ROMIO

MPICH-3.2 came out a few days before SC 2015. It’s been over a year since MPICH-3.1.4. Here are the ROMIO highlights from the release:

• more improvements with IBM-PE compatibility
• more bug fixes for Blue Gene specific optimizations
• non-blocking collective I/O implemented
• fixed an odd bug where on some file systems a file created with MPI_MODE_EXCLUSIVE would fail
• bug fixes in the Panasas PanFS driver
• better manage some odd generated types the HDF5 folks found

releases

The MPICH team released MPICH-3.1.4 , and with it some ROMIO bug fixes.  Not much in the way of new features:

• continue to improve handing of bogus arguments
• several GPFS and Blue Gene optimizations
• continued improvements with IBM PE compatibility
• picked up a few build-related changes from OpenMPI
• worked around a strange MPICH type handling bug related to zero-length struct members

releases

The MPICH team released 3.1.3 on October 8th.  Here are the main new ROMIO-related things in this release (not everything: just the highlights):
– Fix for an unfortunate bug in which I could end up storing too big a value into an Info parameter, causing a crash at larger core counts.
– The MPI-3 type HINDEXED_BLOCK was only nominally supported, and in a buggy way at that.  now HINDEXED_BLOCK will work.
– some fixes for very large file accesses
– improvements to the “deferred open” optimization — now defers the open call in more situations
– a small reduction in number of system calls used to process “systemwide hints”
– some improvements to Blue Gene specific routines.

releases