The problem was caused by the static request optimization. The buffered send case
is much like the isend case in that the request structure may be needed after
MPI_Bsend completes. Fix this case by calling isend and freeing the resulting
request.
cmr=v1.7.5:ticket=trac:4149
This commit was SVN r30601.
The following Trac tickets were found above:
Ticket 4149 --> https://svn.open-mpi.org/trac/ompi/ticket/4149
them, but it's going to take a little time (at least one day). So
Nathan says it's ok to .ompi_ignore coll ml until he's able to fix it.
This commit was SVN r30600.
This change does not appear to increase the small message latency of ping-pong
benchmarks and fixes an issue found by our ibm datatype tests.
Fixes trac:4232
cmr=v1.7.5:ticket=trac:4149
This commit was SVN r30598.
The following Trac tickets were found above:
Ticket 4149 --> https://svn.open-mpi.org/trac/ompi/ticket/4149
Ticket 4232 --> https://svn.open-mpi.org/trac/ompi/ticket/4232
opal does not always define MB. It is recommended that opal_atomic_[rw]mb is
called instead. We will need to address the cases where these functions are
no-ops on weak-memory ordered cpus.
cmr=v1.7.5:ticket=trac:4158
This commit was SVN r30534.
The following Trac tickets were found above:
Ticket 4158 --> https://svn.open-mpi.org/trac/ompi/ticket/4158
Several changes are contained in this commit:
- Clean up tabs and trailing whitespaces
- Use consistent indentation in changed files
- Remove unused code. None of the removed code will ever have been
used in a trunk build.
- Clean up the smcm code quite a bit
- Do not fflush stderr and use opal_output instead of fprintf.
These changes have been tested on Cray XE-6 and PSM systems.
cmr=v1.7.5:ticket=trac:4158
This commit was SVN r30533.
The following Trac tickets were found above:
Ticket 4158 --> https://svn.open-mpi.org/trac/ompi/ticket/4158
During the commits to make the C/R code compile again the
blocking receive calls were replaced by non-blocking
which broke the code. This patch uses ORTE_WAIT_FOR_COMPLETION()
to wait until the non-blocking calls have finished.
This commit was SVN r30486.
This commit fixes one warning that should have caused coll/ml to segfault
on reduce. The fix should be correct but we will continue to investigate.
cmr=v1.7.5:ticket=trac:4158
This commit was SVN r30477.
The following Trac tickets were found above:
Ticket 4158 --> https://svn.open-mpi.org/trac/ompi/ticket/4158
for 32-bit architectures.
This commit also modifies _OMPI_CHECK_HEADER to use AC_CHECK_HEADERS instead
of AC_CHECK_HEADER. This allows components to check for multiple headers
instead of just one. The new semantics of the header check in OMPI_CHECK_PACKAGE
are to return success if at least one of the specified headers exists. The new
semantics will not break current usage.
cmr=v1.7.5:ticket=trac:4053
This commit was SVN r30476.
The following Trac tickets were found above:
Ticket 4053 --> https://svn.open-mpi.org/trac/ompi/ticket/4053
After IM with Nathan, apply patch from ticket after verification by Paul Hargrove that it fixes the problem on non-x86 32-bit platforms
Verified by Paul, RM-approved
cmr=v1.7.4:reviewer=ompi-gk1.7
This commit was SVN r30411.
The following Trac tickets were found above:
Ticket 4143 --> https://svn.open-mpi.org/trac/ompi/ticket/4143
ROMIO and Lustre 2.4.0. It has been solved upstream already; here's
the ticket:
http://trac.mpich.org/projects/mpich/ticket/1973
And here's the commit that fixed it:
a0c4278f14
OMPI does not have the other code referred to in that git commit (in
ad_lustre_hints.c).
Thanks to Adam Moody for reporting the issue.
cmr=v1.7.4:reviewer=hjelmn:subject=Fix ROMIO compile error w/ Lustre 2.4
This commit was SVN r30393.
btl sendi functions currently can not handle the descriptor being NULL. The
send inline optimization was assuming (incorrectly) that NULL was ok.
cmr=v1.7.5:ticket=trac:4149
This commit was SVN r30364.
The following Trac tickets were found above:
Ticket 4149 --> https://svn.open-mpi.org/trac/ompi/ticket/4149
allgather.
The new collectives provide a signifigant performance increase over tuned for
small and medium messages. We are initially setting the priority lower than
tuned until this has had some time to soak in the trunk. Please set
coll_ml_priority to 90 for MTT runs.
Credit for this work goes to Manjunath Gorentla Venkata (ORNL), Pavel Shamis (ORNL),
and Nathan Hjelm (LANL).
Commit details (for reference):
Import ORNL's collectives for MPI_Allreduce, MPI_Reduce, and MPI_Allgather.
We need to take the basesmuma header into account when calculating the
ptpcoll small message thresholds. Add a define to bcol.h indicating the
maximum header size so we can take the header into account while not
making ptpcoll dependent on information from basesmuma.
This resolves an issue with allreduce where ptpcoll overwrites the
header of the next buffer in the basesmuma bank.
Fix reduce and make a sequential collective launcher in coll_ml_inlines.h
The root calculation for reduce was wrong for any root != 0. There are
four possibilities for the root:
- The root is not the current process but is in the current hierarchy. In
this case the root is the index of the global root as specified in the
root vector.
- The root is not the current process and is not in the next level of the
hierarchy. In this case 0 must be the local root since this process will
never communicate with the real root.
- The root is not the current process but will be in next level of the
hierarchy. In this case the current process must be the root.
- I am the root. The root is my index.
Tested with IMB which rotates the root on every call to MPI_Reduce. Consider
IMB the reproducer for the issue this commit solves.
Make the bcast algorithm decision an enumerated variable
Resolve various asset failures when destructing coll ml requests.
Two issues:
- Always reset the request to be invalid before returning it to the
free list. This will avoid an asset in ompi_request_t's destructor.
OMPI_REQUEST_FINI does this (and also releases the fortran handle
index).
- Never explicitly construct or destruct the superclass of an opal
object. This screws up the class function tables and will cause
either an assert failure or a segmentation fault when destructing
coll ml requests.
Cleanup allgather.
I removed the duplicate non-blocking and blocking functions and modeled
the cleanup after what I found in allreduce. Also cleaned up the code
somewhat.
Don't bother copying from the send to the recieve buffer in
bcol_basesmuma_allreduce_intra_fanin_fanout if the pointers are the
same.
The eliminates a warning about memcpy and aliasing and avoids an
unnecessary call to memcpy.
Alwasy call CHECK_AND_RELEASE on memsync collectives.
There was a call to OBJ_RELEASE on the collective communicator but
because CHECK_AND_RECYLCE was never called there was not matching call
to OBJ_RELEASE. This caused coll ml to leak communicators.
Make allreduce use the sequential collective launcher in coll_ml_inlines.h
Just launch the next collective in the component progress.
I am a little unsure about this patch. There appears to be some sort
of race between collectives that causes buffer exhaustion in some cases
(IMB Allreduce is a reproducer). Changing progress to only launch the
next bcol seems to resolve the issue but might not be the best fix.
Note that I see little-no performance penalty for this change.
Fix allreduce when there are extra sources.
There was an issue with the buffer offset calculation when there are
extra sources. In the case of extra sources == 1 the offset was set
to buffer_size (just past the header of the next buffer). I adjusted
the buffer size to take into accoun the maximum header size (see the
earlier commit that added this) and simplified the offset calculation.
Make reduce/allreduce non-blocking. This is required for MPI_Comm_idup
to work correctly.
This has been tested with various layouts using the ibm testsuite and
imb and appears to have the same performance as the old blocking version.
Fix allgather for non-contiguous layouts and simplify parsing the
topology.
Some things in this patch:
- There were several comments to the effect that level 0 of the
hierarchy MUST contain all of the ranks. At least one function
made this assumption but it was not true. I changed the sbgp
components and the coll ml initization code to enforce this
requirement.
- Ensure that hierarchy level 0 has the ranks in the correct
scatter gather order. This removes the need for a separate
sort list and fixes the offset calculation for allgather.
- There were several passes over the hierarchy to determine
properties of the hierarchy. I eliminated these extra passes
and the memory allocation associated with them and calculate the
tree properties on the fly. The same DFS recursion also handles
the re-order of level 0.
All these changes have been verified with MPI_Allreduce, MPI_Reduce, and
MPI_Allgather. All functions now pass all IBM/Open MPI, and IMB tests.
coll/ml: correct pointer usage for MPI_BOTTOM
Since contiguous datatypes are copied via memcpy (bypassing the convertor) we
need to adjust for the lb of the datatype. This corrects problems found testing
code that uses MPI_BOTTOM (NULL) as the send pointer.
Add fallback collectives for allreduce and reduce.
cmr=v1.7.5:reviewer=pasha
This commit was SVN r30363.
Per RFC. There are two optimizations in this commit:
- Allocate requests for blocking sends and receives on the stack. This
bypasses the request free list and saves two atomics on the critical path.
This change improves the small message ping-pong by 50-200ns on both AMD
and Intel CPUs.
- For small messages try to use the btl sendi function before intializing a
send request. If the sendi fails or the btl does not have a sendi function
silently fallback on the standard send path.
cmr=v1.7.5:reviewer=brbarret
This commit was SVN r30343.
Adds coll_hcoll_np mca parameter similar to that of fca component (defaults to 32). Those who use hcoll be aware that from now on the communicators less than 32 procs will run w/o hcoll by default. - Resolves fallback issue in case libhcoll runs out of allowed contexts. The solution is moving hcoll_context_create from comm_enable to comm_query. Shortly, comm_enable should never return OMPI_ERROR in the coll component with highest priority (hcoll). Otherwise the ompi coll_base_select will unselect the coll funtion pointers and module references leaving the communicator w/o coll pointer. This will cause the fail. Same behavior can be reproduced even with tuned if one would hardcore some "return OMPI_ERROR" into it's module_enable funtion. - Additionally, removed all the dead code under #if 0; removed unused variables (path for library, active_modules list) and classes (module list wrapper)
Fixed by Val, Reviewed by Devendar/Josh/Miked
cmr=v1.7.4:reviewer=ompi-rm1.7
This commit was SVN r30341.
This commit fixes an error path that occurs when huge page allocations are
enabled. In this case we allocate a huge page and try to register it but fail.
We then were calling free on the opal object. Fix this by calling the proper free
function.
cmr=v1.7.4:reviewer=rhc
This commit was SVN r30289.
Also add a verbose flag so one can see what devices are selected as well as another flag to override
locality information and use all devices on the node.
This commit was SVN r30287.
NetBSD puts the AIO functions in -lrt, vs. the usual libc. So we
need the fbtl/posix configure.m4 to test for -lrt properly.
Reviewed by Jeff Squyres.
cmr=v1.7.4:reviewer=ompi-rm1.7:subject=Fix NetBSD use of -laio
This commit was SVN r30274.
NOTE: launch performance will be absolutely awful if you do this with BTLs that aren't configured to modex_recv on first message!
Even with "modex on demand", we still have to do a barrier in place of the modex - we simply don't move any data around, which does reduce the time impact. The barrier is required to ensure that the other proc has in fact registered all its BTL info and therefore is prepared to hand over a complete data package. Otherwise, you may not get the info you need. In addition, the shared memory BTL can fail to properly rendezvous as it expects the barrier to be in place.
This behavior will *only* take effect under the following conditions:
1. launched via mpirun
2. #procs is greater than ompi_hostname_cutoff, which defaults to UINT32_MAX
3. mca param rte_orte_direct_modex is set to 1. At the moment, we are having problems getting this param to register properly, so only the first two conditions are in effect. Still, the bottom line is you have to *want* this behavior to get it.
The planned next evolution of this will be to make the direct modex be non-blocking - this will require two fixes:
1. if the remote proc doesn't have the required info, then let it delay its response until it does. This means we need a way for the MPI layer to tell the RTE "I am done entering modex data".
2. adjust the SM rendezvous logic to loop until the required file has been created
Creating a placeholder to bring this over to 1.7.5 when ready.
cmr=v1.7.5:reviewer=hjelmn:subject=Enable direct modex at scale
This commit was SVN r30259.
