The AMD Ryzen Thermal / Power Linux Reporting Improvements Working Well - V2 Up For Testing

[phoronix]

Phoronix: The AMD Ryzen Thermal / Power Linux Reporting Improvements Working Well - V2 Up For Testing

A few days ago I reported on AMD's "k10temp" Linux kernel driver finally seeing the ability to report CCD temperatures and CPU current/voltage readings as a big improvement to this hardware monitoring driver. The work hasn't yet been queued for inclusion into the mainline kernel, but initial testing is working well and a second revision to the patches has been sent out...

http://www.phoronix.com/scan.php?page=news_item&px=AMD-k10temp-Better-Zen-V2

[geearf]

Does this patchset provide anything different from zenpower?

[Ropid]

Does this patchset provide anything different from zenpower?

It basically is providing the same. The comments in the code say that the info about how to get the data out of the CPU hardware came from looking at zenpower's code.

A difference is, this k10temp driver only shows voltage and current readings while the zenpower driver also shows a reading for power usage.

I guess the power usage reading is left out because the hardware really only reports voltage and current. The zenpower driver calculates its power usage reading by multiplying the hardware's voltage and current readings.

[shmerl]

It basically is providing the same. The comments in the code say that the info about how to get the data out of the CPU hardware came from looking at zenpower's code.

How did zenpower author implement it? Is it documented anywhere?

[da_habakuk]

How did zenpower author implement it? Is it documented anywhere?

look at: https://github.com/ocerman/zenpower/...ter/zenpower.c

[shmerl]

look at: https://github.com/ocerman/zenpower/...ter/zenpower.c

Not so clear. It references Librehardware monitor and some experimental formulas. So sounds like partially reverse engineered? Not sure where Librehardware monitor got it.

The main question is, why can't AMD provide patches for for k10temp with all this? It's totally something they should have been doing even before CPUs came out.

[geearf]

It basically is providing the same. The comments in the code say that the info about how to get the data out of the CPU hardware came from looking at zenpower's code.

A difference is, this k10temp driver only shows voltage and current readings while the zenpower driver also shows a reading for power usage.

I guess the power usage reading is left out because the hardware really only reports voltage and current. The zenpower driver calculates its power usage reading by multiplying the hardware's voltage and current readings.

Nothing to really look forward to then :/

Thank you very much!

[rene]

Nothing to really look forward to then :/

Thank you very much!

Other than proper upstream kernel integration and not betrothing out-of-tree stuff?

[rene]

I can confirm it show something somewhat plausible on Ryzen 9 3950x: https://www.youtube.com/watch?v=SNwh9VgynK8

[jrdoane]

This allows core complex tie temperature reporting

Typo: I think you mean die.