The 0.9.2 release includes the following new features:
- The neighbor alltoall exchanger is now expected to fully work on MPI's where MPI 3.x features are available, test it by setting environment variable
XT_CONFIG_DEFAULT_EXCHANGE_METHOD to neigh_alltoall
It also contains new work-arounds for the following compilers:
- gcc/gfortran 10 and 11, especially when using
-flto
- Intel Fortran 19.1 Update 2
- PGI/NVHPC compilers 20 and 21
Also a work-around for bugs in mpich 3.4.x releases is included, but the recommended fix is to apply the patches in source:config/checkpatch
Apart from that the release contains multiple measures to save memory and MPI resources.
This release
- reduces libtool compatibility headaches with the new nvfortran compiler,
- fixes a small memory leak when more than 8 redists are used to form a collection,
- fixes an incorrect computation of position extents when an xmap is constructed manually,
- hides internal library symbols,
- fixes the passing of zero size arrays in the Fortran interface,
- add minor improvements to the cfortran.h header.
Also see commit log at commit:f2eb603483f3b8041.
This release bring the following new features:
- configurable message pumping mechanism, programmers can select a so-called exchanger for new redist constructors with an additional argument or users/administrators by setting the XT_CONFIG_DEFAULT_EXCHANGE_METHOD environment variable
- by setting FC=no at configure time, it is possible to perform a C-language only build
The release also includes various minor improvements to the code, the build and test infrastructure, and reliability of build configuration.
Also see commit log at commit:4334d020c206da33.
This release is a minor update to 0.8.0 that
- fixes some minor build system issues,
- defaults to mpicc/mpifort and other usual wrappers for CC/FC instead of gcc/gfortran,
- adds inquiry functions xt_redist_get_num_send_msg and xt_redist_get_num_recv_msg,
- conditionalizes building of performance test programs,
- detects another failure in datatype handling in Intel MPI and MVAPICH at configure time,
- adds a constructor for xmaps (xt_xmap_intersection_pos_new) when the transferred positions are already known, and
- improves support for pgfortran 14.
Also see commit log at commit:c96c095bb3c6fa6.
This release contains several new/improved features:
- potential integer overflow in stripe overlap computation was fully eliminated
- interfaces (xt_xmap_reorder, xt_xmap_update_positions, xt_xmap_spread) to tune/extend the xmap message element sequence for improved memory access patterns
- make the build much cleaner
- improved support for recent PGI compiler releases
- much improved support for various Cray compiler versions
- tons of smaller tweaks/improvements/fixes, see revision log commit:652f8312be016820 for details
This is a bug-fix release that only
- contains a work-around for automake prior to 1.16 spurious file creation (no breakage)
- improves on build-workarounds for NAG Fortran
- improves libtool interaction with PGI compilers when invoked through a wrapper like mpicc
- adds a workaround for a Cray CCE code generation bug, where the compiler makes incorrect assumptions about pointer alignment.
This is a bug-fix release that only
- adds a file missing from the distribution archive (but is only needed when time-stamps get changed)
- fixes a problem when ifort or icc was used with the -diag-disable option having an argument as next command-line argument instead of separated by an equals sign
- fixes needed if autoreconf is performed with automake 1.16
The new YAXT release includes
- support for asynchronous exchanges (like MPI_Isend) and
- fixes for MPI implementations where MPI_Pack_size returns an actual upper bound and not the exact amount that MPI_Pack advances position by
- document the consequences segmented memory architectures might have
The following improvements have been added to YAXT:
- User-defined redists via the
xt_redist_single_array_base_new function
- Lots of minor cleanup and bug-fixing
The following improvements have been added to YAXT:
- Improved g95 support.
- A memory free bug when concatenating empty redists.
- A minor bug fix for stripe overlap computation when stripe starts are far apart and stride is large.
- New constructor xt_redist_repeat_asym_new allows for different displacements in source and destination arrays.
- Various minor fixes to tests.