Full CPU power consumption

[Roman]

Tell me how the total CPU power consumption is calculated

 

[Zach]

If you mean the entire CPU package then all CPUs have multiple rails. 1 for cores, 1 for SoC and probably 1-2 more for other stuff.
In order to calculate power (Watt) you need voltage (V) times (x) current (A).
So you need to know each rail's voltage and current at any given moment and then you add all rails (VxA=W) together to find the total CPU package power.

For example:
If a specific rail at some point has 1.1V and 100A, then power at that moment on that rail is 110W

Volts and Ampere on a CPU are highly dynamic depending on load.

 

[Roman]

If you mean the entire CPU package then all CPUs have multiple rails. 1 for cores, 1 for SoC and probably 1-2 more for other stuff.
In order to calculate power (Watt) you need voltage (V) times (x) current (A).
So you need to know each rail's voltage and current at any given moment and then you add all rails (VxA=W) together to find the total CPU package power.

For example:
If a specific rail at some point has 1.1V and 100A, then power at that moment on that rail is 110W

Volts and Ampere on a CPU are highly dynamic depending on load.

Hi!
I am interested in how HWiNFO64 determines the total CPU power consumption.
The user's initial data
13900k
cooler Ultra 120EX REV.4 WHITE
The screenshot shows the results
Total power consumption 371W
I can't believe that such a cooler can take away so much heat. Either this is possible, or the program incorrectly outputs the result.

 

[LeDoyen]

Whether the value is correct or not, you shouldn't leave the power limits unlocked with an aircooler. Here PL1 and PL2 are both set to 4096W instead of 125 and 253W respectively.

 

[Roman]

Whether the value is correct or not, you shouldn't leave the power limits unlocked with an aircooler. Here PL1 and PL2 are both set to 4096W instead of 125 and 253W respectively.

Hi!
I asked a specific question earlier. I want to get a specific answer. There is nothing from you but unnecessary advice.
If the cooling system allows you to keep the temperature within the range of +100, then it is quite possible to increase PL1, PL2. Can you tell me about what temperature Intel gives a limit of 252W?

 

[LeDoyen]

Oh i didn't mean to be rude
They don't give temperature figures, it's just the factory power specification, that's why i was pointing at them. The power limit you have is strange because to get that much power applied through you usually need a very good AIO or even custom loop cooling. From the power limits, it seems you haven't touched the bios settings and it runs as it came out of the box, which would be unlocked power limits and maybe slightly high vcore values. But if it works and you're comfortable bumping into temperature limits, yes by all means keep it like that, the CPU is protecting itself as it should.
Now if the cooler noise bothers you, you'll know you can lower the power limits and get it to spin slower of course. Running benchmarks with power limits disabled and enabled may give food for thought as the difference isn't proportional to the power increase.

 

[Roman]

Oh i didn't mean to be rude
They don't give temperature figures, it's just the factory power specification, that's why i was pointing at them. The power limit you have is strange because to get that much power applied through you usually need a very good AIO or even custom loop cooling. From the power limits, it seems you haven't touched the bios settings and it runs as it came out of the box, which would be unlocked power limits and maybe slightly high vcore values. But if it works and you're comfortable bumping into temperature limits, yes by all means keep it like that, the CPU is protecting itself as it should.
Now if the cooler noise bothers you, you'll know you can lower the power limits and get it to spin slower of course. Running benchmarks with power limits disabled and enabled may give food for thought as the difference isn't proportional to the power increase.

I will clarify the situation. The issues of cooling with coolers in the air were discussed at the forum. The participant posted a screenshot with the results in HWiNFO64 when using the Ultra 120EX RAV.4 WHITE cooler for cooling 13900k. I had doubts about the result. So I wanted to find the truth. Where the error comes from. The user has configured the BIOS and the sensors give out false information or HWiNFO64 incorrectly processes the information received from the sensors.

 

[LeDoyen]

Well one could change the value of the sensor. when you right click it you can add a fixed value to the reading or set a correction factor. But that would be weird to do.
HWinfo polls the value straight from the CPU so i don't think it can read wrong.

 

[Roman]

Well one could change the value of the sensor. when you right click it you can add a fixed value to the reading or set a correction factor. But that would be weird to do.
HWinfo polls the value straight from the CPU so i don't think it can read wrong.

It's not entirely clear. Does the processor itself calculate the total power consumption and then output the received data to Hwinfo64 or does the program calculate based on the data received from the processor sensors?

 

[Martin]

The CPU calculates energy consumed and reports this. HWiNFO takes calculates the difference between 2 samples - (delta)Energy / (delta)Time.

 

[Roman]

The CPU calculates energy consumed and reports this. HWiNFO takes calculates the difference between 2 samples - (delta)Energy / (delta)Time.

Hi!
Is such a cooler Ultra 120EX REV.4 WHITE capable of removing 371 W of heat? Do I understand correctly that if the user has made changes to the BIOS settings, then the algorithm for calculating the total CPU power consumption may give an incorrect result?

 

[Martin]

Some overclocking settings or SVID disabling might cause this.

 

[Roman]

Some overclocking settings or SVID disabling might cause this.

Do you confirm that the incorrect indication of the full CPU power consumption is caused by incorrect BIOS settings by the user?Are there any other reasons that may affect the unreliable data on total power consumption that HWiNFO64 reflects?

 

[Zach]

Hi!
Is such a cooler Ultra 120EX REV.4 WHITE capable of removing 371 W of heat? Do I understand correctly that if the user has made changes to the BIOS settings, then the algorithm for calculating the total CPU power consumption may give an incorrect result?

Do you confirm that the incorrect indication of the full CPU power consumption is caused by incorrect BIOS settings by the user?Are there any other reasons that may affect the unreliable data on total power consumption that HWiNFO64 reflects?

No one really can confirm anything but making assumptions unless witnessing the actual BIOS/UEFI settings.

Personally I too find it odd that such a relatively small cooler can remove 370W of power/heat and CPU stays under 100C.
Unless this system is placed in a very very low ambient temp environment I can't see it happening, but I can only assume as I said based on my personal experience of PC systems.

 

[Roman]

No one really can confirm anything but making assumptions unless witnessing the actual BIOS/UEFI settings.

Personally I too find it odd that such a relatively small cooler can remove 370W of power/heat and CPU stays under 100C.
Unless this system is placed in a very very low ambient temp environment I can't see it happening, but I can only assume as I said based on my personal experience of PC systems.

I have one version. The user changed the BIOS settings, which affected the readings of the total power consumption. What is the default frequency of 13900k in multithreaded mode? What versions do you have?

 

[Zach]

I have one version. The user changed the BIOS settings, which affected the readings of the total power consumption. What is the default frequency of 13900k in multithreaded mode? What versions do you have?

First of all I do not own an Intel system.
To you question about default freq of 13900K in multi... at auto boost P-cores peak at 5.5GHz and E-cores at 4.3GHz. But that does not mean that these clocks can be sustained over time under continued long boost, especially with air coolers.

Below there is a test with 14 air coolers of different cooling capacity (from 19 to 125$)
On the best coolers the avg P-core freq was around 5.3GHz and E-cores around 4.2GHz with avg power of 270+W
On the worst one the avg P-core freq was around 4.9GHz and E-cores around 3.9GHz with avg power of 220-230W

On all coolers thermal throttling (100C) did happened either instantly or after a few sec and none made it possible to even reach the 295W limit that was set on the board, let alone sustain it.

 

[LeDoyen]

i ran a CB R23 test, my CPU is locked at 253W PL1 and PL2 and the P-cores cap at 5.3ghz, with Ecores at 4.1. No tweaking at all besides setting PL1 to 253 instead of 125W. That already gives me almost 80° on water.
R23 finishes before the boost expires so a factory set 13900K would most likely have those kinds of clocks by default.

Single thread it peaks at 5.8, basically following the spec https://ark.intel.com/content/www/u...3900k-processor-36m-cache-up-to-5-80-ghz.html

if memory serves, before i restored the power limits, on out of the box motherboard OC, the CPU was throttling slightly beyond 360 - 370W already (the rig is on custom loop, not AIO..)

 

[Roman]

First of all I do not own an Intel system.
To you question about default freq of 13900K in multi... at auto boost P-cores peak at 5.5GHz and E-cores at 4.3GHz. But that does not mean that these clocks can be sustained over time under continued long boost, especially with air coolers.

Below there is a test with 14 air coolers of different cooling capacity (from 19 to 125$)
On the best coolers the avg P-core freq was around 5.3GHz and E-cores around 4.2GHz with avg power of 270+W
On the worst one the avg P-core freq was around 4.9GHz and E-cores around 3.9GHz with avg power of 220-230W

On all coolers thermal throttling (100C) did happened either instantly or after a few sec and none made it possible to even reach the 295W limit that was set on the board, let alone sustain it.

Thanks for the information.

 

[Roman]

i ran a CB R23 test, my CPU is locked at 253W PL1 and PL2 and the P-cores cap at 5.3ghz, with Ecores at 4.1. No tweaking at all besides setting PL1 to 253 instead of 125W. That already gives me almost 80° on water.
R23 finishes before the boost expires so a factory set 13900K would most likely have those kinds of clocks by default.

Single thread it peaks at 5.8, basically following the spec https://ark.intel.com/content/www/u...3900k-processor-36m-cache-up-to-5-80-ghz.html

if memory serves, before i restored the power limits, on out of the box motherboard OC, the CPU was throttling slightly beyond 360 - 370W already (the rig is on custom loop, not AIO..)

Your observations coincide with the opinions of other users with whom I have communicated.
Thank you for the information provided.

 

[Zach]

if memory serves, before i restored the power limits, on out of the box motherboard OC, the CPU was throttling slightly beyond 360 - 370W already (the rig is on custom loop, not AIO..)

Yeah that is the only way I see it for the 13900k to hit 370W... with a custom loop and not some half tower simple air cooler.