- Update to 4.1.6:
- Update embedded PMIx to 3.2.5.
- Fix issue with buffered sends and MTL-based interfaces (Libfabric,
PSM, Portals).
- Add missing MPI_F_STATUS_SIZE to mpi.h.
- Update Fortran mpi module configure check to be more correct.
- Update to properly handle PMIx v>=4.2.3.
- Fix minor issues and add some minor performance optimizations with
OFI support.
- Support the "striping_factor" and "striping_unit" MPI_Info names
recomended by the MPI standard for parallel IO.
- Fixed some minor issues with UCX support.
- Minor optimization for 0-byte MPI_Alltoallw (i.e., make it a no-op).
OBS-URL: https://build.opensuse.org/request/show/1114836
OBS-URL: https://build.opensuse.org/package/show/science:HPC/openmpi4?expand=0&rev=28
- Update to 4.1.5:
- Fix crash in one -sided applications for certain process layouts.
- Update embedded OpenPMIx to version 3.2.4
- Backport patches to Libevent for CVE -2016 -10195, CVE -2016 -10196, and
CVE -2016 -10197. Note that Open MPI's internal libevent does not
use the impacted portions of the Libevent code base.
- SHMEM improvements:
- Fix initializer bugs in SHMEM interface.
- Fix unsigned type comparisons generating warnings.
- Fix use after clear issue in shmem_ds_reset.
- UCX improvements
- Fix memory registration bug that could occur when UCX was built
but not selected.
- Reduce overhead of add_procs with intercommunicators.
- Enable multi_send_nb by default.
- Call opal_progress while waiting for a UCX fence to complete.
- Fix data corruption bug in osc/rdma component.
- Fix overflow bug in alltoall collective
- Fix crash when displaying topology.
- Add some MPI_F_XXX constants that were missing from mpi.h.
- coll/ucc bug fixes.
OBS-URL: https://build.opensuse.org/request/show/1100770
OBS-URL: https://build.opensuse.org/package/show/science:HPC/openmpi4?expand=0&rev=24
- update to 4.1.4:
* Fix possible length integer overflow in numerous non-blocking collective
operations.
* Fix segmentation fault in UCX if MPI Tool interface is finalized before
MPI_Init is called.
* Remove /usr/bin/python dependency in configure.
* Fix OMPIO issue with long double etypes.
* Update treematch topology component to fix numerous correctness issues.
* Fix memory leak in UCX MCA parameter registration.
* Fix long operation closing file descriptors on non-Linux systems that
can appear as a hang to users.
* Fix for attribute handling on GCC 11 due to pointer aliasing.
* Fix multithreaded race in UCX PML's datatype handling.
* Fix a correctness issue in CUDA Reduce algorithm.
* Fix compilation issue with CUDA GPUDirect RDMA support.
* Fix to make shmem_calloc(..., 0) conform to the OpenSHMEM specification.
* Add UCC collectives component.
* Fix divide by zero issue in OMPI IO component.
* Fix compile issue with libnl when not in standard search locations.
* Fixed a seg fault in the smcuda BTL. Thanks to Moritz Kreutzer and
@Stadik for reporting the issue.
* Added support for ELEMENTAL to the MPI handle comparison functions
in the mpi_f08 module. Thanks to Salvatore Filippone for raising
the issue.
* Minor datatype performance improvements in the CUDA-based code paths.
* Fix MPI_ALLTOALLV when used with MPI_IN_PLACE.
* Fix MPI_BOTTOM handling for non-blocking collectives. Thanks to
Lisandro Dalcin for reporting the problem.
* Enable OPAL memory hooks by default for UCX.
* Many compiler warnings fixes, particularly for newer versions of
OBS-URL: https://build.opensuse.org/request/show/992639
OBS-URL: https://build.opensuse.org/package/show/science:HPC/openmpi4?expand=0&rev=19
- Update to version 4.1.1
- Fix a number of datatype issues, including an issue with
improper handling of partial datatypes that could lead to
an unexpected application failure.
- Change UCX PML to not warn about MPI_Request leaks during
MPI_FINALIZE by default. The old behavior can be restored with
the mca_pml_ucx_request_leak_check MCA parameter.
- Reverted temporary solution that worked around launch issues in
SLURM v20.11.{0,1,2}. SchedMD encourages users to avoid these
versions and to upgrade to v20.11.3 or newer.
- Updated PMIx to v3.2.2.
- Disabled gcc built-in atomics by default on aarch64 platforms.
- Disabled UCX PML when UCX v1.8.0 is detected. UCX version 1.8.0 has a bug that
may cause data corruption when its TCP transport is used in conjunction with
the shared memory transport. UCX versions prior to v1.8.0 are not affected by
this issue. Thanks to @ksiazekm for reporting the issue.
- Fixed detection of available UCX transports/devices to better inform PML
prioritization.
- Fixed SLURM support to mark ORTE daemons as non-MPI tasks.
- Improved AVX detection to more accurately detect supported
platforms. Also improved the generated AVX code, and switched to
using word-based MCA params for the op/avx component (vs. numeric
big flags).
- Improved OFI compatibility support and fixed memory leaks in error
handling paths.
- Improved HAN collectives with support for Barrier and Scatter. Thanks
to @EmmanuelBRELLE for these changes and the relevant bug fixes.
- Fixed MPI debugger support (i.e., the MPIR_Breakpoint() symbol).
Thanks to @louisespellacy-arm for reporting the issue.
- Fixed ORTE bug that prevented debuggers from reading MPIR_Proctable.
OBS-URL: https://build.opensuse.org/request/show/889006
OBS-URL: https://build.opensuse.org/package/show/science:HPC/openmpi4?expand=0&rev=15
- Update to version 4.1.0
* collectives: Add HAN and ADAPT adaptive collectives components.
Both components are off by default and can be enabled by specifying
"mpirun --mca coll_adapt_priority 100 --mca coll_han_priority 100 ...".
We intend to enable both by default in Open MPI 5.0.
* OMPIO is now the default for MPI-IO on all filesystems, including
Lustre (prior to this, ROMIO was the default for Lustre). Many
thanks to Mark Dixon for identifying MPI I/O issues and providing
access to Lustre systems for testing.
* Minor MPI one-sided RDMA performance improvements.
* Fix hcoll MPI_SCATTERV with MPI_IN_PLACE.
* Add AVX support for MPI collectives.
* Updates to mpirun(1) about "slots" and PE=x values.
* Fix buffer allocation for large environment variables. Thanks to
@zrss for reporting the issue.
* Upgrade the embedded OpenPMIx to v3.2.2.
* Fix issue with extra-long values in MCA files. Thanks to GitHub
user @zrss for bringing the issue to our attention.
* UCX: Fix zero-sized datatype transfers.
* Fix --cpu-list for non-uniform modes.
* Fix issue in PMIx callback caused by missing memory barrier on Arm platforms.
* OFI MTL: Various bug fixes.
* Fixed issue where MPI_TYPE_CREATE_RESIZED would create a datatype
with unexpected extent on oddly-aligned datatypes.
* collectives: Adjust default tuning thresholds for many collective
algorithms
* runtime: fix situation where rank-by argument does not work
* Portals4: Clean up error handling corner cases
* runtime: Remove --enable-install-libpmix option, which has not
worked since it was added
OBS-URL: https://build.opensuse.org/request/show/882166
OBS-URL: https://build.opensuse.org/package/show/science:HPC/openmpi4?expand=0&rev=12
