0

When I execute some code (from this repo: https://github.com/guicho271828/latplan), I get the following error:

W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'libcudart.so.10.0'; dlerror: libcudart.so.10.0: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH:

I have installed v1.15 of tensorflow (which is mandatory for the project I am running), and to make it compatible with the GPU I am using I also installed "nvidia-tensorflow" and "nvidia-pyindex".

I also tried :

conda install -c nvidia cuda-toolkit

But it just installed cuda v11 in the conda env, and the same error appears again.

So, is it possible to install cuda v10 instead of cuda v11 with conda ?

Any idea to go around this issue ?

Thank you

p.s: I already have checked other similar issues like this one this one but it does not solve my problem as I cannot install another cuda version outside a conda environement (I don't have root access)

talonmies
  • 70,661
  • 34
  • 192
  • 269
RandomFellow
  • 317
  • 1
  • 9
  • You don't really need root access to install cuda (just think of how anaconda is installing cuda without you being root). – Dr. Snoopy Jun 03 '22 at 11:03
  • Yes, This could be the reason. Also please check the Tensorflow, Python, Cuda and cuDnn version with this [tested build configuration](https://www.tensorflow.org/install/source_windows#gpu) to get gpu support enabled. You can follow the instructions mentioned in [this](https://www.tensorflow.org/install/pip) link to install cuda in your conda environment. –  Jul 27 '22 at 17:37

0 Answers0