A new stable release of MPICH, 4.1, is now available for download. This is the first stable release in the 4.1.x series. Major features in this release include:
* Standalone libpmi
* Independent use of MPICH testsuite
* Experimental MPIX stream prototypes to support explicit threading context and asynchronous enqueue to GPU streams. Both CUDA and HIP are supported.
* Significant refactor in PMI and Hydra
You can find the release on our downloads page (https://www.mpich.org/downloads/)
Regards,
The MPICH team
===============================================================================
Changes in 4.1
===============================================================================
# Thread-cs in ch4 changed to per-vci.
# Testsuite (test/mpi) is configured separately from mpich configure.
# Added options in autogen to accelerate CI builds, including using pre-built
sub-modules. Added -yaksa-depth option to generate shallower yaksa pup code
for faster build and smaller binaries.
# Support singleton init using hydra.
# On OSX, link option flat_namespace is no longer turned on by default.
# Generate mpi.mod Fortran interfaces using Python 3. For many compilers,
including gfortran, flags such as -fallow-mismatched-args is no longer
necessary.
# Fixed message queue debugger interface in ch4.
# PMI (src/pmi) is refactored as a subdir and can be separately distributed.
# Added MPIX_Comm_get_failed.
# Experimental MPIX stream API to enable explicit thread contexts.
# Experimental MPIX gpu enqueue API. It currently only supports CUDA streams.
# Delays GPU resource allocation in yaksa.
# CH3 nemesis ofi netmod is removed.
# New collective algorithms. All collective algorithms are listed in
src/mpi/coll/coll_algorithms.txt
# Removed hydra2. We will port unique features of hydra2, including
tree-launching, to hydra in the future release.
# Added in-repository wiki documentation.
# Added stream workq to support optimizations for enqueue operations.
# Better support for large count APIs by eliminating type conversion issues.
# Hydra now uses libpmi (src/pmi) for handling PMI messages.
# Many bug fixes and enhancements.
A full list of changes since 4.0.3 can be found here: https://github.com/pmodels/mpich/compare/v4.0.3...v4.1
A new release candidate of MPICH, 4.1rc4, is now available for download. This release contains fixes for shared memory initialization and MPI_Allgatherv calls on intracommunicators of size=1.
As a reminder, only critical bugs will be fixed before the 4.1 GA release. Please give it a try and report any issues!
You can find the release on our downloads page (https://www.mpich.org/downloads/)
Regards,
The MPICH team
===============================================================================
Changes in 4.1
===============================================================================
# Thread-cs in ch4 changed to per-vci.
# Testsuite (test/mpi) is configured separately from mpich configure.
# Added options in autogen to accelerate CI builds, including using pre-built
sub-modules. Added -yaksa-depth option to generate shallower yaksa pup code
for faster build and smaller binaries.
# Support singleton init using hydra.
# On OSX, link option flat_namespace is no longer turned on by default.
# Generate mpi.mod Fortran interfaces using Python 3. For many compilers,
including gfortran, flags such as -fallow-mismatched-args is no longer
necessary.
# Fixed message queue debugger interface in ch4.
# PMI (src/pmi) is refactored as a subdir and can be separately distributed.
# Added MPIX_Comm_get_failed.
# Experimental MPIX stream API to enable explicit thread contexts.
# Experimental MPIX gpu enqueue API. It currently only supports CUDA streams.
# Delays GPU resource allocation in yaksa.
# CH3 nemesis ofi netmod is removed.
# New collective algorithms. All collective algorithms are listed in
src/mpi/coll/coll_algorithms.txt
# Removed hydra2. We will port unique features of hydra2, including
tree-launching, to hydra in the future release.
# Added in-repository wiki documentation.
# Added stream workq to support optimizations for enqueue operations.
# Better support for large count APIs by eliminating type conversion issues.
# Hydra now uses libpmi (src/pmi) for handling PMI messages.
# Many bug fixes and enhancements.
A full list of changes since 4.0.3 can be found here: https://github.com/pmodels/mpich/compare/v4.0.3...v4.1rc4
A new release candidate of MPICH, 4.1rc3, is now available for download. This release contains a fix for the MPIR_proctable implementation in Hydra. Thanks to the DDT folks for the report.
At this stage, only critical bugs will be fixed before the 4.1 GA release. Please give it a try!
You can find the release on our downloads page (https://www.mpich.org/downloads/)
Regards,
The MPICH team
===============================================================================
Changes in 4.1
===============================================================================
# Thread-cs in ch4 changed to per-vci.
# Testsuite (test/mpi) is configured separately from mpich configure.
# Added options in autogen to accelerate CI builds, including using pre-built
sub-modules. Added -yaksa-depth option to generate shallower yaksa pup code
for faster build and smaller binaries.
# Support singleton init using hydra.
# On OSX, link option flat_namespace is no longer turned on by default.
# Generate mpi.mod Fortran interfaces using Python 3. For many compilers,
including gfortran, flags such as -fallow-mismatched-args is no longer
necessary.
# Fixed message queue debugger interface in ch4.
# PMI (src/pmi) is refactored as a subdir and can be separately distributed.
# Added MPIX_Comm_get_failed.
# Experimental MPIX stream API to enable explicit thread contexts.
# Experimental MPIX gpu enqueue API. It currently only supports CUDA streams.
# Delays GPU resource allocation in yaksa.
# CH3 nemesis ofi netmod is removed.
# New collective algorithms. All collective algorithms are listed in
src/mpi/coll/coll_algorithms.txt
# Removed hydra2. We will port unique features of hydra2, including
tree-launching, to hydra in the future release.
# Added in-repository wiki documentation.
# Added stream workq to support optimizations for enqueue operations.
# Better support for large count APIs by eliminating type conversion issues.
# Hydra now uses libpmi (src/pmi) for handling PMI messages.
# Many bug fixes and enhancements.
A full list of changes since 4.0.3 can be found here: https://github.com/pmodels/mpich/compare/v4.0.3...v4.1rc3
A new release candidate of MPICH, 4.1rc2, is now available for download. This release contains fixes for a few user reported issues, as well as improvements to the release packaging process.
At this stage, only critical bugs will be fixed before the 4.1 GA release. Please give it a try!
You can find the release on our downloads page (https://www.mpich.org/downloads/)
Regards,
The MPICH team
===============================================================================
Changes in 4.1
===============================================================================
# Thread-cs in ch4 changed to per-vci.
# Testsuite (test/mpi) is configured separately from mpich configure.
# Added options in autogen to accelerate CI builds, including using pre-built
sub-modules. Added -yaksa-depth option to generate shallower yaksa pup code
for faster build and smaller binaries.
# Support singleton init using hydra.
# On OSX, link option flat_namespace is no longer turned on by default.
# Generate mpi.mod Fortran interfaces using Python 3. For many compilers,
including gfortran, flags such as -fallow-mismatched-args is no longer
necessary.
# Fixed message queue debugger interface in ch4.
# PMI (src/pmi) is refactored as a subdir and can be separately distributed.
# Added MPIX_Comm_get_failed.
# Experimental MPIX stream API to enable explicit thread contexts.
# Experimental MPIX gpu enqueue API. It currently only supports CUDA streams.
# Delays GPU resource allocation in yaksa.
# CH3 nemesis ofi netmod is removed.
# New collective algorithms. All collective algorithms are listed in
src/mpi/coll/coll_algorithms.txt
# Removed hydra2. We will port unique features of hydra2, including
tree-launching, to hydra in the future release.
# Added in-repository wiki documentation.
# Added stream workq to support optimizations for enqueue operations.
# Better support for large count APIs by eliminating type conversion issues.
# Hydra now uses libpmi (src/pmi) for handling PMI messages.
# Many bug fixes and enhancements.
A full list of changes since 4.0.3 can be found here: https://github.com/pmodels/mpich/compare/v4.0.3...v4.1rc2
