summaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorMuQun Yang <myang6@hdfgroup.org>2008-09-08 22:55:07 (GMT)
committerMuQun Yang <myang6@hdfgroup.org>2008-09-08 22:55:07 (GMT)
commitfc7993a9f068319ed35f13fd153feeb5da3d86a7 (patch)
tree19114d1d29ea3db96c4c9fd4eb6e1726b1eac056
parent2b6b6825ab77d8aaf80ace6a2b26e6614572e5d6 (diff)
downloadhdf5-fc7993a9f068319ed35f13fd153feeb5da3d86a7.zip
hdf5-fc7993a9f068319ed35f13fd153feeb5da3d86a7.tar.gz
hdf5-fc7993a9f068319ed35f13fd153feeb5da3d86a7.tar.bz2
[svn-r15605] The MPI compiler(mvapich2-0.9.8p2patched-intel-ofed-1.2) at NCSA's linux cluster(Abe) fails to handle collective IO calls with very complicate MPI derived datatype. All the values of input parameters from HDF5 are correct before passing to MPI-IO collective calls. Sometimes the following errors happen:
rank 0 in job 77 honest1.ncsa.uiuc.edu_60945 caused collective abort of all ranks exit status of rank 0: killed by signal 9 Up to this point, we decide to turn off the use of advanced MPI derived datatype features inside HDF5 so that complicated selection tests are ensured to be handled by HDF5. This check-in reflects this.
-rw-r--r--config/linux-gnulibc14
1 files changed, 3 insertions, 1 deletions
diff --git a/config/linux-gnulibc1 b/config/linux-gnulibc1
index 015d079..b075e31 100644
--- a/config/linux-gnulibc1
+++ b/config/linux-gnulibc1
@@ -128,4 +128,6 @@ fi
#Comment out the following line if your system supports collective IO when some processes
#don't have any contributions to IOs.
-#hdf5_mpi_special_collective_io_works=${hdf5_mpi_special_collective_io_works='no'}
+hdf5_mpi_special_collective_io_works=${hdf5_mpi_special_collective_io_works='no'}
+hdf5_mpi_complex_derived_datatype_works=${hdf5_mpi_complex_derived_datatype_works='no'}
+