AMDGPU Linux Driver No Longer Lets You Have Unlimited Control To Lower Your Power Limit

[stormcrow]

...some commenters apparently fell asleep in high school science classes...

Basic high school electricity. Under voltage beyond tolerance will kill electronics nearly as fast as an over voltage (where you get arcing). Under voltage increases your amperage to meet the basic power levels required by the electronics. Electronics are rated to a certain voltage, but more importantly, to a certain amperage. When that amperage is exceeded Bad Things happen. Ever wondered why high amperage extension cords are much larger and more expensive than low amperage for the same voltage? (Look it up.) Additionally left to your education, find out why weak(ening) PSUs often scorch power traces on connected boards. (Hint: Power (Watts) = V (voltage) x I (current or amperage) )

Don't expect this to ever be reverted. It was an oversight/bug to begin with.

[stormcrow]

Because "safety".

"You'll shoot your eye out kid"

And he did.

[piotrj3]

Seriously what the?

Nvidia doesn't have problem lowering power limit far (for example my 4070 ti super officially allows lowering power limit up to 36% of orginal TDP. Previous card could 40%. It works as intended.

Now in AMD you have a situation when either power limit you have lower by just few percent, and if you lower it way too far you suddenly can pull 60-70% more than official TDP allows and potentially burn the card. Lmao.

[kiffmet]

AMDs argumentation is nonsensical given that identical chips that are used in notebooks can run at much lower power without entering self-destruct mode.

[xpris]

Guys, just ask your distro kernel maintainers to revert that and I think most of distros that care about users voices can "fix" this.

[Daktyl198]

Isn't setting a lower power cap equivalent to setting a lower maximum gpu clock?

At some point, yes. You can run at the standard clocks up to a point, but eventually the lower power limit is going to impact your performance as the chip can't draw the minimum required power to do what it needs to do.

As for the other comments here, I highly doubt AMD would impose these limits if it wasn't at least somewhat necessary. It's well known that weird/dirty power can damage electronics over time, and I assume running a GPU at too low of a power draw has led eventually to more customer support claims than AMD is happy with. Especially with the undervolting craze that started with the RX400 series which had insane room for lower voltages AND higher clock speeds.

As far as I'm aware, NVidia also has limits on its power limit slider, they're just generally wider range as nvidia has a bit more of a stable architecture.

[skeevy420]

Especially with the undervolting craze that started with the RX400 series which had insane room for lower voltages AND higher clock speeds.

It was necessary with my RX 580. That GPU ran hot enough to thermally crash my PC; above 90C under load. I eventually found a stable undervolt (no overclock) that lowered the temps to 84C under load and created a custom bios because that's 1000x easier than trying to UV ever damn OS I run. I never had a GPU thermal related crash afterwards. I should add that part of the problem was needing to add more case fans. Adding them took my 580s 84C down to 80C. I wonder if they'd have prevented thermal crashing when it went over 90C?

My 6700 XT hits around 70C under extreme loads so I've never bothered with it.

[jeisom]

Isn't setting a lower power cap equivalent to setting a lower maximum gpu clock?

No. Think about how the multi core cpus are. Single core you get 5GHz, but multi core you get like 4.4Ghz. Both are against the power limit. Same can be done with gpus. Furmark for instance will just try to use all the power even getting lower clocks. Depending on your program the gpu can be loaded in such a way than scales across compute units well or not. So say the program used 100% of the resources available at a powerlimit might have a clock of 2000mhz vs something that only uses 50% may get a clock of 2800mhz at the same powerlimit.

[Britoid]

AMD disabled something that was never supposed to be possible and could damage hardware?

Yeh, screw AMD cause it's not like themselves or hardware partners would be responsible for warranty replacements or fire hazard caused by this, oh wait.

Shame on you AMD for doing the legal and responsible thing.

[ssokolow]

AMDs argumentation is nonsensical given that identical chips that are used in notebooks can run at much lower power without entering self-destruct mode.

My reading of that "the minimum limits established by the graphics card AIB partner" line was that it's not the chips that are the risk, but limits imposed by the rest of the components the board builder integrates them with.

(Sort of like how my $10 USB 1.1 isolator isn't going to give me 2500V of isolation just because one of the chips isolating the data lines is rated for that much, because the DC-DC converter/isolator for the 5V power lines is a much weaker link with a much smaller gap between the traces on its two sides.)