Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Test needed for AMD Docker #29

Closed
adamradocz opened this issue Jul 14, 2018 · 7 comments
Closed

Test needed for AMD Docker #29

adamradocz opened this issue Jul 14, 2018 · 7 comments

Comments

@adamradocz
Copy link
Member

Unlike the Nvidia and Intel Dockers, the AMD OpenCL docker doesn't contain the AMD OpenCL Driver, therefore I'm sure it's not working.

Unfortunately, I don't have AMD Radeon GPU to update and test it.

@adamradocz adamradocz changed the title Update and test needed for AMD OpenCL docker Test needed for AMD Docker Jul 20, 2018
@adamradocz
Copy link
Member Author

I updated the AMD Docker, it just has to be tested.
I couldn't test it, because I don't have AMD GPU, but I merged it, because the older version definitely was not working, so this commit could help the situation not worsen.

@marius311
Copy link
Member

Good by me, thanks for doing all this stuff, sorry I've been too busy to really help. By the way, I tried to find your email so I could send you a message about the BOINC Workshop next week, but couldn't. Can you send me an email, you can see mind by just clicking on my Github profile then my website. Thanks!

@sunk818
Copy link

sunk818 commented Oct 28, 2018

@adamradocz I have macBook with AMD Radeon... Is that can acceptable use case? I've never run docker on macOS, but willing to give it a try. Is Collatz Conjecture okay for GPU? I've run tasks on the macBook Pro successfully using the macOS BOINC client.

@adamradocz
Copy link
Member Author

adamradocz commented Oct 28, 2018

Unfortunately MacOS not supported by ROCm.
ROCm/ROCm#262

@kliu128
Copy link

kliu128 commented Nov 1, 2018

I've been using it for a while on an RX 580, kernel 4.18, and it seems to work fine for Einstein@Home. Milkyway gives an error on every GPU job, but that might be their issue: http://milkyway.cs.rpi.edu/milkyway/forum_thread.php?id=4185

Thanks for your work in supporting it!

@adamradocz
Copy link
Member Author

@Pneumaticat thanks for the feedback.

I think the addig yourself to the video group is unnecessary: 'sudo usermod -a -G video $LOGNAME'

If you could confirm, it would be awesome.

@kliu128
Copy link

kliu128 commented Nov 2, 2018

@adamradocz No problem!

I can't test that (I'm on NixOS, so I can't change the groups I'm in, plus I think it would break my GUI setup), but I can see no reason why it should be necessary, as BOINC doesn't run as your user in the docker container anyway, and plus you add --group-add video already to docker run.

Also, I don't think that the host technically has to install the ROCm userspace driver either, as it is installed in the Docker image. (At least, I haven't had to install it, and BOINC still works fine.) I believe the host only needs the kernel component, and according to Phoronix, any mainline kernel >4.17 will also work, as it has the driver built in.

So really, all the user needs is an up-to-date kernel ;)

EDIT: Actually, I can indeed confirm that the video group is unnecessary, at least for me, because my user isn't even in the video group normally.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

4 participants