You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
the README.md in the cuda directory claims,
"Building should be straightforward by using the respective CMakeList.txt file."
But it is not straightforward at all.
when I try "cmake .", it errors with CMAKE_CUDA_ARCHITECTURES must be non-empty if set.
but there is zero instruction about this, and nothing about it in the CMakeList.txt
Please update with relevant information
`
I am using:
Ubuntu 22
Cuda compilation tools, release 12.2, V12.2.140
Update:
It appears if I force the cuda tools to version 11, it is more forgiving.
But if you arent going to update to support the latest, then please update README to say this requires cuda version 11.
The text was updated successfully, but these errors were encountered:
Thanks for the heads up. Yes, this parameter looks that has been introduced to CMake after my past tests.
A workaround is to set its value in the cmake command with the default CC (compute capability) of your device, e.g. for GTX-1060 that would be cmake ../mixbench-cuda -DCMAKE_CUDA_ARCHITECTURES=61
the README.md in the cuda directory claims,
"Building should be straightforward by using the respective
CMakeList.txt
file."But it is not straightforward at all.
when I try "cmake .", it errors with
CMAKE_CUDA_ARCHITECTURES must be non-empty if set.
but there is zero instruction about this, and nothing about it in the CMakeList.txt
Please update with relevant information
`
I am using:
Ubuntu 22
Cuda compilation tools, release 12.2, V12.2.140
Update:
It appears if I force the cuda tools to version 11, it is more forgiving.
But if you arent going to update to support the latest, then please update README to say this requires cuda version 11.
The text was updated successfully, but these errors were encountered: