llmodel: add CUDA to the DLL search path if CUDA_PATH is set #2357
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
I tested this change with Qt Creator on Windows. Without it, the CUDA runtime libraries are only found if they are copied into the build directory. With this change, the CUDA runtime libraries are found as long as CUDA_PATH is set correctly, which is done automatically by the CUDA Toolkit installer.
As far as the chat UI is concerned, this change only matters for users that build GPT4All from source themselves, which already requires the CUDA Toolkit. Once GPT4All is installed, the CUDA libraries are copied into the install directory.
This change should also benefit the bindings, by allowing the system CUDA libraries to be used automatically if they are installed.