AMDGPU For Linux 5.1 Tweaks The Golden Settings For Vega, Corrects Fiji Power Reading

[aufkrawall]

Sounds rather bad. In idle, my MSI 580 Armor goes down to ~29°C on Linux and ~25°C on Windows with both 450rpm.
I also had an Asus 580 Dual card, which had a cheating bios regarding temperature thresholds and way worse fans than MSI. It also consumed more power by having lower clock and voltage at the same time. In my experience, Asus is the biggest trash manufacturer around in PC hardware business.

[aufkrawall]

And how would broken fan control be connected to wrong reported GPU power? Totally different topic, I'd say.
Needless to say that ZeroFan should work on Linux as well, at least it did so for an Asus Vega card I had briefly installed. It's probably just another notorious Polaris Linux issue, among others.

But I don't care anyway, as I use a custom fan curve without ZF on any OS.

[Brisse]

I have read that windows driver even disables the gpu fan. Windows destroys your hardware.

You mean the semi-passive fan control that is a feature of many modern graphics cards? That's not just Windows by the way, mine does that on Debian as well.

[dwagner]

Not only does the current amd-staging-drm-next head still crash quickly when performing the "3 fps video replay test", it also fails to allow setting the shader clock manually, keeping it at an unreasonable second-highest of its 6 different levels. So no progress in sight for me, just more regressions.

[shmerl]

If you mean this patch: https://lkml.org/lkml/2019/2/25/747

The AMD wip kernel does not have that: https://cgit.freedesktop.org/~agd5f/...t-5.2-wip#n644

Regular 5.0 didn't have it either, until it was disabled. I'm asking whether it's still planned for 5.1 or not.

[M@GOid]

Why in english people say "to X from Y" instead of the opposite?

Doesn't make sense for me as portuguese speaker, as we say "from Y to X", where Y is the initial value/position and X is new value/destination

Eles dirigem do lado errado da estrada. O que você esperava?

[Veerappan]

Why in english people say "to X from Y" instead of the opposite?

Doesn't make sense for me as portuguese speaker, as we say "from Y to X", where Y is the initial value/position and X is new value/destination

As a native English speaker, I almost always use "from X to Y", which is the same as your preferred method.

"To Y from X" makes sense, but it's definitely awkward to use and takes a little more mental effort for me to parse.

[Brisse]

That override temperature seems too high. I thought 14/10/7nm chips couldn't withstand that much heat?

That number is for the Radeon VII specifically which has a new way to measure temperature. Traditionally, temperature was read by a probe at the edge of the chip, which will read much lower than the hottest parts of the chip, but the Radeon VII has many probes all over the chip and will display the hottest part of the chip. Actually, I don't think the fact that it has many probes all over is new, but the fact that it displays the hottest spot to the user is new. Take care not to compare apples and oranges when comparing to other GPU's.

[tildearrow]

That override temperature seems too high. I thought 14/10/7nm chips couldn't withstand that much heat?

[andrei_me]

Why in english people say "to X from Y" instead of the opposite?

Doesn't make sense for me as portuguese speaker, as we say "from Y to X", where Y is the initial value/position and X is new value/destination