Re:Pop_OS 21.04
by edgyquant on Wednesday June 30, 2021 @01:02AM
(#27685065)
I can’t get cuda to work for me on my nvidia 1660x. I’ve followed their steps straight from nvidia, am using the Ubuntu LTS, etc. I’m sure I must be doing something wrong but I always get a dependency issue (cuda depends on X which isn’t going to be installed) or when I try to import tensorflow it claims it can’t find the .so files. I’ve added the cuda path in opt to the LD path with no luck. I bought nvidia specifically to be able to use tensorflow (had an amd prior) and I’m just kinda upset about it.
[ Parent
| on HN ]
Installing CUDA (with kernel driver, CuDNN, Tensorflow/PyTorch) has not been easy for me either. Usually you want to have the latest version of all of them. But sometimes the very latest version of these packages is not yet supported by the other. This is where it get difficult. Also the fact that the CUDA instruction from TensorFlow [1] still are at Ubuntu 18.04 does not help.
In my experience installing PyTorch with CUDA is easier than TensorFlow (and much better documented).
The best experience, in my opinion, is however with CUDA and Flux.jl or Knet.jl on Julia. All you need there is the kernel driver (and julia). CUDA and CuDNN will be installed per default if necessary. I am wondering why CUDA is not simply a pip-installable package (like it is in Julia or in Conda).
Re:Pop_OS 21.04
by eigenspace on Wednesday June 30, 2021 @12:38PM
(#27690613)
My understanding is that the dependency analysis in Pip just isn’t very good, so it’s easy to end up with incompatible versions.
You may want to try this guide: https://medium.com/@dun.chwong/the-ultimate-guide-ubuntu-18-...
I also had to resort to looking at guides on Medium.com for installing CUDA, as the install is not as straightforward as it looks. Alternatively, you can search on DuckDuckGo (for privacy reasons) with this as the query: "NVIDIA" AND "CUDA" AND "install" site:medium.com
Re:Pop_OS 21.04
by rahimnathwani on Wednesday June 30, 2021 @02:18AM
(#27685515)
The fast.ai student community forums have good tips for people having trouble getting CUDA to work.