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

According to https://www.gigabyte.com/Motherboard/B550-AORUS-ELITE-V2-rev-10/support#support-cpu the CPU is supported since the very first BIOS version. CM Hyper 212 cooler. It's not anything magical but still better than stock. ~35C idle temps. Full load is ~80C. With accurate power reporting full load temps would probably be lower and clocks higher.
It’s not exactly how it works. Higher clock requires more voltage and that would increase power and current(A) under load and temp eventually.

Yes the cooler is not the best, it’s considered a “budget” one, but still ambient temp and case airflow are also in play. It’s very common for Ryzens 3000/5000 to reach easily the 80~85C region because of their thermal density and their off centered position of core chiplets.
 
Hi folks, I'm another one with very strange power reporting results. Just thought i'd throw my results into the mix as well.

R5 3600 on an Aorus Elite V2 (rev 1) All stock bios settings for CPU. XMP 1 for memory @3200. (infinity fabric 1600) Pic shows HWinfo while Cinebench R23 is running multi core test. CPU temp maxes out around 71c (idles at 24ish)Power rep test.png
 
Hi folks, I'm another one with very strange power reporting results. Just thought i'd throw my results into the mix as well.

R5 3600 on an Aorus Elite V2 (rev 1) All stock bios settings for CPU. XMP 1 for memory @3200. (infinity fabric 1600) Pic shows HWinfo while Cinebench R23 is running multi core test. CPU temp maxes out around 71c (idles at 24ish)View attachment 6115
You must have really low ambient and the CPU cooler is not the stock one. Maybe using AIO?
Anyway your CPU's true power consumption (PPT) as shown (current value) is 75.7 / 0.695 = 109W. Stock PPT for R5 3600 is 88W.

Did you updated the BIOS to the latest?
At least to the F12 for your board (AGESA Combo V2 1.1.0.0 D).
 
You must have really low ambient and the CPU cooler is not the stock one. Maybe using AIO?
Anyway your CPU's true power consumption (PPT) as shown (current value) is 75.7 / 0.695 = 109W. Stock PPT for R5 3600 is 88W.

Did you updated the BIOS to the latest?
At least to the F12 for your board (AGESA Combo V2 1.1.0.0 D).
Yup, I'm on F12. As for temps, yes pretty low ambient temps and an aftermarket fan cooler in an airflow priority case... very happy with the temps, not so happy with the cpu power consumption!
 
Yup, I'm on F12. As for temps, yes pretty low ambient temps and an aftermarket fan cooler in an airflow priority case... very happy with the temps, not so happy with the cpu power consumption!
We're in the same boat. We both have B550 Aorus Elite v2 boards and our PRD is the same.
 
Yes I think that Gigabyte should address this with this specific board, but on the other hand you all should not worry about it too much. Even if you did run 100% load for a living, and I don’t mean benchmarks. Even if the CPU is exceeding its stock limits under 100% load, doesn‘t mean it would be damaged. It’s still under its internal silicon limits by the silicon FITness controller. The board all it does is to give it false feedback so the CPU “thinks“ it’s drawing less power than it really does, and by this it’s pushing it self (beyond AMD rated specs) to the point it can without compromising longevity. Temperature is the key for what this point would be. As I said before it’s like having PBO Enabled and let it do it’s thing.

When running simple every day tasks, including gaming, its behavior is not affected by this discrepancy. So if you don’t like this just don’t run 100% loads unless you have to.
 
Yes I think that Gigabyte should address this with this specific board, but on the other hand you all should not worry about it too much. Even if you did run 100% load for a living, and I don’t mean benchmarks. Even if the CPU is exceeding its stock limits under 100% load, doesn‘t mean it would be damaged. It’s still under its internal silicon limits by the silicon FITness controller. The board all it does is to give it false feedback so the CPU “thinks“ it’s drawing less power than it really does, and by this it’s pushing it self (beyond AMD rated specs) to the point it can without compromising longevity. Temperature is the key for what this point would be. As I said before it’s like having PBO Enabled and let it do it’s thing.

When running simple every day tasks, including gaming, its behavior is not affected by this discrepancy. So if you don’t like this just don’t run 100% loads unless you have to.
Thanks for explaining that clearly :) Puts my mind at rest.
 
Hi, I didn't pay much attention till yesterday, but I really need some help with the numbers reported in Power Deviation.
They are really high, do you think I should do or try sthg.? Feel free ask me anything that might help me solve this issue.
I have attached a screenshot while IDLE, but there can be seen some even bigger numbers at a certain point.

BTW this is my RIG:

Lian Li TU150x
ASRock B550 PG ITX/ax
Ryzen 9 3900x
ID-Cooling SE-224-XT Black
ASRock RX6800 Reference Design
Corsair LPX 3600 32GB
AORUS NVMe Gen4 2TB
PNY CS3030 NVMe 2TB
HDD WD Blue 4TB
Corsair SF750
Redragon Vishnu K596
Logitech G604
Samsung UJ590 32"
Huion GT-221 Pro
Scarlett 2i2_Yamaha HS5

Thanks in advance!
 

Attachments

  • InkedCaptura de pantalla 2021-04-08 140050_LI.jpg
    InkedCaptura de pantalla 2021-04-08 140050_LI.jpg
    222.4 KB · Views: 1
Hi, I didn't pay much attention till yesterday, but I really need some help with the numbers reported in Power Deviation.
They are really high, do you think I should do or try sthg.? Feel free ask me anything that might help me solve this issue.
I have attached a screenshot while IDLE, but there can be seen some even bigger numbers at a certain point.

BTW this is my RIG:

Lian Li TU150x
ASRock B550 PG ITX/ax
Ryzen 9 3900x
ID-Cooling SE-224-XT Black
ASRock RX6800 Reference Design
Corsair LPX 3600 32GB
AORUS NVMe Gen4 2TB
PNY CS3030 NVMe 2TB
HDD WD Blue 4TB
Corsair SF750
Redragon Vishnu K596
Logitech G604
Samsung UJ590 32"
Huion GT-221 Pro
Scarlett 2i2_Yamaha HS5

Thanks in advance!
Reread https://www.hwinfo.com/forum/thread...ng-deviation-metric-in-hwinfo.6456/post-25045
 

Not really sure, but I guess that you've send me to that post because I took the measurements while on IDLE (to tell you the truth I do not fully grab the whole text, mostly because english ain't my first language, so I've got a bit confused). Anyway, my bad, I allways thought that while bieng in IDLE was the way check it out. I've just checked the results with Cinebench running and it seemed fine (I've attached a JPG). Thanks man if that was the intention behind when directing me to that post.

PD: Not really related to this thread, nor HWINFO, but I find my chip a bit underwhelming regarding the Cinebench resultas. It's hovering around 17500 and slightly above, while I've seen that most people is arround 18500 and above. Someone here witha 3900x? should I try some setting in BIOS?
 

Attachments

  • InkedCaptura de pantalla 2021-04-09 011937_LI.jpg
    InkedCaptura de pantalla 2021-04-09 011937_LI.jpg
    552.1 KB · Views: 3
Not really sure, but I guess that you've send me to that post because I took the measurements while on IDLE (to tell you the truth I do not fully grab the whole text, mostly because english ain't my first language, so I've got a bit confused). Anyway, my bad, I allways thought that while bieng in IDLE was the way check it out. I've just checked the results with Cinebench running and it seemed fine (I've attached a JPG). Thanks man if that was the intention behind when directing me to that post.

PD: Not really related to this thread, nor HWINFO, but I find my chip a bit underwhelming regarding the Cinebench resultas. It's hovering around 17500 and slightly above, while I've seen that most people is arround 18500 and above. Someone here witha 3900x? should I try some setting in BIOS?
First, your PRD is almost perfect and as you understood it only makes sense under 100% CPU load and not at idle or even at middle load.

Second, your CPU's metrics under the CB R23 run are normal. The only thing that you could improve is the cooling. Ryzen 3000/5000 CPUs are very temp sensitive when boosting. Maybe the other 3900Xs you are referring to are running cooler and so boosting more.
If you look that screenshot of yours you'll see the CPU Precision Boost (PB) values reaching stock limits under normal operation.
PPT: 142W
EDC: 140A

Depending on your usage case scenario you could do one or both(best) of the following
1. Improve cooling
2. Enable PBO (Precision Boost Overdrive). Now it must be on Auto and by design its Disabled. If you Enable PBO then the above stock limits (PPT/EDC) will be increased to a much higher values. Then the CPU will try to run to max performance and the only limit will be its internal limits that the board can't control.

Given the temp level your CPU is operating PBO won't make much of a difference if you can't improve cooling.

If your usage is gaming or other tasks that don't involve a 100% load then leave it as it is.
 
Not really sure, but I guess that you've send me to that post because I took the measurements while on IDLE (to tell you the truth I do not fully grab the whole text, mostly because english ain't my first language, so I've got a bit confused). Anyway, my bad, I allways thought that while bieng in IDLE was the way check it out. I've just checked the results with Cinebench running and it seemed fine (I've attached a JPG). Thanks man if that was the intention behind when directing me to that post.

PD: Not really related to this thread, nor HWINFO, but I find my chip a bit underwhelming regarding the Cinebench resultas. It's hovering around 17500 and slightly above, while I've seen that most people is arround 18500 and above. Someone here witha 3900x? should I try some setting in BIOS?
You got it. For your 2nd question, that's a tough one. I have the same thing, but my power deviation isn't anywhere near as good as yours. Do an indepth comparison with whoever is getting the additional 1,000 points. It's probably memory timing. Comparing my CBR20 results with others, I think results can vary by these sorts of amounts due to the motherboard used, implementation details for each. I like what I see in your screenshots though, I don't see a problem worth looking into. I'd use it and enjoy it, if you really want to play around with it try flashing older vs newer BIOS versions and check for any differences. Just be sure to clear the CMOS RAM if you do flash around, and be warned once you're on the latest ComboAM4v2 AGESAs you can't flash around any longer, they've put in read/write protections.
 
You got it. For your 2nd question, that's a tough one. I have the same thing, but my power deviation isn't anywhere near as good as yours. Do an indepth comparison with whoever is getting the additional 1,000 points. It's probably memory timing. Comparing my CBR20 results with others, I think results can vary by these sorts of amounts due to the motherboard used, implementation details for each. I like what I see in your screenshots though, I don't see a problem worth looking into. I'd use it and enjoy it, if you really want to play around with it try flashing older vs newer BIOS versions and check for any differences. Just be sure to clear the CMOS RAM if you do flash around, and be warned once you're on the latest ComboAM4v2 AGESAs you can't flash around any longer, they've put in read/write protections.

Really? I've thought I could roll back to older BIOS versions. There is a new BETA BIOS version with AMD AGESA ComboAM4v2 1.2.0.1 Patch A, I was thinking of trying it, but if I can't roll back to previous versiones I might avoid it for the moment.

PD: thanks BTW!
 
First, your PRD is almost perfect and as you understood it only makes sense under 100% CPU load and not at idle or even at middle load.

Second, your CPU's metrics under the CB R23 run are normal. The only thing that you could improve is the cooling. Ryzen 3000/5000 CPUs are very temp sensitive when boosting. Maybe the other 3900Xs you are referring to are running cooler and so boosting more.
If you look that screenshot of yours you'll see the CPU Precision Boost (PB) values reaching stock limits under normal operation.
PPT: 142W
EDC: 140A

Depending on your usage case scenario you could do one or both(best) of the following
1. Improve cooling
2. Enable PBO (Precision Boost Overdrive). Now it must be on Auto and by design its Disabled. If you Enable PBO then the above stock limits (PPT/EDC) will be increased to a much higher values. Then the CPU will try to run to max performance and the only limit will be its internal limits that the board can't control.

Given the temp level your CPU is operating PBO won't make much of a difference if you can't improve cooling.

If your usage is gaming or other tasks that don't involve a 100% load then leave it as it is.

Thanks for the help man!
 
Really? I've thought I could roll back to older BIOS versions. There is a new BETA BIOS version with AMD AGESA ComboAM4v2 1.2.0.1 Patch A, I was thinking of trying it, but if I can't roll back to previous versiones I might avoid it for the moment.

PD: thanks BTW!
https://www.overclock.net/threads/r...ocking-thread.1624603/page-2353#post-28763934
He's talking about Asus but I know it's in MSI board's BIOS updates too. I'm stuck on the latest bios for my board which broke things that worked before.. I'd ask around on your exact board before updating. I think it may be all AGESA 1.2.0.0 releases that has AmdSpiRomProtectDxE, but I'm not sure.
 
You got it. For your 2nd question, that's a tough one. I have the same thing, but my power deviation isn't anywhere near as good as yours.
PRD value doesn't say much by it self other than the board is giving false feedback to the CPU about its power consumption. Actually when the value is around 80~95% the CPU is exceeding its stock performance, or at least the board is "trying" to do so. That depends strongly on the values of PPT, EDC and temp during the 100% load.

Thanks for the help man!
Just to add,
Ryzen 3000 series has a temp operating limit at 95°C. This is the throttling point, reducing speed and voltage drastically but until that point it also cuts boost proportionally while getting near 95°C. This is mainly for all core loads.
Some rare test shown that the (all core load) boost cutting is starting around 50°C and its maximized at 94°C. Of course having a CPU cooling device that can keep a 140+Watt chip below 50~55°C is unrealistic for the majority of users. I would say having it around 70°C is a more feasible case scenario if someone is willing to spend the extra $$ for the cooling it requires. Still though getting under 80°C is perfectly fine and safe, and if the user doesn't really care for all core loads (like having a life dependency on it) can skip the more powerful cooling.

Also, Cinebench benchmark scores in general don't care to much about DRAM settings. Its mostly (99+%) a CPU calculation matter than memory performance. Its all about CPU speed/IPC/thread matter. Anything above 16GB 3200MHz CL16 kit in dual channel won't do anything really for CB R other than 1~2% that falls under margin of error. CPU boost does and that is primarily temp related.

For the BIOS flash matter to older version you can search your board specifications and look for a BIOS flash button. Almost all B550s have it and a lot of X570s. 400series boards not so much. With this you can force a BIOS flash no matter what. Even in case of BIOS corruption/damage that renders the board unbootable.
It requires a very specific procedure, defined(differently) by each board vendor. Its on the manual.
 
Back
Top