Ranter
Join devRant
Do all the things like
++ or -- rants, post your own rants, comment on others' rants and build your customized dev avatar
Sign Up
Pipeless API
From the creators of devRant, Pipeless lets you power real-time personalized recommendations and activity feeds using a simple API
Learn More
Comments
-
running it inside a vm wouldn't help?
The last time I ran ubuntu as a default installation was 13 years ago, ever since I just throw it in a VM and use it -
I have a txt I used to ctrl+c/v commands into a Ubuntu 16 server. Sets up Cuda with a virtualenv to use Keras with Theano backend, all the latest. Maybe that could help you?
-
@gitpush I usually do the WSL or VM route but they don't have the GPU pass-through on Windows :(
-
@polaroidkidd Okay, so this might not be the cleanest, but it worked on my dual-booted ubuntu and parts were applied to an azure-hosted ubuntu 16. 04 VM.
I installed stuff on Windows and Ubuntu a bunch of times, so if something comes up I might be able to help beyond this :)
Hope it works out!
https://github.com/d4rkspir1t/... -
mpower2697yUbuntu(mint) is really bad in that part. Took me a week of struggle to make it work with the nvidia driver. I don't know why is this still so hard when every user should go through it.
-
I dont recommend using 16.04. 17.10 worked right out of the box on my laptop (NVidia optimus). 18.04 LTS is right around the corner too
-
@bahua some laptops are just unlucky
A few of my friends suffered similar problems too
I think its more to do with the motherboard and the laptops design than the GPU and linux itself -
bahua129057y@brasorexia
There's also the fact that for me, laptops are for work, not personal use. So I have no particular need for big 3D performance on it. For games, I have a full size computer at home. -
@Seshpenguin Oh nice? I've tried 14 and 16 but since I'm also bound by optimus, I'll try that as well. Have you had any bad experiences with 17?
-
Forside14597yFor me setting up cuda on my Ubuntu and Matlab wasn't hard. Exceptionally easy in fact. But finding the right package was horror. Nvidia has no useful landing page for cuda. I found multiple official sites with years old releases until I finally found the newest one, which was just run setup and it's done. Another part is missing documentation about the versions. You can define a cuda version to use in code, but there is no clear listing about available versions. Had to increment it one by one to find out the newest.
-
@polaroidkidd no, not really. Just make sure you go to the driver manager after installation and pick the NVidia driver (not nouveau). One thing to note is you won't be able to use hybrid graphics (intel and NVidia at the same time), but you can pick which card to use using the 'prime-select' command. Good luck!
-
@Seshpenguin so I've managed to get it all working and it runs like a dream (mostly). I can only use the Nvidia card because when I use the intro card I can't log in anymore (logs me straight out again) so that's a bit annoying. Also, I'm not sure if that's the reason for it, but the fan is always running. Using sensors and pwmconfig fails because pwmconfig claims there are no components which can be configured. Other than that, it's great and I am starting to seriously enjoy it! The UI was a nice surprise as well!
Thanks very much for the tip!
Related Rants
Seriously, Ubuntu can go burn in hell far as I care.
I've spent the better part of my morning attempting to set it up to run with the correct Nvidia drivers, Cuda and various other packages I need for my ML-Thesis.
After countless random freezes, updates,. Downgrades and god-knows-what, I'm going back to Windows 10 (yes, you read that right). It's not perfect but at least I don't have to battle with my laptop to get it running. The only thing which REALLY bothers me about it is the lack of GPU pass-through, meaning running local docker containers rely solely on the CPU. In itself not a huge issue if only I didn't NEED THE GOD DAMN GPU FOR THE TRAINING
rant
windows 10
ubuntu 16.04
machine learning