Power Reporting Deviation (Accuracy)

[xBLUEDOGx]

 Are these values normal? If not, what do i need to do?

[Guest]

Quote

Power Reporting Deviation is a new feature available on AMD Ryzen CPUs that tells how much the CPU telemetry seen by the CPU differs from real world (expected) data. This value has a useful meaning only under full CPU load and values around 100% (95 - 105 %) mean the telemetry is working correctly. On systems with a higher deviation under full load this means the CPU thinks it's working at lower or higher power than expected for the given SKU, hence out of specification. This is usually caused by the 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.

https://www.guru3d.com/news-story/hwinfo-application-beta-introduces-power-reporting-deviation-sensorcpu-lifespan-reducing-mobo-enhancements.html#:~:text=The HWInfo tool now introduces,real world (expected) data.

[Mark Kaine]

1 hour ago, ShrimpBrime said:

caused by the mainboard vendor (often intentionally)

Sounds like my MSI board alright! 

 

So the only fix is buying a 300$ mobo to save my 200$ cpu?  Lol.

 

I'm so going intel next time, I had it with AMD's shenanigans.  (and yeah that's on them,  they apparently allow it ...)

[Guest]

1 minute ago, Mark Kaine said:

Sounds like my MSI board alright! 

 

So the only fix is buying a 300$ mobo to save my 200$?

 

I'm so going intel next time, I had it with AMD's shenanigans.  (and yeah that's on them,  they apparently allow it ...)

There has been some talk in OC forums about this issue/non issue. 

Basically it's a bios level tweak, has nothing to do with AMD actually. You want to blame the guys that write the bios's for these boards.

 

So then you would try different bios revisions to see which one works best within the Cpu specifications, obviuosly the reading is only accurate under a 100% load and would likely have around a 5% give or take. So 95 to 105% would be within the normal range.

 

 

[Mark Kaine]

39 minutes ago, ShrimpBrime said:

There has been some talk in OC forums about this issue/non issue. 

Basically it's a bios level tweak, has nothing to do with AMD actually. You want to blame the guys that write the bios's for these boards.

 

So then you would try different bios revisions to see which one works best within the Cpu specifications, obviuosly the reading is only accurate under a 100% load and would likely have around a 5% give or take. So 95 to 105% would be within the normal range.

 

 

I'm just saying if the mobo guys do this on purpose, AMD should take their license...  of course AMD isn't going to do that since from their perspective it'd be like shooting their own foot  - but that doesn't make it right, imo.

 

PS: I just checked , my guy says actually 100% at 100% load...  that was a great explanation,  and makes sense too !  When I saw this recently it was all over the place (like 500%) and didn't make any sense to me. 

 

 

[Guest]

2 minutes ago, Mark Kaine said:

I'm just saying if the mobo guys do this on purpose, AMD should take their license...  of course AMD isn't going to do that since from their perspective it'd be like shooting their own foot  - but that doesn't make it right, imo.

 

PS: I just checked , my guy says actually 100% at 100% load...  that was a great explanation,  and makes sense too !  When I saw this recently it was all over the place (like 500%) and didn't make any sense to me. 

Well that one is pretty much out of my hands. AMD may not know or care much about it. If it cause actual cpu failures, that would be a different story.

 

After all these are just reports from software from systems that use an algorithm vs a direct reading.

And the temps and voltage fluctuations you see already happened about 1000ms's ago. So none of this is actually in real time either. 

So for example, when a Cpu reaches therm trip temp and the board shuts off, you have like a 10% chance (at best) of actually seeing the temp it shut down at.

 

[Mark Kaine]

53 minutes ago, ShrimpBrime said:

Well that one is pretty much out of my hands. AMD may not know or care much about it. If it cause actual cpu failures, that would be a different story.

Yeah, true,  maybe it's not as bad as it sounds and generally there doesn't seem to be a widespread issue with CPUs dying either...