.
 2011-12-09 21:54:05 +00:00
Nathan Hjelm
e03d23d96e Intial support for Cray's uGNI interface (XE-6/XK-6) 
This commit was SVN r25608.
 2011-12-09 21:24:07 +00:00
Nathan Hjelm
87b7e85d53 rfc timeout. retry registration after removing old registration from lru 
This commit was SVN r25587.
 2011-12-07 18:20:44 +00:00
Josh Hursey
e56b4de2c9 Fixes trac:2550 : Cleanup comment in crcp_bkmrk_pml.h 
This commit was SVN r25585.

The following Trac tickets were found above:
  Ticket 2550 --> https://svn.open-mpi.org/trac/ompi/ticket/2550
 2011-12-07 14:50:04 +00:00
Jeff Squyres
c10f41c87e Do not build these frameworks when --disable-mpi-io is specified. 
Fixes some Cisco MTT MPI install errors.

This commit was SVN r25566.
 2011-12-02 22:11:23 +00:00
Ralph Castain
07655e2945 Handle the case where the allocator "fibs" to us about the node names. In some cases (ahem...you know who you are!), the allocator will tell us a node number (e.g., "16"). However, the daemon will return a node name (e.g., "nid0016") - leaving us not recognizing its location. 
So provide a new parameter (can't have too many!) that handles this situation by stripping the prefix from the returned node name. Also do a little cleanup to ensure we cleanly exit from errors, without generating too many annoying messages.

This commit was SVN r25562.
 2011-12-02 14:10:08 +00:00
Ralph Castain
357ac14530 Can't return a numerical value here 
This commit was SVN r25559.
 2011-12-02 10:36:57 +00:00
Nathan Hjelm
bb1fec0407 added put/get btl descriptor flags 
This commit was SVN r25553.
 2011-11-30 21:37:23 +00:00
Ralph Castain
c56acf60ca Although we never really thought about it, we made an unconscious assumption in the mapper system - we assumed that the daemons would be placed on nodes in the order that the nodes appear in the allocation. In other words, we assumed that the launch environment would map processes in node order. 
Turns out, this isn't necessarily true. The Cray, for example, launches processes in a toroidal pattern, thus causing the daemons to wind up somewhere other than what we thought. Other environments (e.g., slurm) are also capable of such behavior, depending upon the default mapping algorithm they are told to use.

Resolve this problem by making the daemon-to-node assignment in the affected environments when the daemon calls back and tells us what node it is on. Order the nodes in the mapping list so they are in daemon-vpid order as opposed to the order in which they show in the allocation. For environments that don't exhibit this mapping behavior (e.g., rsh), this won't have any impact.

Also, clean up the vm launch procedure a little bit so it more closely aligns with the state machine implementation that is coming, and remove some lingering "slave" code.

This commit was SVN r25551.
 2011-11-30 19:58:24 +00:00
Jeff Squyres
6fbbfd0f7a Gah! r25545 acidentally included ''waaaay'' more stuff than it was 
supposed to.  I.e., half-baked/not complete stuff.

This commit backs out all of r25545.  Sorry folks!

This commit was SVN r25546.

The following SVN revision numbers were found above:
  r25545 --> open-mpi/ompi@7f9ae11faf
 2011-11-29 23:24:52 +00:00
Jeff Squyres
7f9ae11faf Per http://www.open-mpi.org/community/lists/users/2011/11/17862.php, 
to make MPI_IN_PLACE (and other sentinel Fortran constants) work on OS
X, we need to use the following compiler (linker) flag:

    -Wl,-commons,use_dylibs 

So if we're compiling on OS X, test to see if that flag works with the
compiler.  If so, add it to the wrapper FFLAGS and FCFLAGS (note that
per a future update, we'll only have one Fortran compiler anyway).

Fixes trac:1982.  

This commit was SVN r25545.

The following Trac tickets were found above:
  Ticket 1982 --> https://svn.open-mpi.org/trac/ompi/ticket/1982
 2011-11-29 23:05:54 +00:00
Terry Dontje
5209de048c add code to service_thread_start to handle EBADF returns from select. This commit fixes trac:2922. 
This commit was SVN r25520.

The following Trac tickets were found above:
  Ticket 2922 --> https://svn.open-mpi.org/trac/ompi/ticket/2922
 2011-11-29 16:49:59 +00:00
Samuel Gutierrez
375162c693 this commit fixes a few things. 1. silence warning in common sm. 2. remove unneeded config code in common sm. 3. move opal_shmem_base_close to a better place in opal_finalize. 4. fix opal_path_nfs output. 
This commit was SVN r25518.
 2011-11-28 23:41:19 +00:00
George Bosilca
0bd2bf9aae The number of segments accepted should be bounded by MCA_BTL_DES_MAX_SEGMENTS 
and not by 2.

This commit was SVN r25515.
 2011-11-28 17:19:12 +00:00
Nathan Hjelm
f8c8c641f1 added asserts to warn developers that ob1/csum match fragments do not support more than 2 segments 
This commit was SVN r25514.
 2011-11-28 16:12:25 +00:00
Samuel Gutierrez
b4edf0ff5c getting ready for 1.5 port of the shared memory enhancements. remove some unused/unneeded stuff and minor style update. 
This commit was SVN r25513.
 2011-11-28 16:08:32 +00:00
Ralph Castain
9b59d8de6f This is actually a much smaller commit than it appears at first glance - it just touches a lot of files. The --without-rte-support configuration option has never really been implemented completely. The option caused various objects not to be defined and conditionally compiled some base functions, but did nothing to prevent build of the component libraries. Unfortunately, since many of those components use objects covered by the option, it caused builds to break if those components were allowed to build. 
Brian dealt with this in the past by creating platform files and using "no-build" to block the components. This was clunky, but acceptable when only one organization was using that option. However, that number has now expanded to at least two more locations.

Accordingly, make --without-rte-support actually work by adding appropriate configury to prevent components from building when they shouldn't. While doing so, remove two frameworks (db and rmcast) that are no longer used as ORCM comes to a close (besides, they belonged in ORCM now anyway). Do some minor cleanups along the way.

This commit was SVN r25497.
 2011-11-22 21:24:35 +00:00
Ralph Castain
6310361532 At long last, the fabled revision to the affinity system has arrived. A more detailed explanation of how this all works will be presented here: 
https://svn.open-mpi.org/trac/ompi/wiki/ProcessPlacement

The wiki page is incomplete at the moment, but I hope to complete it over the next few days. I will provide updates on the devel list. As the wiki page states, the default and most commonly used options remain unchanged (except as noted below). New, esoteric and complex options have been added, but unless you are a true masochist, you are unlikely to use many of them beyond perhaps an initial curiosity-motivated experimentation.

In a nutshell, this commit revamps the map/rank/bind procedure to take into account topology info on the compute nodes. I have, for the most part, preserved the default behaviors, with three notable exceptions:

1. I have at long last bowed my head in submission to the system admin's of managed clusters. For years, they have complained about our default of allowing users to oversubscribe nodes - i.e., to run more processes on a node than allocated slots. Accordingly, I have modified the default behavior: if you are running off of hostfile/dash-host allocated nodes, then the default is to allow oversubscription. If you are running off of RM-allocated nodes, then the default is to NOT allow oversubscription. Flags to override these behaviors are provided, so this only affects the default behavior.

2. both cpus/rank and stride have been removed. The latter was demanded by those who didn't understand the purpose behind it - and I agreed as the users who requested it are no longer using it. The former was removed temporarily pending implementation.

3. vm launch is now the sole method for starting OMPI. It was just too darned hard to maintain multiple launch procedures - maybe someday, provided someone can demonstrate a reason to do so.

As Jeff stated, it is impossible to fully test a change of this size. I have tested it on Linux and Mac, covering all the default and simple options, singletons, and comm_spawn. That said, I'm sure others will find problems, so I'll be watching MTT results until this stabilizes.

This commit was SVN r25476.
 2011-11-15 03:40:11 +00:00
Jeff Squyres
e8dcad6017 This typo has been here since August 2005. :-) 
This commit was SVN r25468.
 2011-11-11 03:01:52 +00:00
Brian Barrett
45a27e4f9f For now, ignore LINK event 
This commit was SVN r25467.
 2011-11-11 02:49:03 +00:00
Brad Benton
96395c916e de-tab'd 
This commit was SVN r25465.
 2011-11-09 19:45:12 +00:00
Brad Benton
0712b911a5 Updated IBM copyright 
This commit was SVN r25464.
 2011-11-09 19:38:53 +00:00
Mike Dubman
00c27afd52 fix pid 
This commit was SVN r25463.
 2011-11-09 17:53:59 +00:00
Nathan Hjelm
d603f31976 removed ptr member from seg_key union 
This commit was SVN r25460.
 2011-11-08 15:44:54 +00:00
Mike Dubman
71398b658e fix: OMPI_ERR_CONNECTION_FAILED available in v1.5, unavailable in trunk 
This commit was SVN r25459.
 2011-11-08 12:34:01 +00:00
Mike Dubman
4cf9e1323d fix: return correct error on connection failure 
This commit was SVN r25452.
 2011-11-07 06:13:17 +00:00
Nathan Hjelm
8962ce25b0 fixed some compiler errors caused by seg_key changes. osc/rdma may need to be updated to use btls that use 128 bit segment keys 
This commit was SVN r25448.
 2011-11-06 20:19:14 +00:00
Samuel Gutierrez
e03bc93fb7 only use pmi grpcomm and pubsub during the direct launch case. use PMI environment variable to setup vpid in ess alps on cray xe systems. add pmi test code. 
This commit was SVN r25447.
 2011-11-06 17:28:40 +00:00
Nathan Hjelm
520a7c570e changes to seg_key needed for a new btl 
This commit was SVN r25445.
 2011-11-06 16:19:09 +00:00
Rolf vandeVaart
f777fe8eba Change tab to spaces. 
This commit was SVN r25433.
 2011-11-04 17:18:30 +00:00
Christopher Yeoh
fb57a74a40 Removes pointless memmove which because of a previous memcpy will always 
have identical source and destination pointers. See #2871
Plugs a couple of minor memory leaks related to remote qp info

This commit was SVN r25431.
 2011-11-04 00:15:08 +00:00
Mike Dubman
7595a80a63 fix self pid 
This commit was SVN r25424.
 2011-11-03 06:46:20 +00:00
Nathan Hjelm
211e2dbdf3 clean up tab characters 
This commit was SVN r25413.
 2011-11-02 15:07:57 +00:00
Ralph Castain
14966e0f8f Cleanup PMI startup - if a component isn't selected, it should finalize PMI IFF it started it. Otherwise, components that aren't selected can finalize PMI when it is in use by other parts of the system. 
This commit was SVN r25407.
 2011-11-01 16:25:12 +00:00
Mike Dubman
3edd77ea25 update mxm plugin to mxm api change: pass synchronous request as an opcode instead of a flag 
This commit was SVN r25403.
 2011-10-31 22:36:15 +00:00
Mike Dubman
6b50ba22a6 select mxm ptl based on user preferences 
This commit was SVN r25401.
 2011-10-31 10:17:43 +00:00
Samuel Gutierrez
0ba13e2f8e fix typo. use PMI_Initialized for init status instead of PMI_Init. 
This commit was SVN r25378.
 2011-10-27 22:41:50 +00:00
Nathan Hjelm
ee087de073 added fast boxes to vader 
This commit was SVN r25376.
 2011-10-27 20:22:46 +00:00
Mike Dubman
f96ae43e23 pass jobid to mxm/sm module 
This commit was SVN r25375.
 2011-10-27 13:14:52 +00:00
Nathan Hjelm
82efe131dc made btl_vader_max_inline_send a configurable parameter and updated and enabled sendi 
This commit was SVN r25374.
 2011-10-26 22:15:42 +00:00
Nathan Hjelm
033179d6ac fixed bug in frag initialization 
This commit was SVN r25373.
 2011-10-26 19:29:37 +00:00
George Bosilca
6fdb040eef ORTE_ERROR to OPAL_ERROR. 
This commit was SVN r25372.
 2011-10-26 15:59:43 +00:00
George Bosilca
9d8e84142f Survivor!!! 
This commit was SVN r25371.
 2011-10-26 00:58:55 +00:00
Nathan Hjelm
05114ffb51 fixed off by one error 
This commit was SVN r25369.
 2011-10-25 22:07:47 +00:00
George Bosilca
72f731f25f The SM2 collective component has not been updated in a long 
time. Rich, the original developer, agrees with this removal.

This commit was SVN r25368.
 2011-10-25 22:07:09 +00:00
Nathan Hjelm
e887d595c7 fix potential bug with non-contiguous sends 
This commit was SVN r25367.
 2011-10-25 19:21:45 +00:00
Nathan Hjelm
433cfa3665 use single copy for some sends 
This commit was SVN r25365.
 2011-10-25 18:38:42 +00:00
Mike Dubman
9ffeeb69d9 fix help message 
This commit was SVN r25364.
 2011-10-25 14:02:43 +00:00
Samuel Gutierrez
663f4546f5 fix define typo in psm mtl. 
This commit was SVN r25362.
 2011-10-24 18:38:12 +00:00
Ralph Castain
955d8e7d46 Allow apps to use pmi when launched by mpirun, if desired, without affecting daemons 
This commit was SVN r25359.
 2011-10-23 15:57:13 +00:00
Nathan Hjelm
fb19f56965 Cray doesn't define PMI2_SUCCESS 
This commit was SVN r25354.
 2011-10-21 16:34:22 +00:00
Nathan Hjelm
cd68dbe2b8 only try to build vader if xpmem is installed. unignore vader 
This commit was SVN r25352.
 2011-10-21 15:45:05 +00:00
Ralph Castain
3e72fccacf Cray's PMI implementation is quite different from slurm's - they extended PMI-1 by adding some, but not all, of the PMI-2 APIs. So you can't just switch to using PMI-2 functions as it isn't a complete implementation. Instead, you have to selectively figure out which ones they have in PMI-2, and use any missing ones from PMI-1. What fun. 
Modify the configure logic and the PMI components to accommodate Cray's approach. Refactor the PMI error reporting code so it resides in only one place. Cray actually decided -not- to define the PMI-2 error codes, so we have to use the PMI-1 codes instead. More fun.

This commit was SVN r25348.
 2011-10-21 04:54:38 +00:00
Ralph Castain
e2adc8fa3a Ignore until Nathan can fix - probably configure problem 
This commit was SVN r25347.
 2011-10-21 03:43:01 +00:00
Ralph Castain
5947f61b86 Remove windows reference for now 
This commit was SVN r25346.
 2011-10-21 01:19:03 +00:00
Nathan Hjelm
414677a082 default to no xpmem support 
This commit was SVN r25345.
 2011-10-20 22:13:45 +00:00
Nathan Hjelm
808a73a5c5 removed erroneous add of .deps 
This commit was SVN r25343.
 2011-10-20 21:41:51 +00:00
Nathan Hjelm
3dbaaf6879 initial commit of vader (xpmem) btl 
This commit was SVN r25342.
 2011-10-20 21:39:44 +00:00