Closed Yi-Lynn closed 8 months ago
A quick note on how I address this issue, hopefully useful for people running into similar situations:
CUDA_TOOLKIT_ROOT_DIR
to the installation path of the specific version of CUDA I need. Then the error Could NOT find CUDA: Found unsuitable version "9.1", but required is at least "10" (found /usr)
was resolved.non-zero exit status 2
error as 37#issue.
Many thanks for your work!
I am trying to re-implemented this project: SceneSketch, which replied on the diffvg. But I was unable to install diffvg on my Ubuntu.
My Environment Ubuntu 18.04 Python 3.8.0 torch 1.11.0+cu113 CUDA 9.1 highest CUDA version supported by the GPU driver: 11.4
Install procedure I followed
Error log
The highlighted red lines says:
It seems like my CUDA version is too old to build diffvg. I am new to this tool and quite confused about this cmake issue. Is there any way to fix this without having to install a newer version of the CUDA? Thanks!!!
p.s.Some previous issues have already discussed similar error log: 37#issue and 33#issue, but none of them can solve my problem.