Can someone explain like I am 5

[keblin]

This damn Power Reporting Deviation (Accuracy), it seems everyone is confused as I see many threads on it.
I've read the important thread explaining it and I am still left confused if I have a problem or not lol.

I don't think I do but it is alarming when you see red. So when I play games this "Power Reporting Deviation (Accuracy)" is red and showing as 80% quite often. It only used to do it now and then but the past 2 weeks it's constantly done it!

My performance is fine, my temperatures are fine. If I didn't have this software I'd be none the wiser and have no clue there was anything odd going on.
So I am assuming it's probably fine? but honestly who knows and it's really bugging me to the point I wish I didn't have this software now or an AMD cpu lol.

Specs:
5600x
RTX 3070
32gb Ram
MSI Mag B550 Tomahawk

 

[Zach]

It’s very simple.
As the sensor description indicates this has a meaning while CPU is ONLY under 100% load.
Any other time, like gaming, is literally useless.
So unless you got any red under 100% CPU load there is nothing to be worry about.

 

[keblin]

It’s very simple.
As the sensor description indicates this has a meaning while CPU is ONLY under 100% load.
Any other time, like gaming, is literally useless.
So unless you got any red under 100% CPU load there is nothing to be worry about.

Oh my bad I am so silly I assumed my CPU was under full load when I was gaming but haha I realise now that wouldn't be the case.
So would running cinebench be 100% and if it's a bad result then something is wrong?

Speaking of, if it was wrong, what does it mean is wrong is it a cpu problem?
Thanks for the reply

 

[Zach]

1. Run a cinebench multicore test
2. While test is running take notes about the PowerReportingDeviation and CPU PPT values.
3. Post here and I will explain what is the meaning.

 

[keblin]

1. Run a cinebench multicore test
2. While test is running take notes about the PowerReportingDeviation and CPU PPT values.
3. Post here and I will explain what is the meaning.

Running it now, does it matter that the PowerReportingDeviation started on 89% Minimum
Should I restart the PC first? That usually makes the figure reset.

 

[Zach]

When CB is running you just observe the current value. It’s shouldnt fluctuate a lot, maybe a few %.
You can also hit the clock button down on right to reset all values, after test starts.

 

[keblin]

When CB is running you just observe the current value. It’s shouldnt fluctuate a lot, maybe a few %.
You can also hit the clock button down on right to reset all values, after test starts.

Ah ok thanks.

Attached photos during test and after the test finished + clicking the clock
Is this enough?

Or should I keep making notes during the test of the changes every so often?

 

[Zach]

Ok the useful one is during.
I will assume that CPU PPT and PRD were stable while the benchmark was running, at least within a small margin like 2-3%.

The R5 5600X has a PPT (PowerPackageTracking) value from AMD the 76W which is the default max power consumption for the entire CPU package.
The board hosting the CPU is responsible to measure CPU operating parameters like voltage, current, total power and I’m sure a few more and feed it back to CPU so that it can regulate it self accordingly to what AMD specifies.
A PRD value of 106% means that your board over-states (through feedback) the current that the CPU is actually drawing, by 6%.
And this has an impact of the actual CPU power draw.
This means that the true power consumption of the CPU (PPT) is ~6% less that the PPT value we see in that screenshot, and we can calculated it like this:

Benchmark values
PPT value: 75.998
PRD value: 106

75.998 / 1.06 = 71.696

So the true power consumption of the CPU is 71.696W

It is unknown why boards are doing this. Some of them overstate the current and other understate it. Some of them by a small amount (within +/-10%) and others by a lot (up to +/-20% or more).

 

[keblin]

Ok the useful one is during.
I will assume that CPU PPT and PRD were stable while the benchmark was running, at least within a small margin like 2-3%.

The R5 5600X has a PPT (PowerPackageTracking) value from AMD the 76W which is the default max power consumption for the entire CPU package.
The board hosting the CPU is responsible to measure CPU operating parameters like voltage, current, total power and I’m sure a few more and feed it back to CPU so that it can regulate it self accordingly to what AMD specifies.
A PRD value of 106% means that your board over-states (through feedback) the current that the CPU is actually drawing, by 6%.
And this has an impact of the actual CPU power draw.
This means that the true power consumption of the CPU (PPT) is ~6% less that the PPT value we see in that screenshot, and we can calculated it like this:

Benchmark values
PPT value: 75.998
PRD value: 106

75.998 / 1.06 = 71.696

So the true power consumption of the CPU is 71.696W

It is unknown why boards are doing this. Some of them overstate the current and other understate it. Some of them by a small amount (within +/-10%) and others by a lot (up to +/-20% or more).

I see thank you. I wonder if they understate to make it look like they use less power as that can be appealing. Overstating though, can't figure a reason for that one? lol

So based on my results above, I would say the margin is so small that it's probably nothing to worry about or do you think something is wrong here? As from my understanding that's not 100% but only slightly off so assume it's ok?

 

[Zach]

I see thank you. I wonder if they understate to make it look like they use less power as that can be appealing. Overstating though, can't figure a reason for that one? lol

So based on my results above, I would say the margin is so small that it's probably nothing to worry about or do you think something is wrong here? As from my understanding that's not 100% but only slightly off so assume it's ok?

Usually board vendors want to show to the market (consumers) that their board can run the CPU to the highest performance possible, as being the best on the market.
At least this is what’s happening with intel platform(s).
I seriously doubt that that’s the case with Ryzen CPUs.
After all the years I run my AM4 system I do not come to any conclusion why this is happening.
At this point I think no one can really answer this.
Some boards are overstating power, others are understating it, and even sometimes the same board is stating power differently on different CPU models.
Currently I run a 5900X, X570 system. On the same board previously I run R5 3600.
With that CPU PRD was around 90% and now with the 5900X PRD is 102% under full load.

Because most of times over/under stating is within the 10% people do not notice it unless you check HWiNFO sensors and start to wonder about this specific reading.
Some times though PRD is way off (like 125%) and user starts to wonder why his CPU underperforms in scores, and/or why core clocks are low and not hitting the advertised speed.

For example check the link below and start reading from post #553 a case with PRD at 125-128% and how the user was able to get the performance he paid.

IMPORTANT - Explaining the AMD Ryzen "Power Reporting Deviation" -metric in HWiNFO

Ryzen 5700x Asus ROG Strix B550 -F Gamig Wifi II Gskill 2x 16gb 3600 CL16 PSU Seasonic Core GC-650 Gold Only XMP turned on rest is stock/default in Bios Reading still tread about this and asking is this fine with my pc numbers in HW with cb23 multicore test for Ryzen 5700X?Thanks

www.hwinfo.com

 

[keblin]

Usually board vendors want to show to the market (consumers) that their board can run the CPU to the highest performance possible, as being the best on the market.
At least this is what’s happening with intel platform(s).
I seriously doubt that that’s the case with Ryzen CPUs.
After all the years I run my AM4 system I do not come to any conclusion why this is happening.
At this point I think no one can really answer this.
Some boards are overstating power, others are understating it, and even sometimes the same board is stating power differently on different CPU models.
Currently I run a 5900X, X570 system. On the same board previously I run R5 3600.
With that CPU PRD was around 90% and now with the 5900X PRD is 102% under full load.

Because most of times over/under stating is within the 10% people do not notice it unless you check HWiNFO sensors and start to wonder about this specific reading.
Some times though PRD is way off (like 125%) and user starts to wonder why his CPU underperforms in scores, and/or why core clocks are low and not hitting the advertised speed.

For example check the link below and start reading from post #553 a case with PRD at 125-128% and how the user was able to get the performance he paid.

IMPORTANT - Explaining the AMD Ryzen "Power Reporting Deviation" -metric in HWiNFO

Ryzen 5700x Asus ROG Strix B550 -F Gamig Wifi II Gskill 2x 16gb 3600 CL16 PSU Seasonic Core GC-650 Gold Only XMP turned on rest is stock/default in Bios Reading still tread about this and asking is this fine with my pc numbers in HW with cb23 multicore test for Ryzen 5700X?Thanks

www.hwinfo.com

Thanks a lot, I will do a few more tests in future make sure it's still within margin.
I think I can rest easy knowing there is nothing majorly wrong with my hardware other than maybe not getting my full performance sometimes. If it becomes a big gap then I'd look into it more.