lammps CUDA 編譯

cuda7.5.18 lammps10Aug15  compiled on 2015/09/15
-------------------
tar –zxvf openmpi-1.4.5.tar.gz
cd openmpi-1.4.5
./configure --prefix=/opt/opnmpi
make
make install
設置環境變量
gedit ~/.bashrc
PATH=/opt/cuda/bin:/opt/openmpi/bin:$PATH
LD_LIBRARY_PATH=/opt/cuda/lib64:/opt/openmpi/lib:/opt/fftw2/lib:$LD_LIBRARY_PATH
最後source ~/.bashrc
測試openmpi是否安裝成功
which mpicc
which mpiexec
which mpirun


配置lammps
http://lammps.sandia.gov/tars/lammps.tar.gz
tar xvf lammps.tar.gz


首先編譯gpu package
cd lammps/lib/gpu
修改Makefile.linux
CUDA_HOME = /opt/cuda
# Kelper CUDA
CUDA_ARCH = -arch=sm_30    ##########這個值參考這裏https://developer.nvidia.com/cuda-legacy-gpus
CUDA_LIB = -L$(CUDA_HOME)/lib64 -L/opt/cuda/lib64/stubs  ####這裏貌似與以前不同了
（將其他CUDA_ARCH註釋掉）
最後make -f Makefile.linux
生成nvc_get_devices，可以運行一下，看看GPU的信息


修改Makefile.lammps
gpu_SYSINC = -I/opt/cuda/include
gpu_SYSLIB = -lcudart -lcuda
gpu_SYSPATH = -L/opt/cuda/lib64  -L/opt/cuda/lib64/stubs
然後編譯自定義包，我們需要用到user-cuda
cd ../cuda
修改Makefile.common
CUDA_INSTALL_PATH = /opt/cuda
然後make：
make cufft=2 precision=2 arch=21
最後會生成liblammpscuda.a


然後安裝所需要的包:
make yes-asphere
make yes-class2
make yes-colloid
make yes-dipole
make yes-granular
make yes-user-misc
make yes-user-cg-cmm
安裝GPU和USER-CUDA package
make yes-gpu
make yes-user-cuda


編譯lammps
使用/src/MAKE/Makefile.openmpi作爲模版
cp MAKE/Makefile.mpi MAKE/Makefile.gpu
vi MAKE/Makefile.gpu


#MPI_INC =  -I/opt/openmpi/include     ###這個按前面的配置路徑來
#MPI_PATH = 
#MPI_LIB = -L/opt/openmpi/lib -lmpi


JPG_INC =  -I/share/apps/jpeg/include -DLAMMPS_JPEG   ###需要事先安裝這個庫
JPG_PATH =
JPG_LIB = -L/share/apps/jpeg/lib -ljpeg




然後
make gpu
於是生成 lmp_gpu