-
Notifications
You must be signed in to change notification settings - Fork 27
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
CMake error when running dace_gpu
framework
#23
Comments
It looks like a CMake/DaCe configuration issue. I believe DaCe's default C/C++ compilation flags are for the GNU compiler, and they do not work with nvcc/nvc++. DaCe will use nvcc to compile CUDA kernels but will call gcc (or the C/C++ compiler found on the system by CMake, nvc++ in your case) for host code. You can probably solve the issue either by explicitly telling CMake to use a different C/C++ compiler (CC and CXX environment variables) or by editing the DaCe configuration file (~/.dace.conf) to use nvc++-compatible flags for the host (CPU) compiler (this link may be helpful; also this). |
If you want to change the flags, then the
You may need to clean up the build folder (in your example above, that would be You can also explicitly set the host compiler in the DaCe configuration file, which should hopefully be properly passed to CMake:
|
Hi,
I got the following CMake error when running
dace_gpu
with any of the benchmarks:I think the issue is somehow the discovered C++ compiler is not able to compile a simple test program. However, when I tried to manually compile a simple test CUDA program using the same compiler
/opt/nvidia/hpc_sdk/Linux_x86_64/23.9/compilers/bin/nvc++
, it worked fine./opt/nvidia/hpc_sdk/Linux_x86_64/23.9/compilers/bin/
is my CUDA toolkit directory. Any idea what could be the issue? Thanks!The text was updated successfully, but these errors were encountered: