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
When CUDA is installed as a system package, it creates a symbolic link: /usr/local/cuda-12.4/include -> targets/x86_64-linux/include.
However, when installing CUDA via Anaconda, it cannot create this link because Anaconda has its own include directory: ~/anaconda3/env/<env-name>/include.
By default, when using Anaconda, CMake's FindCUDA only checks ~/anaconda3/env/<env-name>/include, which
will lead to issues if the necessary headers are not found.
Expected behavior or outcome
x
Conda info
Conda config
Conda list
Additional information
No response
The text was updated successfully, but these errors were encountered:
Checklist
Impacted product
What happened?
#13393 (comment)
pytorch/pytorch#134331 (comment)
When CUDA is installed as a system package, it creates a symbolic link:
/usr/local/cuda-12.4/include -> targets/x86_64-linux/include
.However, when installing CUDA via Anaconda, it cannot create this link because Anaconda has its own include directory:
~/anaconda3/env/<env-name>/include
.By default, when using Anaconda, CMake's
FindCUDA
only checks~/anaconda3/env/<env-name>/include
, whichwill lead to issues if the necessary headers are not found.
Expected behavior or outcome
x
Conda info
Conda config
Conda list
Additional information
No response
The text was updated successfully, but these errors were encountered: