-
Notifications
You must be signed in to change notification settings - Fork 69
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
Implement USE_BBSWITCH option #137
base: master
Are you sure you want to change the base?
Conversation
Signed-off-by: Yousef Sultan <[email protected]>
I would very much like this and I have adjusted my setup to use With the included method, the card sometimes comes up on its own (maybe the system does a PCI rescan) and also there is no way to check if the card is off. Of note is the fact that there is no need for nvidia-xrun to run anything with bbswitch on startup (like the systemd service it has right now). You can set the bbswitch module to load on boot by itself and turn the card off. On Arch you can do this wih 2 files:
and
|
@JpegXguy nvidia-xrun is designed specifically to be used in another tty, so no, bbswitch is mandatory for that. That said, I also had the auto power on issue, it turned out to happen when I unplug my laptop, removing I still eventually had to use bbswitch though due to the said zombie process issue. |
Oh lol I just did the same in #139 -_- |
Funny how the implementation is the same too lol
|
Your implementation is fine, and what I'd be using if it got merged. My previous comment was about the handy reality that we wouldn't need a systemd service at boot to turn off the card, like the one used right now by nvidia-xrun |
@JpegXguy ah yeah that's right. Though relevant to your issues I'm also having a problem where GPU fails to unload modules unless I close all other running sessions. So running in a tty would end up failing to unload modules unless I am signed out of other sessions. Not sure if you had that issue |
Something makes me think gdm or something is locking the modules as soon as they are loaded |
per my observation it happens only if you try to switch to running session
when you have nvidia-xrun running.
Though I didn't test it throughly and it might require more testing.
…On Thu, 24 Oct 2019, 12:49 Yousef Sultan, ***@***.***> wrote:
Something makes me think gdm or something is locking the modules as soon
as they are loaded
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
<#137?email_source=notifications&email_token=AABSW2MNRLR3373ZTPMIKPDQQFVSDA5CNFSM4JBMRAE2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOECEN2KY#issuecomment-545840427>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AABSW2L24WPK357GHKRAHF3QQFVSDANCNFSM4JBMRAEQ>
.
|
I had this when I tried using nouveau for the powermanagement. |
Ahh yes, you both were right! I'll make another PR to put this in README as well |
On my system, both methods crash and do not turn off GPU correctly (it stays on, despite being not even visible in
lspci
). Other times it even causestee
to become a zombie process using 100% of CPU forever, making me have to restart my entire machine.After some trial and error,
bbswitch
seemed to be the only stable and working solution. Theturn_off_gpu.sh
script bundled withacpid
didn't work either.Thus, I implemented this feature and tested it on my machine, so far it works flawlessly without issues.