AMD Frequency Invariance Support Comes With Linux 5.11

[phoronix]

Phoronix: AMD Frequency Invariance Support Comes With Linux 5.11

The previously reported on work for frequency invariance calculations for AMD CPUs with a focus on the AMD EPYC 7002 series has been merged for Linux 5.11 as part of the "sched/core" material...

http://www.phoronix.com/scan.php?page=news_item&px=AMD-Freq-Invariance-Linux-5.11

[Guest]

This support for server CPUs is nice and all, but the lack of proper support for consumer products is disappointing. I genuinely regret purchasing AMD hardware.

[cbxbiker61]

This support for server CPUs is nice and all, but the lack of proper support for consumer products is disappointing. I genuinely regret purchasing AMD hardware.

I'm puzzled about this? Do you know if Intel on Linux does this? And a "new" feature being rolled out for Server processors first is bad, why?

[agd5f]

As far as I know, this is not specific to server, but regardless, the Linux server market is much larger than the Linux desktop market.

[intelfx]

I'm puzzled about this? Do you know if Intel on Linux does this?

Oh yeah, Intel’s support of power management on Linux is generally very good through all ranges of their products. (A notable exception would be the DPTF mess leading to crippling levels of throttling with ultrabooks.) The logical equivalent would be the intel_pstate cpufreq driver — it’s been there how long, 8 years I think?

And a "new" feature being rolled out for Server processors first is bad, why?

Because it’s not “servers first”, but rather “servers sometime, consumers never”.

[Guest]

I'm puzzled about this? Do you know if Intel on Linux does this? And a "new" feature being rolled out for Server processors first is bad, why?

It's not just a feature, it's support for better for better process scheduling, and hence better performance/power management. And such features also also never gets rolled out to the consumer version.

I tried to get the amd_energy driver enabled for desktop Zen 2 CPUs, but apparently some firmware/driver bug that results in bad energy values for Zen 2 desktop CPUs still hasn't been fixed 1.5 years after release. Even all of the voltage and temperature monitoring was done by the community or other companies, after hardware release. AMD didn't bother doing so themselves. And 1.5 years after Zen 2 release, they still haven't bothered. They just don't give a crap about consumer hardware.

[Guest]

Because it’s not “servers first”, but rather “servers sometime, consumers never”.

Exactly! That's the problem. If I was running a business)data center and needed server hardware, sure I'd pick EPYC (i.e the already supported stuff). But AMDs support for their consumer hardware (whether Windows or Linux) is pretty crap.

[cbxbiker61]

The logical equivalent would be the intel_pstate cpufreq driver — it’s been there how long, 8 years I think

The question isn't whether Intel has scheduling support (AMD has had scheduling support for a long time too).

The question is whether or not it uses a frequency invariant strategy. If Intel does, I'd sure like to see the documentation to back that up.

[Linuxxx]

The question isn't whether Intel has scheduling support (AMD has had scheduling support for a long time too).

The question is whether or not it uses a frequency invariant strategy. If Intel does, I'd sure like to see the documentation to back that up.

Sure:

https://www.phoronix.com/scan.php?page=news_item&px=Linux-5.7-Schedutil-P-State

You & the one who liked your post should better keep up with your Phoronix reading...