Z590 Elite Anomalous Current(IOUT) and Power(POUT) Readings

krunk

New Member
Hello!

I noticed that POUT and IOUT are much lower than they should be in 7.00, 7.02, and now 7.04 beta hwinfo64 releases. In the imgur album below(3 pictures), it can be seen that there is a 100w+ difference software readings and a Kill-A-Watt power meter that the computer is plugged into. This also happens in both BIOS versions available as well as if the BIOS is fully at default settings. The test ran to compare power and current usages was Cinebench R23.


Specs: Z590 Elite @ BIOS version F1 and F2
10900kf @ Stock & Overclocked to 5.4ghz/Override/1.4v/Hyperthreading off
2x8gb Gskill 3600 cl15-15-15 1.35v @ 4500mhz cl16-17-17 1.58v
GTX 660 OEM
Corsair 750w RMX

Here is someone else I know with a Z590 Elite and 10600k seeing the same incorrect behavior.



Here is someone I know with a Z590 Ultra seeing correct behavior. I had them run similar(ended up being lower) voltage and clockspeed with hyperthreading disabled(10900k/f).

I could not find an option to create a debug file under options, if someone could let me know how that would also be fantastic!
 
You don't understand the meaning of those values. They don't represent the power consumption of the entire system, but only one voltage rail (the CPU core). The whole system consists of many other rails that usually can't be monitored. To measure the power consumption of the entire system would require a PSU with such capability and check it on that side.
 
You don't understand the meaning of those values. They don't represent the power consumption of the entire system, but only one voltage rail (the CPU core). The whole system consists of many other rails that usually can't be monitored. To measure the power consumption of the entire system would require a PSU with such capability and check it on that side.
Go ahead and ignore the wall reported power as it is detrimental to my explanation of what's occurring as I'm seeing now.

#1
Here are two people with Asus motherboards that I've asked to run Cinebench with Hyperthreading disabled on a 10900k/f. Asus EC easily near 190a. Yes, different motherboards. This was to initially show that something was definitely off with the readings.

#2
This again is a repeat of the post above, the Z590 Ultra. Idle current of 25a, load current of 158a. I asked them to run with Hyperthreading disabled on their 10900k and they're showing 1.38v die-sense load voltage. As we all know, more voltage means more current due to Ohm's Law (I = V/R). Okay so we have someone with the same manufacture(Gigabyte) and the same ISL 69629's as my Z590 Elite.

#3
This is my setup with a 10900kf with Hyperthreading disabled with an idle voltage of 0a(notice how it's 25a for the Z590 Ultra) and a load current of 109a(notice how it is 158a for the Z590 Ultra). VR VOUT under load is at 1.42v, more than the Z590 Ultra setup with 1.38v die-sense load voltage. As we all know, more voltage means more current.

Now you may be thinking to yourself, maybe it's the cold temperatures in #3 that lower resistance and therefore current! Well, I found someone else with the same Z590 Elite motherboard and a 10600k.

#4
This is a Z590 Elite(same as the one I have with the issue) and a 10600k. Notice how they're experiencing the same 0a at idle, despite no other board going down to 0a at idle. I also had them run Cinebench r23 as is with their current 5ghz overclock without throttling as well as 1.323v load voltage with VR VOUT. According to POUT, only 51w was drawn. Even according to CPU Package Power(which I do realize is inaccurate for when you have access to better sensors) is only 10w. If I were to tell anyone else that an overclocked 10600k with hyperthreading enabled at 1.323v load voltage and 5ghz only shows at a max of 51w, I would be told I'm insane and that it's impossible.

There is an obvious power discrepancy with the Z590 Elite.


Now the real question is if this is a Gigabyte issue or one with Hwinfo.
 
You don't understand the meaning of those values. They don't represent the power consumption of the entire system, but only one voltage rail (the CPU core). The whole system consists of many other rails that usually can't be monitored. To measure the power consumption of the entire system would require a PSU with such capability and check it on that side.

Hi Martin,
I looked at his readings and there's definitely something off. I doubt it's anything to do with hwinfo. Has to be something causing the VRM to report strange readings, because it's impossible for for a 5.3 ghz HT off 10900k to report anywhere near that low of 109 amps at full load. And watts is a function of amps, so if the IOUT value is wrong, the POUT value follows it like a lemming.

The only thing I recall is Elmor mentioning some sort of power state interfering with VRM reporting. But that was only at *IDLE*, when using C-states and Intel auto voltages and the Internal AC/DC Loadlines. That was on a 9900k.

I have no idea what krunk's problem is here. All I can say is "Call in the doomsday squad", because this is beyond my knowledge.
 
I'm not sure either what's going on with that PWM. Let's ask some bigger experts in power delivery ;)
 
The difference might be due to some BIOS tweak settings. Try to switch the following options to default how it will behave then:
"Enhanced Multi-Core Performance" = Disabled
"CPU Upgrade" = Default
 
The difference might be due to some BIOS tweak settings. Try to switch the following options to default how it will behave then:
"Enhanced Multi-Core Performance" = Disabled
"CPU Upgrade" = Default


Here is the result, similar/same results. I'll update this post in a moment with more testing as well.

I tried the same thing but with a different profile and Hyperthreading enabled just for fun, no difference.

Here is the image of my BIOS for verification. (Thanks capture card!)
 
Bump! Issue still happening on latest BIOS and latest hwinfo beta(7.03-4470). Also ran into more Z590 Elite users with similar power behaviors, not even CPU Package Power is spared. I will be sending a ticket with this thread to Gigabyte to see what can be done.
 
Back
Top