philipp/hdf5
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

[svn-r5660] Purpose:

Browse Source 
Code optimization

Description:
    Avoid creating MPI types (and thus requiring a MPI_File_set_view() call)
    when contiguous selections are used for dataset I/O.  This should be a
    performance improvement for those sorts of selections.

Platforms tested:
    Linux 2.2.x (eirene) w/parallel && IRIX64 6.5 (modi4) w/parallel & FORTRAN
This commit is contained in:
Quincey Koziol
2002-06-18 09:02:17 -05:00
parent ad9ca47cec
commit 972707dcd3
4 changed files with 242 additions and 130 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
release_docs/RELEASE.txt
View File

@@ -181,11 +181,15 @@ Documentation
New Features
============
* Changed MPI I/O routines to avoid creating MPI derived types (and thus
needing to set the file view) for contiguous selections within datasets,
which should result in some performance improvement for those types of
selections. QAK - 2002/06/18
* Enable MPI type support for collective I/O to be enabled by default.
This can be disabled by setting the HDF5_MPI_OPT_TYPES environment
variable to the value "0". QAK - 2002/06/14
* Allow chunks in chunked datasets to be cached when file is opened for
read-only access (bug #709). QAK - 2002/06/10
* Allow chunks in chunked datasets to be cached when parallel file is
opened for read-only access (bug #709). QAK - 2002/06/10
* Added internal "small data" aggregation, which can reduce the number of
actual I/O calls made, improving performance. QAK - 2002/06/05
* Improved internal metadata aggregation, which can reduce the number of