We are pleased to announce the alpha release of OSHMPI-2.1 (version 2.1a1) and the release candidate of OSHMPI-2.0 (version 2.0rc1). OSHMPI is a portable OpenSHMEM implementation on top of MPI.
Release tarballs are available at:
https://pmodels.github.io/oshmpi-www/downloads.html
### A list of new features in version 2.1a1:
- Support a prototype of memory space with CUDA memory kind.
- Enable MPI dynamic window based optimization.
- Add configure option to force inline performance critical routines.
### A list of new features in version 2.0rc1:
- Do not inline performance noncritical paths.
- Enable operation tracking to avoid unnecessary network flush.
- Enable datatype cache to reduce MPI derived datatype creation overhead at strided operations.
- Disable internal assertion when --enable-fast configure option is set.
- Update openpa git repository URL.
- Bug fixes:
+ Unsupported MPI datatype with complexd reduce calls.
+ Compile openpa only when multithread safety is set.
Details of each feature can be found in CHANGES (https://github.com/pmodels/oshmpi/blob/main/CHANGES) and README (https://github.com/pmodels/oshmpi/blob/main/README.md)
Regards,
The OSHMPI Team
A new release candidate of MPICH, 3.4rc1, is now available for download. This release previews the 3.4 major series. In the 3.4 series, ch4 replaces ch3 as the default device configuration. New features include GPU support, multi-VCI communication (experimental), XPMEM support, DAOS support in ROMIO (thanks to Intel contribution), and business card exchange enhancements.
Also in this release are several bug fixes and improvements in ch3, hydra, and common MPICH code. You can find the release on our downloads
page (https://www.mpich.org/downloads/)
Unless critical bugs are found, this release would be identical to the final 3.4 GA release (apart from the version number change).
Regards,
The MPICH team
===============================================================================
Changes in 3.4
===============================================================================
# ch4 replaces ch3 as the default device configuration.
# Add support for Yaksa datatype engine (default in ch4).
# Add support for GPU buffers (CUDA, Level Zero) in pt2pt,
collectives, and one-sided communication.
# Add support for XPMEM.
# Add support for multiple virtual communication interfaces for more
efficient MPI_THREAD_MULTIPLE (experimental).
# Add DAOS ADIO driver to ROMIO (contributed by Intel).
# Add Quobyte ADIO driver to ROMIO (contributed by Quobyte).
# Add support for Arm compiler toolchain
# Add support for flang/f18 Fortran compiler
# Add support for AddressSanitizer and UndefinedBehaviorSanitizer to
debug configuration
# Remove mxm, llc, and portals4 netmods from ch3.
# Remove support for logical reduction operations on floating point
types.
# Remove MPIX_Mutex interfaces.
# Further improvements to ch4 business card exchange: extra
long address support and fixes for PMIx integration.
# Un-inline non-critical ch4 code for improved build times.
# Fix several test program bugs.
# Fix several static analysis and compiler warnings.
# Change the signature of MPID_Init to include requested and provided
thread levels.
A full list of changes is available at the following link:
https://github.com/pmodels/mpich/compare/v3.3...v3.4rc1
