Paddle
Paddle copied to clipboard
English documentation needs information about how to select CUDA version
文档链接&描述 Document Links & Description
The documentation is here https://www.paddlepaddle.org.cn/documentation/docs/en/install/index_en.html
Screenshot shows support for CUDA 11, but the pip install command later down the page only works for CUDA 10.2.
python -m pip install paddlepaddle==2.4.2 -i https://mirror.baidu.com/pypi/simple
The paddle.utils.run_check() will fail with PreconditionNotMetError: The third-party dynamic library (cublas64_102.dll;cublas64_10.dll) that Paddle depends on is not configured correctly. (error code is 126)
I happened to find the alternate command here python -m pip install paddlepaddle-gpu==2.4.2.post117 -f https://www.paddlepaddle.org.cn/whl/windows/mkl/avx/stable.html for CUDA 11.7 for example, using the graphical configuration selector at paddlepaddle.org.cn
This command works for my system which has CUDA 11.7.
请提出你的建议 Please give your suggestion
Maybe the graphical selector can be copied over to the documentation page. Or the documentation page could mention that if CUDA version != 10.2 is needed, then we should use the graphical selector paddlepaddle.org.cn. Thanks!
I noticed that the installation instructions actually do have the more advanced commands if you click into the "Install ... with PIP" subpage (1)! They are just unlisted on the main page. In that case, it would be helpful to just remove the pip command from the main page (2).
Did you solve the isse ? i am facing same issue `RuntimeError: (PreconditionNotMet) The third-party dynamic library (cublas64_102.dll;cublas64_10.dll) that Paddle depends on is not configured correctly. (error code is 126) Suggestions:
- Check if the third-party dynamic library (e.g. CUDA, CUDNN) is installed correctly and its version is matched with paddlepaddle you installed.
- Configure third-party dynamic library environment variables as follows:
- Linux: set LD_LIBRARY_PATH by
export LD_LIBRARY_PATH=... - Windows: set PATH by
set PATH=XXX; (at ..\paddle\phi\backends\dynload\dynamic_loader.cc:303)cuDNN Version: 7.6. CUDA Version 12.3