Hi,
We have been having problems with compiling 6.3 on the BlueGene/Q platform using MPI-TS as advised on http://nwchemgit.github.io/index.php/Compiling_NWChem.
It appears to be a GA problem as simple MD runs that don't use GA run OK, but other tests fail along the lines of:
MA_verify_allocator_stuff: starting scan ...
stack block 'gai_diag_std:z', handle 1, address 0x11ae499bb8:
current right signature 1060908322 != proper right signature 1431655765
stack block '?Q¥Ò£Ó?GÍX?', handle 0, address 0x11ae4c4898:
current checksum 13660927947873495360 != stored checksum 4563766883317136985
We get this error with any number of processors including nproc=1.
The environment variables set before compilation were:
export NWCHEM_TOP=/gpfs/ .... /nwchem-6.3
export LARGE_FILES TRUE
export USE_NOFSCHECK TRUE
export LIB_DEFINES="-DDFLT_TOT_MEM=16777216"
export NWCHEM_TARGET=BGQ
export NWCHEM_MODULES="all"
export USE_MPI=y
export USE_MPIF=y
export USE_MPIF4=y
export MPI_INCLUDE="/bgsys/drivers/ppcfloor/comm/xl/include"
export LIBMPI=" "
export ARMCI_NETWORK=MPI-TS
export DISABLE_GAMIRROR=y
We have also tried using ARMCI-MPI instead of the built-in version but get similar problems.
Has anyone encountered an error of this kind before on BGQ?
Tom
|