A new preview release of MPICH, 3.3a2, is now available for download.
This preview release is for the 3.3 major release series. Of major focus
in this alpha is an new (non-default) device layer implementation - CH4.
CH4 is designed for low software overheads to better exploit
next-generation hardware. An OFI (http://libfabric.org) or UCX
(http://openucx.org) library is required to build CH4. Example configure
lines:
./configure --with-device=ch4:ofi
--with-libfabric=<path/to/ofi/install>
./configure --with-device=ch4:ucx --with-ucx=<path/to/ucx/install>
CH4 is still in alpha stages, meaning there are known build issues and
bugs, but most tests and common benchmarks will complete on 64-bit Linux
systems. Since 3.3a1, there have been major stability improvements, bug
fixes, and code cleanup in CH4. CH4 support for Blue Gene/Q is added in
this release (via the OFI/BGQ provider), as well as Mac OS builds.
Also in this release are several bug fixes and improvements to CH3,
Hydra, and common MPICH code. You can find the release on our downloads
page (http://www.mpich.org/downloads)
Regards,
The MPICH team
===============================================================================
Changes in 3.3
===============================================================================
# CH4 Device: A new device layer implementation designed for low
software
overheads. CH4 has experimental support for OFI and UCX network
libraries,
and POSIX shared memory. Thanks to Intel, Mellanox, and RIKEN AICS for
participating in the CH4 coding effort.
# Removed the PAMI device and poe PMI client.
# Several other minor bug fixes, memory leak fixes, and code cleanup.
A full list of changes is available at the following link:
http://git.mpich.org/mpich.git/shortlog/v3.2..v3.3a2
A full list of bugs that have been fixed is available at the
following link:
https://github.com/pmodels/mpich/milestone/25
We are proud the announce the first preview release of Casper (1.0b1).
Please visit the Casper homepage for download.
http://www.mcs.anl.gov/project/casper/
Casper is a process-based asynchronous progress model for MPI one-sided
communication. It allows the user to keep aside a small number of cores
on a multicore or many-core environment as "ghost processes," which are
dedicated to help asynchronous progress for user processes through
appropriate memory mapping from those user processes. Casper is designed
as an external library through the PMPI name-shifted profiling interface
of MPI, it can transparently link with most MPI implementation that
supports MPI-3.0 RMA features.
Best Regards,
The MPICH team
Preview releases of Argobots (1.0a1) and BOLT (1.0a1) are now available for download here:
http://www.argobots.org/downloads/http://www.bolt-omp.org/downloads/
Argobots is a lightweight low-level threading and tasking infrastructure and can be used as a low-level threading mechanism for diverse high-level programming models or domain-specific libraries.
BOLT is a high-performing OpenMP implementation that utilizes Argobots as the primary threading model. With the ABI compatibility with the Intel OpenMP runtime, BOLT can be used with LLVM/Clang, Intel OpenMP compiler, or GCC with a simple LD_PRELOAD environment change (no recompilation needed).
Best Regards,
The MPICH team
