| 8:58:20 AM PST - Wed, Feb 10th 2016 |
|
My installation nwchem on cuda is fail. I have standart problems with SLP the basic block and this
/MD/azat/nwchem-6.6/lib/LINUX64/libtce.a(memory.o): In function `std::_Rb_tree<void*, std::pair<void* const, int>, std::_Select1st<std::pair<void* const, int> >, std::less<void*>, std::allocator<std::pair<void* const, int> > >::_M_erase_aux(std::_Rb_tree_const_iterator<std::pair<void* const, int> >)':
/MD/azat/nwchem-6.6/lib/LINUX64/libtce.a(memory.o): In function `__gnu_cxx::new_allocator<std::_Rb_tree_node<void*> >::deallocate(std::_Rb_tree_node<void*>*, unsigned long)':
/MD/azat/nwchem-6.6/lib/LINUX64/libtce.a(memory.o): In function `__gnu_cxx::new_allocator<std::_Rb_tree_node<void*> >::allocate(unsigned long, void const*)':
/MD/azat/nwchem-6.6/lib/LINUX64/libtce.a(memory.o): In function `__gnu_cxx::new_allocator<std::_Rb_tree_node<std::pair<void* const, int> > >::allocate(unsigned long, void const*)':
/MD/azat/nwchem-6.6/lib/LINUX64/libtce.a(memory.o): In function `__gnu_cxx::new_allocator<std::_Rb_tree_node<std::pair<int const, std::set<void*, std::less<void*>, std::allocator<void*> > > > >::allocate(unsigned long, void const*)':
collect2: error: ld returned 1 exit status
make: *** [all] Error 1
without cuda Nwchem worked.
My steps:
export ARMCI_NETWORK="MPI-PR"
export BLASOPT="-L/MD/azat/Openblas/lib -lopenblas -lpthread -lrt"
export BLAS_SIZE="4"
export CUDA="nvcc"
export CUDA_FLAGS="-arch sm_20 "
export CUDA_INCLUDE="-I. -I/usr/local/cuda-6.5/include"
export CUDA_LIBS="-L/usr/local/cuda-6.5/lib64 -L/usr/local/cuda-6.5/lib -lcudart"
export FC="gfortran"
LD_LIBRARY_PATH="/usr/local/cuda-6.5/lib64:/root/.local/Tcl/lib:/home/azat/Downloads/amber14/lib:/usr/lib64/mpi/gcc/openmpi/lib64:/MD/azat/Openblas/lib:/usr/local/lib:/usr/lib64/mpi/gcc/openmpi/lib64:/MD/azat/Openblas/lib"
export NWCHEM_MODULES="all"
export NWCHEM_TARGET="LINUX64"
export NWCHEM_TOP="/MD/azat/nwchem-6.6"
export TCE_CUDA="Y"
export USE_64TO32="y"
export USE_MPI="y"
make nwchem_config
make 64_to_32
make > make.log
linux432:/MD/azat/nwchem-6.6/src # nvcc --version
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2014 NVIDIA Corporation
Built on Thu_Jul_17_21:41:27_CDT_2014
Cuda compilation tools, release 6.5, V6.5.12
NVIDIA 650 Ti
Gromacs programm package work with CUDA.
When I tryed install with
LAPACK_SIZE=4
SCALAPACK_SIZE=4
SCALAPACK=-lsci_pgi_mp
BLASOPT=-lsci_pgi_mp
I had error: cannot found: -lsci_pgi_mp.
|
|
|