Skip to content
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

cannot remove ‘BeagleCUDA_kernels.h’: No such file or directory #150

Open
biozhajie opened this issue Jul 14, 2020 · 2 comments
Open

Comments

@biozhajie
Copy link

Hi, Doctor

I encountered an error when I am compling beagle-lib, as the prompt below:

nvcc fatal : Value 'compute_30' is not defined for option 'gpu-architecture'
rm: cannot remove ‘BeagleCUDA_kernels.h’: No such file or directory
make[4]: *** [BeagleCUDA_kernels.h] Error 1
make[4]: Leaving directory /usr/local/beagle-lib-3.1.2/libhmsbeagle/GPU/kernels' make[3]: *** [all-recursive] Error 1 make[3]: Leaving directory /usr/local/beagle-lib-3.1.2/libhmsbeagle/GPU'
make[2]: *** [all-recursive] Error 1
make[2]: Leaving directory /usr/local/beagle-lib-3.1.2/libhmsbeagle' make[1]: *** [all] Error 2 make[1]: Leaving directory /usr/local/beagle-lib-3.1.2/libhmsbeagle'
make: *** [all-recursive] Error 1

Please help me fix this issue, thanks!

Jie

@ayamadanu
Copy link

ayamadanu commented Aug 19, 2020

I also found the same error. In my case, this was because "compute_30" was not allowed by installed version of cuda (11.0.2). In addition, SM supported by my GPU (RTX 2070 super) was not "compute_30" but "compute_75". Accordingly, I modified ./beagle-lib-3.1.2/libhmsbeagle/GPU/kernels/Makefile as follows:

NVCCFLAGS = -O3 -arch compute_30 -Wno-deprecated-gpu-targets
->
NVCCFLAGS = -O3 -arch compute_75 -Wno-deprecated-gpu-targets

Then, I could build beagle 3.1.2 without errors, while I'm not sure beagle can work properly or not.
You can check gpu-code supported by your cuda by "nvcc --help", and know gpu arch on site below:
http://arnon.dk/matching-sm-architectures-arch-and-gencode-for-various-nvidia-cards/

@nlabrad
Copy link

nlabrad commented Mar 1, 2021

I can confirm this is an issue. I ran into the same thing with an NVIDIA 1070. Changed it to compute_61 and it compiled.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

3 participants