# Ubuntu 18.04 with RTX 2080Ti

**URL:** <https://forums.developer.nvidia.com/t/ubuntu-18-04-with-rtx-2080ti/71516>\
**Category:** cuDNN\
**Created:** [March 14, 2019, 2:04pm UTC](https://forums.developer.nvidia.com/t/ubuntu-18-04-with-rtx-2080ti/71516 "2019-03-14T14:04:53Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![ith.user](https://developer.download.nvidia.com/images/forums/profile-default-devtalk-84.png) [@ith.user](https://forums.developer.nvidia.com/u/ith.user)\
**Post date:** [March 14, 2019, 2:04pm UTC](https://forums.developer.nvidia.com/t/ubuntu-18-04-with-rtx-2080ti/71516/1 "2019-03-14T14:04:53Z")

</div>

hi,  
i installed CUDA and cuDNN on my pc and would like to use it to start some ML projects.  
i installed the following:  
dirver: 410.57  
CUDA: 10.01 V10.1.105

but i read that Tensorflow for GPU only works on CUDA 10.0

I have two questions:

1  
which is the most stable and combination that can be installed in order to use

Tensorflow for GPU + Keras

CUDA 10.0 with 410.57 driver and cuDNN 7.5 or which should i use?

2  
Can i use other cuda + cuDNN setups in virtualenvs?  
how should i link the path in my virtual env?

Thanks