diff options
author | Mike McGreevy <mamcgree@hdfgroup.org> | 2011-07-14 20:33:21 (GMT) |
---|---|---|
committer | Mike McGreevy <mamcgree@hdfgroup.org> | 2011-07-14 20:33:21 (GMT) |
commit | dc9288a44706fa89558d218325f34dbfdad31d07 (patch) | |
tree | 04fd24b7f047af98763c5b4b8a830622312b43fd /config | |
parent | 3397a114ced762dd088c674ff9bca5a950428927 (diff) | |
download | hdf5-dc9288a44706fa89558d218325f34dbfdad31d07.zip hdf5-dc9288a44706fa89558d218325f34dbfdad31d07.tar.gz hdf5-dc9288a44706fa89558d218325f34dbfdad31d07.tar.bz2 |
[svn-r21101] HDFFV-7639
Purpose:
Remove H5_MPI_SPECIAL_COLLECTIVE_IO_WORKS and
H5_MPI_COMPLEX_DERIVED_DATATYPE_WORKS #defines from source.
Description:
Two advanced parallel functionalities, special collective IO and
complex derived datatypes, are not supported by older
implementations of mpi, and thus our code limits the use of these
features with #ifdefs and has checks in configure to set them (or
not). Unfortunately, configure can't actually run a parallel check
to see if these features are working (nor not) so it resorts to
looking in the config files where they are explicity enabled or
disabled based on versions of mpi, sytems being built on, or for
no documented reason at all (i.e. just set to on or off as some
'default'). Overriding these settings is easy if need be, provided
it is known that it needs to be done to get improved performance,
and oftentimes it is not.
Most new MPI implementations successfully handle the functionality
requested when these #defines are set, and many of the "turn these
features off" cases in the config files are for old (> 5 years)
versions of MPI and retired systems (such as NCSA's tungsten).
Therefore, the decision has been made to remove the support for
these old versions of MPI and systems that cannot handle these
behaviors. The #ifdefs and supporting setup in the config/ files
and configure script has been removed, and the code executed when
these options were not set removed from the source.
In passing, this commit also cleans up some whitespace issues in
both t_mpi.c and H5Dmpio.c. Furthermore, in t_mpi.c, the special
collective IO test was not getting regularly run due to it being
written to work only with four processes (we regularly test with
six, previously with three), and thus it failed when actually run
due to an out of bounds data buffer assignment. It has been
modified to run at any number of processes greater than four, and
the memory problem has been fixed so the test passes.
Tested:
jam, h5committest, ember
Diffstat (limited to 'config')
-rw-r--r-- | config/cmake/H5pubconf.h.in | 7 | ||||
-rw-r--r-- | config/ia64-linux-gnu | 9 | ||||
-rw-r--r-- | config/ibm-aix | 5 | ||||
-rw-r--r-- | config/irix6.x | 20 | ||||
-rw-r--r-- | config/linux-gnulibc1 | 17 | ||||
-rw-r--r-- | config/mpich | 83 | ||||
-rw-r--r-- | config/x86_64-redstorm-linux-gnu | 2 |
7 files changed, 0 insertions, 143 deletions
diff --git a/config/cmake/H5pubconf.h.in b/config/cmake/H5pubconf.h.in index 8b8bf30..ae2e1a0 100644 --- a/config/cmake/H5pubconf.h.in +++ b/config/cmake/H5pubconf.h.in @@ -514,17 +514,10 @@ /* Define if the metadata trace file code is to be compiled in */ #cmakedefine H5_METADATA_TRACE_FILE @H5_METADATA_TRACE_FILE@ -/* Define if your system can handle complicated MPI derived datatype - correctly. */ -#cmakedefine H5_MPI_COMPLEX_DERIVED_DATATYPE_WORKS @H5_MPI_COMPLEX_DERIVED_DATATYPE_WORKS@ - /* Define if your system's `MPI_File_set_size' function works for files over 2GB. */ #cmakedefine H5_MPI_FILE_SET_SIZE_BIG @H5_MPI_FILE_SET_SIZE_BIG@ -/* Define if your system can handle special collective IO properly. */ -#cmakedefine H5_MPI_SPECIAL_COLLECTIVE_IO_WORKS @H5_MPI_SPECIAL_COLLECTIVE_IO_WORKS@ - /* Define if we can violate pointer alignment restrictions */ #cmakedefine H5_NO_ALIGNMENT_RESTRICTIONS @H5_NO_ALIGNMENT_RESTRICTIONS@ diff --git a/config/ia64-linux-gnu b/config/ia64-linux-gnu index 519d0fc..9f1b262 100644 --- a/config/ia64-linux-gnu +++ b/config/ia64-linux-gnu @@ -142,12 +142,3 @@ case $CXX_BASENAME in PROFILE_CPPFLAGS= ;; esac - -# Check MPICH settings -. $srcdir/config/mpich -#Uncomment the next line if your system doesn't support MPI complex derived datatype. -#hdf5_cv_mpi_complex_derived_datatype_works=${hdf5_cv_mpi_complex_derived_datatype_works='no'} -#Comment out the following line if your system supports collective IO when some processes -#don't have any contributions to IOs. -hdf5_cv_mpi_special_collective_io_works=${hdf5_cv_mpi_special_collective_io_works='no'} - diff --git a/config/ibm-aix b/config/ibm-aix index a94be4b..7022ab6 100644 --- a/config/ibm-aix +++ b/config/ibm-aix @@ -142,11 +142,6 @@ if test "X-" = "X-$f9x_flags_set"; then f9x_flags_set=yes fi -# With poe version 3.2.0.19 or lower(using lpp -l all | grep ppe.poe to check the version number, -# IBM MPI-IO implementation has a bug, -#it cannot generate correct MPI derived datatype. Please uncomment the following line: -#hdf5_cv_mpi_complex_derived_datatype_works=${hdf5_cv_mpi_complex_derived_datatype_works='no'} - # The default C++ compiler # Use AIX supplied C++ compiler by default. diff --git a/config/irix6.x b/config/irix6.x index 9bd9c4c..8179b86 100644 --- a/config/irix6.x +++ b/config/irix6.x @@ -171,26 +171,6 @@ hdf5_cv_ulong_to_fp_bottom_bit_accurate=${hdf5_cv_ulong_to_fp_bottom_bit_accurat hdf5_cv_ldouble_to_integer_accurate=${hdf5_cv_ldouble_to_integer_accurate='no'} hdf5_cv_integer_to_ldouble_accurate=${hdf5_cv_integer_to_ldouble_accurate='no'} -# For IRIX 6.5, any version that is older than MIPSpro 7.3.1.3m, -# the MPI derived datatype is not working. -# Versions 7.4.2m or newer work. -# Up to version 7.4.4m, it cannot handle collective IO with non-contribution -# of some processes. -# Fix $hdf5_cv_mpi_complex_derived_datatype_works if it is not set and is using cc. -if [ -z "$hdf5_cv_mpi_complex_derived_datatype_works" -a $CC_BASENAME = cc ]; then - ccversion=`$CC -version 2>&1 | sed -e 's/.*Version //p'` - ccversion1=`echo $ccversion | cut -f1 -d.` - ccversion2=`echo $ccversion | cut -f2 -d.` - # Assume all versions 7.4.* or newer are okay - # and assume ccversion2 is never larger than 99. - ccversionval=`expr $ccversion1 \* 100 + $ccversion2` - hdf5_cv_mpi_special_collective_io_works='no' - if [ $ccversionval -lt 704 ]; then - hdf5_cv_mpi_complex_derived_datatype_works='no' -# hdf5_cv_mpi_special_collective_io_works='no' - fi -fi - # Set flag to generate alternate code for H5V_log2_gen, to avoid # problems with the MIPSpro compiler 7.30 and IRIX64 6.5 (ie. other # combinations might work, but haven't been tested) diff --git a/config/linux-gnulibc1 b/config/linux-gnulibc1 index 8f04cf0..9dcc411 100644 --- a/config/linux-gnulibc1 +++ b/config/linux-gnulibc1 @@ -125,23 +125,6 @@ if test -z "$CXX"; then CXX_BASENAME=g++ fi -# Check MPICH settings -. $srcdir/config/mpich - -# The ChaMPIon on NCSA tungsten doesn't work correctly for either of the following -# cases: -# 1. collective I/O when some processes don't have any contributions; -# 2. complex derived MPI data type. -if test $CC_BASENAME = cmpicc; then - hdf5_cv_mpi_special_collective_io_works=${hdf5_cv_mpi_special_collective_io_works='no'} - hdf5_cv_mpi_complex_derived_datatype_works=${hdf5_cv_mpi_complex_derived_datatype_works='no'} -fi - -#Comment out the following line if your system supports collective IO when some processes -#don't have any contributions to IOs. -hdf5_cv_mpi_special_collective_io_works=${hdf5_cv_mpi_special_collective_io_works='no'} -hdf5_cv_mpi_complex_derived_datatype_works=${hdf5_cv_mpi_complex_derived_datatype_works='no'} - # compiler version strings case $CC in # whatever matches *pgcc* will also match *gcc*, so this one must come first diff --git a/config/mpich b/config/mpich deleted file mode 100644 index 575859e..0000000 --- a/config/mpich +++ /dev/null @@ -1,83 +0,0 @@ -# -*- shell-script -*- -# -# Copyright by The HDF Group. -# Copyright by the Board of Trustees of the University of Illinois. -# All rights reserved. -# -# This file is part of HDF5. The full HDF5 copyright notice, including -# terms governing use, modification, and redistribution, is contained in -# the files COPYING and Copyright.html. COPYING can be found at the root -# of the source code distribution tree; Copyright.html can be found at the -# root level of an installed copy of the electronic HDF5 document set and -# is linked from the top-level documents page. It can also be found at -# http://hdfgroup.org/HDF5/doc/Copyright.html. If you do not have -# access to either file, you may request a copy from help@hdfgroup.org. - -# This file should be sourced into configure if the compiler is the -# MPICH compiler script. It is careful not to do anything if the compiler -# is not MPICH. - -# Check if mpicc can support complicated derived datatype correctly. -# We know that mpich 1.2.4 and 1.2.5 do not support it correctly. We assume -# older versions do not work either. We don't know of a way of testing its -# correctness without the risk of hanging the configure process. So, we -# set the configure variable hdf5_cv_mpi_complex_derived_datatype_works to no. -# Notice that this code works only if the mpicc compiler shows its $MPIVERSION -# properly. It is confirmed mpicc does that as far back as v1.2.3. - -# mpich2 do not support it correctly. But mpich2 reports small version number -# indistiguishable from mpich. Some mpich2 reports blank, some reports 1.0.2. - -if [ -z "$hdf5_cv_mpi_complex_derived_datatype_works" ]; then - ccversion=`$CC -v 2>/dev/null` - # mpich compiler will give "mpicc for 1.2.x ..." - if echo "$ccversion" | grep '^mpicc for' > /dev/null ; then - # $CC is an MPICH compiler. Grab the version numbers. - ccversion=`echo $ccversion | cut -f3 -d' '` - case "$ccversion" in - 1.2.[0-5]*) - hdf5_cv_mpi_complex_derived_datatype_works='no' - ;; - 1.0.2) - # mpich2 is recycling the version number, some report - # 1.0.2, some report blank. - hdf5_cv_mpi_complex_derived_datatype_works='no' - ;; - "") - # got blank ccversion. Assume it is bad a mpich2. - hdf5_cv_mpi_complex_derived_datatype_works='no' - ;; - *) - # assume okay - ;; - esac - fi -fi - -if [ -z "$hdf5_cv_mpi_special_collective_io_works" ]; then - ccversion=`$CC -v 2>/dev/null` - # mpich compiler will give "mpicc for 1.2.x ..." - if echo "$ccversion" | grep '^mpicc for' > /dev/null ; then - # $CC is an MPICH compiler. Grab the version numbers. - ccversion=`echo $ccversion | cut -f3 -d' '` - case "$ccversion" in - 1.2.[0-6]*) - hdf5_cv_mpi_special_collective_io_works='no' - ;; - 1.0.2) - # mpich2 is recycling the version number, some report - # 1.0.2, some report blank. - hdf5_cv_mpi_special_collective_io_works='no' - ;; - "") - # got blank ccversion. Assume it is bad a mpich2. - hdf5_cv_mpi_special_collective_io_works='no' - ;; - *) - # assume okay - ;; - esac - fi -fi - - diff --git a/config/x86_64-redstorm-linux-gnu b/config/x86_64-redstorm-linux-gnu index 0cd2d73..76f9711 100644 --- a/config/x86_64-redstorm-linux-gnu +++ b/config/x86_64-redstorm-linux-gnu @@ -162,5 +162,3 @@ hdf5_cv_ullong_to_fp_cast_works=${hdf5_cv_ullong_to_fp_cast_works='yes'} hdf5_cv_ullong_to_ldouble_precision_works=${hdf5_cv_ullong_to_ldouble_precision_works='yes'} hdf5_cv_vsnprintf_works=${hdf5_cv_vsnprintf_works='yes'} hdf5_cv_fp_to_integer_overflow_works=${hdf5_cv_fp_to_integer_overflow_works='yes'} -# mpich2 used has derived datatype errors. -hdf5_cv_mpi_complex_derived_datatype_works=${hdf5_cv_mpi_complex_derived_datatype_works='no'} |