AMD Delivers Many Fixes For Polaris GPUs On Linux - Finally Enables ZeroRPM Fan Mode

[pqwoerituytrueiwoq]

Is there a clean way to get rid of 0 RPM mode? My cooling solution does not support this feature and does not work as intended (0% PWM = 100% fan)
* The 5.11 kernel was recently pushed out to Ubuntu 20.04
if there is no kernel feature to disable this feature where do i file a bug report

[amdtesterman]

Michael Have it been merged into linux for 5.11 at the end or not?
I didn't see any problems but maybe I'm not in the right place to watch this! Thanks in advance

[AsuMagic]

In fairness, dwagner has put a lot of effort into careful bug reporting, interacting with developers and testing potential fixes. I'm starting to think that there might just be a bad board causing the issues dwagner is seeing since we are having so much trouble reproducing them in-house.

EDIT - re-reading I guess your post is probably sarcastic so feel free to ignore the above. First read I thought you were criticizing dwagner.

sorry if the intent wasn't clear. to clarify, i was just annoyed in general by people telling others to fix code themselves when they bring up an issue they have (makes sense for smaller, unpaid projects, certainly, or for detailed bug reports in this context).
which was seemingly not the case in the end anyway because i misinterpreted what sandy said. my bad

[bridgman]

oh, so you have issues with a card you bought for over a hundred dollars? just get paid 0$ spending a few tens of hours of your time fixing a multi-billion dollar company's code on an officially supported platform!

i have a 580 and i don't really have any such stability issues, but ffs can people just quit with this horrible argument?

In fairness, dwagner has put a lot of effort into careful bug reporting, interacting with developers and testing potential fixes. I'm starting to think that there might just be a bad board causing the issues dwagner is seeing since we are having so much trouble reproducing them in-house.

EDIT - re-reading I guess your post is probably sarcastic so feel free to ignore the above. First read I thought you were criticizing dwagner.

[dwagner]

I wouldn't really count on stable Xe drivers from Intel, the current crop of drivers for integrated GPUs have various stability issues as well. Intel also famously broke stability for, well, pretty much everyone, like a year ago, and it took them months to fix.

Since among the computers I use are also ones with Intel iGPUs, I can tell about my experience: None of those ever had any significant stability issue, I found only one rare configuration (switching off a laptop's internal display, only enabling an external one, then sending the laptop into S3 sleep and resuming a few times) at which I once had issues with the Intel driver sometimes crashing - but that was easy to avoid, and not in the slightest way as disastrous as the "crashes every other day while idle or under light desktop use" that I had with amdgpu now for 4 years, with the stability getting just worse, not better.

Even if it took Intel 6 months to stabilize Xe drivers after the hardware is being sold, that would still be paradise in comparison to the stability nightmare that I experience with amdgpu.

I wouldn't really count on stable Xe drivers from Intel, the current crop of drivers for integrated GPUs have various stability issues as well. Intel also famously broke stability for, well, pretty much everyone, like a year ago, and it took them months to fix.

???? What stability problem are you talking about? Been using Intel GPUs for several years now, no stability issues here. Definitely none in the last year. Maybe you were specifically affected, but your claim that everyone was affected certainly seems false.

oh, so you have issues with a card you bought for over a hundred dollars? just get paid 0$ spending a few tens of hours of your time fixing a multi-billion dollar company's code on an officially supported platform!

i have a 580 and i don't really have any such stability issues, but ffs can people just quit with this horrible argument?

Oh trust me I totally agree and feel the same way. But, AMD hadn't bothered fixing many of the issues I had with my card. And the reality is that all software will always have bugs, and many of them that affect you will go unfixed because the developers don't have the same priorities that you do.

Instead of waiting around for someone else to hopefully, maybe fix your bug, you can take action yourself and just get it working. Don't need permission or approval for that.

Or you could just wait until Intel releases proper dGPUs with proper open source driver support that's ready in time and just works, and switch to using those. Atleast, that's my plan.

[AsuMagic]

Best way to get things working is to fix it yourself.

oh, so you have issues with a card you bought for over a hundred dollars? just get paid 0$ spending a few tens of hours of your time fixing a multi-billion dollar company's code on an officially supported platform!

i have a 580 and i don't really have any such stability issues, but ffs can people just quit with this horrible argument?

[brent]

I wouldn't really count on stable Xe drivers from Intel, the current crop of drivers for integrated GPUs have various stability issues as well. Intel also famously broke stability for, well, pretty much everyone, like a year ago, and it took them months to fix.

I should have known better than to waste my time applying the patch-set to amd-staging-drm-next and testing a kernel compiled from it.
The resulting kernel crashed in the amdgpu driver when running my "standard" 3-fps-video-playback test-script after about 20 seconds.
I really should have known better, given that for about 4 years now I have seen nothing but a decline in stability of amdgpu driving my Polaris based XFX RX 460 GPU.
Still waiting for Intel to release its first "Xe" discrete GPU to get out of this tragedy.

Best way to get things working is to fix it yourself.