0
$\begingroup$

I've managed to build Tensorflow 2.3 from source, following these instructions: https://towardsdatascience.com/how-to-compile-tensorflow-2-3-with-cuda-11-1-8cbecffcb8d3

But, when I install obtained pip package in new conda environment, and import tensorflow, I get following error: Could not load dynamic library 'libcudart.so.11.1'; dlerror: libcudart.so.11.1: cannot open shared object file: No such file or directory

I've managed to use GPU support with CUDA 11.1 for Tensorflow 2.5 nightly, without creating soft links between libs (I get Successfully opened dynamic library libcudart.so.11.0 message).

Any help appreciated.

Improve this question
$\endgroup$

1 Answer 1

Reset to default
0
$\begingroup$

It's working with symbolic link: libcudart.so.11.1 -> libcudart.so.11.0

I guess it's because the nightly build has CUDA 11.1 fully supported.

Improve this answer
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.