ports/openmpi
ports/openmpi
1
1
Форкнуть 0
Код Релизы Активность
6 765 коммитов 3 Ветки 0 Теги 130 MiB
1abe8ef368
Граф коммитов

50 Коммитов

Автор SHA1 Сообщение Дата
George Bosilca
83f83e5730 Specialize the MCA_PML_OB1_FREE macro. When we call this macro we already know what kind 
of request we are playing with (send or receive). Therefore, it's useless to have another
switch inside this macro and make the code bigger. Now, we have 2 versions
MCA_PML_OB1_SEND_REQUEST_FREE and MCA_PML_OB1_RECV_REQUEST_FREE.

This commit was SVN r8945.
 2006-02-08 22:42:00 +00:00
George Bosilca
600f664db2 No comment !!! But this time I really remove the second lock ... 
This commit was SVN r8943.
 2006-02-08 21:43:46 +00:00
George Bosilca
89db0be4a8 Remove the second lock. 
This commit was SVN r8940.
 2006-02-08 19:19:57 +00:00
Galen Shipman
18bbb049d1 fix bsend bug, need to check that data length before packing. 
This needs to go to the release branch

This commit was SVN r8937.
 2006-02-08 17:39:33 +00:00
George Bosilca
e9706e6db0 Remove the macro to compute the length of the segments from the send header 
and add a new macro that can be used for both sends and receives.
Move to atomic operations to manage the length of the sended or received
status. There is one instance where the atomic operation is not required
as the code can cannot be executed in same time by 2 differents threads.

This commit was SVN r8933.
 2006-02-08 06:03:54 +00:00
George Bosilca
a2f31b2bf6 Use the MCA_PML_BASE_REQUEST_MPI_COMPLETE macro to set the completed state to true. 
Add back the mutex protection around the size computations as they are not yet atomic
operations.

This commit was SVN r8812.
 2006-01-25 23:17:17 +00:00
Rainer Keller
7442a641ea - Fix locking, not related to the current deadlocks. 
This commit was SVN r8777.
 2006-01-21 13:43:36 +00:00
Tim Woodall
ebbac05f3c remove questionable sched_yield - probably left over from debugging 
This commit was SVN r8457.
 2005-12-12 16:00:27 +00:00
George Bosilca
c6eb429a9a Wondows work: 
- remove windows socket initialization (it's already in the TCP component)
 - protect all used header files
 - remove the unused ones.

This commit was SVN r8434.
 2005-12-10 21:38:48 +00:00
Tim Woodall
607f62accd - pass a flag to the peer indicating wether data is contiguous at the soure 
- only attempt to schedule rdma if contiguous at both src/dst
- need to review this for next release 

This commit was SVN r8119.
 2005-11-11 15:33:25 +00:00
Jeff Squyres
42ec26e640 Update the copyright notices for IU and UTK. 
This commit was SVN r7999.
 2005-11-05 19:57:48 +00:00
Tim Woodall
e45f4744ee do not return these descriptors to cache 
This commit was SVN r7986.
 2005-11-03 23:20:38 +00:00
Tim Woodall
26003bc952 fix from release branch - don't use get protocol if more 
than one btl is available

This commit was SVN r7984.
 2005-11-03 20:52:56 +00:00
George Bosilca
d916e0c5b4 The (I hope) final solution for the convertor problem. As all the PML inherit 
the base send and receive request from the pml_base, we can solve our problem
if we construct the convertor attached to any request in the pml_base_construct
function. At the end of the life time for each request (here life time is 
related to one utilisation, without taking in account the cache) we release
all information attached to the convertors in the _FINI macro by calling the
ompi_convertor_cleanup.

This commit was SVN r7910.
 2005-10-28 03:26:36 +00:00
Brian Barrett
bf67c9387b * initialize send request convertor with the correct type (convertor instead 
of request).  This fixes at least the bug with NetPIPE in 64bit land that
  Troy was seeing. 

This commit was SVN r7904.
 2005-10-27 23:08:27 +00:00
George Bosilca
1d75b7972f Solve thee problem with the reference count on the datatype (RT bug 1492). The problem is that the 
convertor (when prepared) increase the reference count on the used datatype. This reference count
will be released only when the OBJ_DESTRUCT is called on a convertor. However, having to call
OBJ_CONSTRUCT and OBJ_DESTRUCT on each request every time we want to use it (even when it come
from the cache) is an expensive operation. This can be avoided is the OBJ_DESTRUCT will leave the
convertor in exactly the same state as OBJ_CONSTRUCT. With this approach we just have to call
OBJ_CONSTRUCT for each convertor once when we initially create the request.

This commit was SVN r7813.
 2005-10-19 20:57:39 +00:00
Jeff Squyres
f9974f72e0 construct/destruct convertor when requests are 
constructed and allocated to free lists

This commit was SVN r7791.
 2005-10-18 12:19:43 +00:00
Tim Woodall
d859855dea merge in changes from 1.0 
This commit was SVN r7728.
 2005-10-12 15:54:35 +00:00
George Bosilca
f8355ec104 Cast the right side member to void* before assignment. 
This commit was SVN r7608.
 2005-10-04 12:37:23 +00:00
Tim Woodall
9b96ecf9b1 correction for rdma case - currently rdma entire message 
This commit was SVN r7392.
 2005-09-15 19:36:56 +00:00
Tim Woodall
6ae2ae4d1a - code cleanup 
- correct mpi semantics

This commit was SVN r7390.
 2005-09-15 18:47:59 +00:00
Tim Woodall
6ffa947a85 dont try and use convertor if not initialized 
This commit was SVN r7375.
 2005-09-14 22:03:52 +00:00
Tim Woodall
c25fb5dab0 - fixed issue w/ btl send-in-place option that was affecting tcp 
- reduced size of match header by an additional 4 bytes to 16 bytes
- corrections for buffered send (work in progress)

This commit was SVN r7371.
 2005-09-14 17:08:08 +00:00
Galen Shipman
b3db6fc2d1 Add code for mem_free hooks, commented out right now as there are performance 
issues. 
In PML-OB1 - fix for sync-sends of zero length. 

This commit was SVN r7362.
 2005-09-14 02:17:04 +00:00
Galen Shipman
d932cfd342 merge of rcache work into the trunk.. lotsa fun ;-).. 
I regression tested before the merge, I will regression test tonight and
correct issues that might have crept in. 

This commit was SVN r7329.
 2005-09-12 22:28:23 +00:00
Tim Woodall
7ccc43a052 correct for btls that don't support get 
This commit was SVN r7201.
 2005-09-06 19:01:35 +00:00
Tim Woodall
4fbf7965e3 check for rdma btls when leave pinned is set 
This commit was SVN r7151.
 2005-09-02 15:04:47 +00:00
Tim Woodall
47b41c7081 correct regression 
This commit was SVN r7098.
 2005-08-30 19:38:56 +00:00
Tim Woodall
47ee552220 can't mark request as complete until local completion as buffers point to user data 
This commit was SVN r7054.
 2005-08-26 20:18:16 +00:00
Tim Woodall
95bf77fc8d corrections for rdma pipeline 
This commit was SVN r6926.
 2005-08-18 17:06:35 +00:00
Tim Woodall
f274f524ab - added get based protocol (if supported by btl) for pre-registered memory 
- removed 8 bytes from the majority of the pml headers 

This commit was SVN r6916.
 2005-08-17 18:23:38 +00:00
Tim Woodall
adce576d5e corrections for eager send of data in place 
This commit was SVN r6892.
 2005-08-15 21:31:52 +00:00
Jeff Squyres
cf16a521c8 Ensure to get ompi/include/constants.h 
This commit was SVN r6845.
 2005-08-12 21:42:07 +00:00
Tim Woodall
0ab588e5d2 added support for sending eager messages directly from user buffer 
if supported by btl - btl must set MCA_BTL_FLAGS_SEND_INPLACE

This commit was SVN r6844.
 2005-08-12 21:33:01 +00:00
Tim Woodall
15c5fd6092 cleanup 
This commit was SVN r6838.
 2005-08-12 19:29:05 +00:00
Galen Shipman
9cbcdae14b more BML related stuff.. static_components.h is autogen'd configure.params is 
not. Compile warning fixed in pml_ob1_sendreq.c

This commit was SVN r6821.
 2005-08-12 14:40:06 +00:00
Galen Shipman
b01ebf45c9 Fixed build error related to direct call (bml_direct_call.h). Misc bug fixes 
and compiler warning issues. Fixed threaded build issue. 

This commit was SVN r6819.
 2005-08-12 14:08:40 +00:00
Galen Shipman
c3c83aa3e1 BML (BTL Managment Layer). Allows BTL's to be used outside of the PML. See 
bml.h and PML-OB1 for usage. 

This commit was SVN r6815.
 2005-08-12 02:41:14 +00:00
Brian Barrett
24116a3935 * fix up a bunch of threading issues when progress and/or mpi threads 
are enabled.  Mostly just ADD32 -> ADD_SIZE_T issues and naming of
  variables in THREAD_{LOCK,UNLOCK}

This commit was SVN r6706.
 2005-08-02 17:36:01 +00:00
Tim Woodall
2324f36065 call ob1 progress rather than endpoint - as it may not have one 
This commit was SVN r6696.
 2005-08-01 22:30:25 +00:00
Galen Shipman
168274db45 Call progress on send schedule so that prepare_src can run if the rdma ack 
lands between sending fragments up to rdma_offset. 

This commit was SVN r6646.
 2005-07-28 15:38:31 +00:00
Brian Barrett
6aa464b67e More changes from Red Storm port 
- only call sched_yield if it exists
  - don't fail out if modex doens't work in ob1
  - bunch of fixes for Portals BTL
  - add cnos rml component
  - add NULL gpr component (should only be used if replica AND proxy
    fail to load)  

This commit was SVN r6629.
 2005-07-27 23:07:14 +00:00
Tim Woodall
14ba3322e9 cleanup 
This commit was SVN r6546.
 2005-07-18 21:22:55 +00:00
Tim Woodall
0423d414ef - correction for sync send 
- now passing all of the intel p2p list

This commit was SVN r6543.
 2005-07-18 18:54:25 +00:00
Tim Woodall
025dcfa80b dont try and pack zero byte message 
This commit was SVN r6532.
 2005-07-15 20:58:11 +00:00
Tim Woodall
eabdb860bc tuning 
This commit was SVN r6370.
 2005-07-07 20:58:57 +00:00
Brian Barrett
a13166b500 * rename ompi_output to opal_output 
This commit was SVN r6329.
 2005-07-03 23:31:27 +00:00
Brian Barrett
39dbeeedfb * rename locking code from ompi to opal 
This commit was SVN r6327.
 2005-07-03 22:45:48 +00:00
Brian Barrett
761402f95f * rename ompi_list to opal_list 
This commit was SVN r6322.
 2005-07-03 16:22:16 +00:00
Jeff Squyres
4ab17f019b Rename src -> ompi 
This commit was SVN r6269.
 2005-07-02 13:43:57 +00:00