libtorch_NVC_COMPILE

#!/bin/bash

rm nvc_main.x

CXX="nvc++"
#NOTE: TO RUN THE EXECUTABLE, YOU NEED TO LOAD THE ENVIRONMENT OF NVIDIA-HPC-SDK
export PATH=/opt/nvidia/hpc_sdk/Linux_x86_64/24.5/compilers/bin:$PATH

torchDir="/$HOME/libtorch/"

LINKFLAG="-D_GLIBCXX_USE_CXX11_ABI=1 -L${torchDir}/lib -I${torchDir}/include/ -I${torchDir}/include/torch/csrc/api/include -Wl,-R${torchDir}/lib -ltorch -ltorch_cpu -lc10 -ltorch_cuda -L/opt/nvidia/hpc_sdk/Linux_x86_64/22.5/cuda/lib64 -L/usr/lib64/ -L/opt/local/lib/gcc11/"

NVCFLAG="-O3 -std=c++20 -target=gpu -Minline -fopenmp -cuda -stdpar=multicore"

$CXX $NVCFLAG $LINKFLAG -c axpy.cu  && echo "Finished axpy.cu" &

$CXX $NVCFLAG $LINKFLAG -c main.cpp  && echo "Finished main.cpp" &

$CXX $NVCFLAG $LINKFLAG -c read_data.cpp  && echo "Finished read_data.cpp" &

$CXX $NVCFLAG $LINKFLAG -c data_struct.cpp  && echo "Finished data_struct.cpp" &

$CXX $NVCFLAG $LINKFLAG -c VDW_Coulomb.cu  && echo "Finished VDW_Coulomb.cu" &

wait

$CXX $NVCFLAG $LINKFLAG main.o read_data.o axpy.o data_struct.o VDW_Coulomb.o -o nvc_main.x ; echo "Finished Linking"

rm *.o