This code is the implementation of Software-base Performance Counters as described in the paper 'Using Software-Base Performance Counters to Expose Low-Level Open MPI Performance Information' in EuroMPI/USA '17 (http://icl.cs.utk.edu/news_pub/submissions/software-performance-counters.pdf). More practical usage information can be found here: https://github.com/davideberius/ompi/wiki/How-to-Use-Software-Based-Performance-Counters-(SPCs)-in-Open-MPI.
All software events functions are put in macros that become no-ops when SOFTWARE_EVENTS_ENABLE is not defined. The internal timer units have been changed to cycles to avoid division operations which was a large source of overhead as discussed in the paper. Added a --with-spc configure option to enable SPCs in the Open MPI build. This defines SOFTWARE_EVENTS_ENABLE. Added an MCA parameter, mpi_spc_enable, for turning on specific counters. Added an MCA parameter, mpi_spc_dump_enabled, for turning on and off dumping SPC counters in MPI_Finalize. Added an SPC test and example.
Signed-off-by: David Eberius <deberius@vols.utk.edu>
Per MPI-3.1:8.7.1 p361:11-13, it's valid for MPI_FINALIZED to be
invoked during an attribute destruction callback (e.g., during the
destruction of keyvals on MPI_COMM_SELF during the very beginning of
MPI_FINALIZE). In such cases, MPI_FINALIZED must return "false".
Prior to this commit, we hung in FINALIZED if it were invoked during
a COMM_SELF attribute destruction callback in FINALIZE. See
https://github.com/open-mpi/ompi/issues/5084.
This commit converts the MPI_INITIALIZED / MPI_FINALIZED
infrastructure to use a single enum (ompi_mpi_state, set atomically)
to represent the state of MPI:
- not initialized
- init started
- init completed
- finalize started
- finalize past COMM_SELF destruction
- finalize completed
The "finalize past COMM_SELF destruction" state is what allows us to
return "false" from MPI_FINALIZED before COMM_SELF has been fully
destroyed / all attribute callbacks have been invoked.
Since this state is checked at nearly every MPI API call (to see if
we're outside of the INIT/FINALIZE epoch), care was taken to use
atomics to *set* the ompi_mpi_state value in ompi_mpi_init() and
ompi_mpi_finalize(), but performance-critical code paths can simply
read the variable without needing to use a slow call to an
opal_atomic_*() function.
Thanks to @AndrewGaspar for reporting the issue.
Signed-off-by: Jeff Squyres <jsquyres@cisco.com>
This commit fixes#4795
- Fixed typo that sometimes causes deadlock in change of protocol.
- Redesigned out of sequence ordering and address the overflow case of
sequence number from uint16_t.
Signed-off-by: Thananon Patinyasakdikul <tpatinya@utk.edu>
This commit fixes a flaw in the eager limit check in pml/ob1. The
check was incorrectly checking if RDMA-only BTLs (BTLs without the
send flag) has a valid eager limit. This commit fixes the check by
adding an additional check for the send flag on the BTL module.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
The osc monitoring component needed to include other OSC components
header in order to be able tu access communicator through the
component specific ompi_osc_*_module_t structures. This commit remove
the dependency, and resolve the issue #4523.
Extend the common monitoring API.
* Now it's possible to translate from local rank to world rank from
both the communicator and the group.
* Remove useless hashtable as we directly use the w_group contained
in window structure.
Add automatic generation at config time.
The templates are expanded at configure time. It creates a new header
file that generates all the variables/functions needed. Adding this
during the autogen automagicaly generates for each of the available
modules the proper functions.
Only keep a generated argv-style array.
Following Jeff's advice, the configure.m4 file generate a simple array
of module variables to be iterated over to find the proper module.
Signed-off-by: Clement Foyer <clement.foyer@inria.fr>
This commit renames the arithmetic atomic operations in opal to
indicate that they return the new value not the old value. This naming
differentiates these routines from new functions that return the old
value.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
Turns out there are edge cases where an MTL's isend
method may end up marking a send request complete prior
to returning to the CM code. The would end up causing
problems in the bsend path since the ompi_request_complete
would end up getting invoked a second time on this request.
This ended up causing segfaults, etc. in ompi_request_complete .
Signed-off-by: Howard Pritchard <howardp@lanl.gov>
Rework the logic to handle the out-of-sequence fragments on the receiver
side. A large number of OOS messages are still arriving even in single
threaded scenarios.
Signed-off-by: George Bosilca <bosilca@icl.utk.edu>
If not the pvars will remain valid after the OB1 PML is unloaded, and
any access will segfault (the callbacks associated with the pvar will
point to the memory of the dlclosed module).
Signed-off-by: George Bosilca <bosilca@icl.utk.edu>
In multithreaded case, it is expensive to release the lock, call the slow match
and retake the lock again just to queue the frag. This patch will eliminate number of
lock taken by queueing the frag right away and return.
Signed-off-by: Thananon Patinyasakdikul <tpatinya@utk.edu>
If Open MPI is configured with CUDA, then user also should be using a CUDA build of
PSM2 and therefore be setting PSM2_CUDA environment variable to 1 while using
CUDA buffers for transfers. If we detect this setting to be missing, force set
it. If user wants to use this build for regular (Host buffer) transfers, we
allow the option of setting PSM2_CUDA=0, but print a warning
message to user that it is not a recommended usage scenario.
Signed-off-by: Aravind Gopalakrishnan <Aravind.Gopalakrishnan@intel.com>
PSM2 enables support for GPU buffers and CUDA managed memory and it can
directly recognize GPU buffers, handle copies between HFIs and GPUs.
Therefore, it is not required for OMPI to handle GPU buffers for pt2pt cases.
In this patch, we allow the PSM2 MTL to specify when
it does not require CUDA convertor support. This allows us to skip CUDA
convertor init phases and lets PSM2 handle the memory transfers.
This translates to improvements in latency.
The patch enables blocking collectives and workloads with GPU contiguous,
GPU non-contiguous memory.
Signed-off-by: Aravind Gopalakrishnan <Aravind.Gopalakrishnan@intel.com>
they are supposed to be unsigned, casting them to a signed
value for all atomic operations is as errorprone as handling
them as signed entities.
Signed-off-by: George Bosilca <bosilca@icl.utk.edu>
* Resolves#3705
* Components should link against the project level library to better
support `dlopen` with `RTLD_LOCAL`.
* Extend the `mca_FRAMEWORK_COMPONENT_la_LIBADD` in the `Makefile.am`
with the appropriate project level library:
```
MCA components in ompi/
$(top_builddir)/ompi/lib@OMPI_LIBMPI_NAME@.la
MCA components in orte/
$(top_builddir)/orte/lib@ORTE_LIB_PREFIX@open-rte.la
MCA components in opal/
$(top_builddir)/opal/lib@OPAL_LIB_PREFIX@open-pal.la
MCA components in oshmem/
$(top_builddir)/oshmem/liboshmem.la"
```
Note: The changes in this commit were automated by the script in
the commit that proceeds it with the `libadd_mca_comp_update.py`
script. Some components were not included in this change because
they are statically built only.
Signed-off-by: Joshua Hursey <jhursey@us.ibm.com>
