run "./main ../tests/input/supremacy_28.qasm" in directory build, but got "unrecognized token version", so what's the right command line with GPU device ?
pengfzhou opened this issue · 4 comments
when with command "./run-single.sh", also found “unrecognized token version”
[ 79%] Linking CXX executable two-group-h
[ 83%] Linking CXX executable local-ctr
[ 87%] Linking CXX executable bench-blas
[ 91%] Linking CXX executable local-single
[ 91%] Built target two-group-h
[ 91%] Built target local-ctr
[ 91%] Built target bench-blas
[ 91%] Built target local-single
[ 95%] Linking CXX executable process
[ 95%] Built target process
[100%] Linking CXX executable main
[100%] Built target main
unrecognized token version
As mentioned in readme, you should
Specify the compute capability in CMakeLists.txt (CUDA_NVCC_FLAGS) and third-party/cutt/Makefile (GENCODE_FLAGS)
before compile the project.
my GPU is Tesla V100 SXM2 16G, it seems the default CUDA_NVCC_FLAGS setting "-arch=compute_70 -code=sm_70" and default GENCODE _FLAFGS setting "GENCODE_FLAGS := $(GENCODE_SM70)" both are ok.
Seems like you haven't downloaded the qasm files properly. They are stored with git lfs. You can refer to git lfs's document for downloading the files.
thanks for your patience, I have solved the problem!