nwchem-6.3 compiling error with gcc 4.8.0 and mvapich2-1.9


Clicked A Few Times
Dear Developers

I was try to compile nwchem-6.3 with gcc 4.8.0 and mvapich2-1.9. The compilation stopped at screened_op.F, the error message looks like "

gfortran -c -m64 -ffast-math -fdefault-integer-8 -Warray-bounds -O3 -mfpmath=sse -ffast-math -fprefetch-loop-arrays -fno-aggressive-loop-optimizations -march=opteron -mtune=opteron -I. -I/n/home/apps/NWCHEM/nw6.3-mvapich2-1.9-gcc4.8.0/src/include -I/n/home/apps/NWCHEM/nw6.3-mvapich2-1.9-gcc4.8.0/src/tools/install/include -DGFORTRAN -DEXT_INT -DLINUX -DLINUX64 -DCHKUNDFLW -DGCC4 -DGCC46 -DPARALLEL_DIAG -DUSE_TEXAS screened_op.F
screened_op.F: In function ‘density_average’:
screened_op.F:155:0: internal compiler error: Illegal instruction
      end
^
0x8687ef crash_signal
       ../.././gcc/toplev.c:332
0x5f3b7d do_mpfr_arg1
       ../.././gcc/builtins.c:13680
0x6022ca fold_builtin_cbrt
       ../.././gcc/builtins.c:7601
0x60614c fold_builtin_1
       ../.././gcc/builtins.c:10409
0x5ff5d3 fold_builtin_n
       ../.././gcc/builtins.c:11054
0x60792d fold_call_expr(unsigned int, tree_node*, bool)
       ../.././gcc/builtins.c:11181
0x71e926 gimple_fold_stmt_to_constant_1(gimple_statement_d*, tree_node* (*)(tree_node*))
       ../.././gcc/gimple-fold.c:2628
0x8e5b38 ccp_fold
       ../.././gcc/tree-ssa-ccp.c:1060
0x8e5b38 evaluate_stmt
       ../.././gcc/tree-ssa-ccp.c:1505
0x8e6e7c visit_assignment
       ../.././gcc/tree-ssa-ccp.c:1985
0x93cb42 simulate_stmt
       ../.././gcc/tree-ssa-propagate.c:314
0x93d025 simulate_block
       ../.././gcc/tree-ssa-propagate.c:437
0x93d025 ssa_propagate(ssa_prop_result (*)(gimple_statement_d*, edge_def**, tree_node**), ssa_prop_result (*)(gimple_statement_d*))
       ../.././gcc/tree-ssa-propagate.c:844
0x8e3efb do_ssa_ccp
       ../.././gcc/tree-ssa-ccp.c:2109
Please submit a full bug report,
"

My configuration is

export TCGRSH=/usr/bin/ssh
export CC=gcc
export FC=gfortran
export LARGE_FILES=TRUE
export USE_NOFSCHECK=TRUE
export NWCHEM_TARGET=LINUX64
export NWCHEM_MODULES="all python"
export MRCC_THEORY=TRUE
export BLAS_I8=yes
export USE_MPIF=y
export USE_MPI=y
export USE_MPIF4=y

export MPI_LOC=/n/centos6/mvapich2-1.9_gcc-4.8.0
export MPI_LIB=$MPI_LOC/lib
export MPI_INCLUDE=$MPI_LOC/include
export LIBMPI="-L/lib -Wl,-rpath,/lib -L/lib -Wl,-rpath,/lib -I/n/centos6/mvapich2-1.9_gcc-4.8.0/include -L/n/centos6/mvapich2-1.9_gcc-4.8.0/lib -lmpich -lopa -lmpl -lrt -libumad -libverbs -lm -lpthread"

export BLAS_SIZE=8
export BLASOPT="-L/n/home/apps/MYMATH/acml5.3.1/gfortran64_fma4_int64/lib -lacml"

export ARMCI_NETWORK=OPENIB
export IB_INCLUDE=/usr/include/infiniband
export IB_LIB=/usr/lib64
export IB_LIB_NAME="-libverbs -libumad -lpthread -lrt"

Could you please give some suggestions? Thanks!

best,

johr

Forum Vet
John
Unfortunately, I am not able to reproduce your error by using the gcc 4.8.0 I got from gnu.org.
Could you try the following command (that disable the optimization flags)

make FC=gfortran FOPTIMIZE=-O1

Cheers, Edo

Clicked A Few Times
Thanks! Edo

Now I found out that this error might be related to gcc 4.8.0 installed.

best,

John

Quote:Edoapra Aug 20th 10:55 pm
John
Unfortunately, I am not able to reproduce your error by using the gcc 4.8.0 I got from gnu.org.
Could you try the following command (that disable the optimization flags)

make FC=gfortran FOPTIMIZE=-O1

Cheers, Edo


Forum >> NWChem's corner >> Compiling NWChem