facebook rss twitter

AMD responds to motherboards misreporting power telemetry

by Mark Tyson on 12 June 2020, 12:31

Tags: AMD (NYSE:AMD)

Quick Link: HEXUS.net/qaemcb

Add to My Vault: x

Yesterday, in the wake of reports that AMD motherboard vendors have applied BIOS firmware to purposely misreport key power telemetry, AMD issued a statement to Tom's Hardware. It said that it has seen these reports, and that it is going to investigate further. The main worry is that while motherboard vendors have made these changes to squeeze the best performance from AMD silicon, a side effect could possibly be felt in processor stability or longevity.

Winding back a little, this story emanated from a hardware analysis and reporting feature introduced by the developers of HWinfo recently. The feature is dubbed Power Reporting Deviation, and in the words of the authors it "tells how much the CPU telemetry seen by the CPU differs from real world (expected) data". Furthermore, HWinfo fields an accusation against vendors, saying the deviation "is usually caused by mainboard vendor (often intentionally) providing wrong calibration data in BIOS (AGESA) to fool the CPU to run at a higher power than the limit for the SKU".

Of course the motherboard market is very competitive, better performance sells, and the manufacturers appear to be indulging in one-upmanship, by hook or by crook. In effect these motherboard vendors are secretly overclocking your AMD processor, by fiddling the data sent to AMD's Adaptive Voltage and Frequency Scaling (AVFS) algorithms.

HWinfo devs observed the above CPU power reporting deviation on a Ryzen 9 3900X running Cinebench

AMD's response is to the above revelations at this time is to soothe user-concerns. It says that while it is looking into the accuracy of the reports, AMD processors contain an array of safeguards which should still maintain CPU safety and longevity. For clarity the full statement is below:

"We want to be clear with our customers: AMD Ryzen processors contain a diverse array of internal safeguards that operate independently of external data sources. These safeguards enforce the safety and reliability of the processor during stock operation. Based on our initial assessment, we do not believe that altering external telemetry in the manner described by those public reports would have a material impact on the longevity or safety of a user's processor."

Going forward it will be interesting to see if further investigations by AMD finds that these motherboard firmware misreporting tweaks are indeed damaging, and whether its board partners will continue to implement them now this info is in the open. Lastly, change will depend on whether AMD puts pressure on board partners to alter their behaviour.



HEXUS Forums :: 14 Comments

Login with Forum Account

Don't have an account? Register today!
AMD has put a ton of R&D into power management and operational analysis for PBO and XFR, it would seem unlikely that the motherboard would have that much control over how the processors environmental analysis.

It would be interesting into seeing a proper analysis of misreporting versus properly reporting and if there really is a perf difference.
“These safeguards enforce the safety and reliability of the processor during stock operation.”

What is considered to be stock operation though, if the board is feeding bad/incorrect info to the CPU, does that mean its not stock…?
“Could shorten CPU life” is needlessly provocative when the result is just a raised power limit. I bet you don't word it like that when Intel boards push tau to infinity.

Predictably the number's being fully misunderstood in the official HWinfo thread about it, even after repeated explanations.
'[GSV
Trig;4219639']“These safeguards enforce the safety and reliability of the processor during stock operation.”

What is considered to be stock operation though, if the board is feeding bad/incorrect info to the CPU, does that mean its not stock…?

CPUs have internal power draw limiters so just because the source is providing one value, the CPU itself will be observing another. The CPU itself will also only rely on external variables in a limited fashion.

Basically, what stock means is that AMD provided a minimum specification to mobo manufacturers that the boards have to be able to provide X amount of watts to the CPU, at stock a CPU will only really be drawing that amount. What the mobo manufacturers are doing are telling the CPU that mobo is being underutilised and therefore increases the potential headroom for something like PBO etc.

Except at stock, PBO isn't enabled IIRC?

I'd like someone to correct me if i'm wrong, but the above seems logical?
Tabbykatze
AMD has put a ton of R&D into power management and operational analysis for PBO and XFR, it would seem unlikely that the motherboard would have that much control over how the processors environmental analysis.

It would be interesting into seeing a proper analysis of misreporting versus properly reporting and if there really is a perf difference.

https://www.anandtech.com/show/15839/electromigration-amd-ryzen-current-boosting-wont-kill-your-cpu