Update CUDA search path to pick up `cudatoolkit` in Conda installs
See original GitHub issueBased on discussion starting here, it appears cudatoolkit
is not getting picked up. This could easily be fixed by adding sys.prefix
to the CUDA search paths.
Issue Analytics
- State:
- Created 3 years ago
- Comments:9 (9 by maintainers)
Top Results From Across the Web
how to set the cuda path in the conda environment? [closed]
You can set the CUDA_HOME environment variable with this command. ... this command to find the path of your conda-installed CUDA library.
Read more >Managing CUDA dependencies with Conda | by David R. Pugh
You can use the conda search command to see what versions of the NVIDIA CUDA Toolkit are available from the default channels.
Read more >Where is CUDAToolKit path when installed via conda?
I installed my PyTorch 1.0 using the command conda install pytorch torchvision cudatoolkit=9.0 -c pytorch while my system has an existing cudatoolkit ......
Read more >NVIDIA CUDA Installation Guide for Linux
The installation instructions for the CUDA Toolkit on Linux. ... If you do not see any settings, update the PCI hardware database that...
Read more >Install conda and set up a Pytorch 1.7, CUDA 11.1 ...
The first line creates our environment called “PyTorch” and you can select the python version (I choose version 3.7). · The second line...
Read more >Top Related Medium Post
No results found
Top Related StackOverflow Question
No results found
Troubleshoot Live Code
Lightrun enables developers to add logs, metrics and snapshots to live code - no restarts or redeploys required.
Start FreeTop Related Reddit Thread
No results found
Top Related Hackernoon Post
No results found
Top Related Tweet
No results found
Top Related Dev.to Post
No results found
Top Related Hashnode Post
No results found
Top GitHub Comments
For my own note:
It seems these are used in build time, though. Will try to dig the runtime ones.
For discussions on potential changes in cupy-feedstock, I’m redirecting to conda-forge/cupy-feedstock#46.