CPU Utilization seems to be incorrect.

[ValhallasAshes]

CPU:  AMD Ryzen 7 9800X3D

OS:  Windows 11 23H2

AIDA64 version: 7.50.7236 Beta (needed this version as current mainline version doesn't support RTX 5080)

I was configuring some fan curves today (recently did a wipe and was getting all my settings dialed in from fan curves to other things) and while testing my temps in a game loop in X4 Foundations I know can really drive CPU temps up, I noticed my CPU cooler was saying my CPU usage was anywhere from 30%-40% utilization.  My CPU cooler is a Deepcool Assassin IV VC Vision, which has a screen with some various stats on it for the CPU (Temp, Utilization, Watts and Frequency). 

I didn't think anything of it, because 30%-40% sounded really normal under that kind of game load.  But then I looked at my WigiDash display, which I have an AIDA64 sensor panel integrated into and in the CPU section, the load reading was only bouncing around 1%-3%.  Sometimes a jump of 11%.  I know what you're thinking, because it was the first thing I thought.  I must have the incorrect setting in the sensor panel preferences.  So that's the first thing I checked.  I definitely have the correct one selected. 

There are 17 possible entries for my CPU in the list.  The first one being "CPU Utilization", followed by listings for "CPU1 Utilization" - "CPU16 Utilization".  Now I know it's not any of the 1-16 options because I actually have individual bars for each one of my cores/threads on my sensor panel.  So not only can I see them, I can see each one of their behavior to tell they are not behaving abnormally (ie giving a reading more representative of overall utilization vs just one thread).

So now I know I not only have the correct entry selected, but I have two different pieces of hardware reporting contradicting values for the same parameter.  So I needed to figure out which one was wrong.  I've now checked and compared the CPU utilization values in the Deepcool software for the cooler, my Lian Li L-Connect 3 software, CPUID HWMonitor, and Windows Task Manager Performance tab.  Even the WigiDash's specific software is reading the correct value.  All of them are showing the same values.  The only software not showing the same value is AIDA64's sensor panel.

I don't know if it's a known bug, or if it's an issue specific to the beta branch, but I thought I should report it.  Not sure if this is the correct place to do it nor what format you prefer for bug reports.  Hopefully I got it close enough.

[Fiery]

7 hours ago, ValhallasAshes said:

CPU:  AMD Ryzen 7 9800X3D

OS:  Windows 11 23H2

AIDA64 version: 7.50.7236 Beta (needed this version as current mainline version doesn't support RTX 5080)

I was configuring some fan curves today (recently did a wipe and was getting all my settings dialed in from fan curves to other things) and while testing my temps in a game loop in X4 Foundations I know can really drive CPU temps up, I noticed my CPU cooler was saying my CPU usage was anywhere from 30%-40% utilization.  My CPU cooler is a Deepcool Assassin IV VC Vision, which has a screen with some various stats on it for the CPU (Temp, Utilization, Watts and Frequency). 

I didn't think anything of it, because 30%-40% sounded really normal under that kind of game load.  But then I looked at my WigiDash display, which I have an AIDA64 sensor panel integrated into and in the CPU section, the load reading was only bouncing around 1%-3%.  Sometimes a jump of 11%.  I know what you're thinking, because it was the first thing I thought.  I must have the incorrect setting in the sensor panel preferences.  So that's the first thing I checked.  I definitely have the correct one selected. 

There are 17 possible entries for my CPU in the list.  The first one being "CPU Utilization", followed by listings for "CPU1 Utilization" - "CPU16 Utilization".  Now I know it's not any of the 1-16 options because I actually have individual bars for each one of my cores/threads on my sensor panel.  So not only can I see them, I can see each one of their behavior to tell they are not behaving abnormally (ie giving a reading more representative of overall utilization vs just one thread).

So now I know I not only have the correct entry selected, but I have two different pieces of hardware reporting contradicting values for the same parameter.  So I needed to figure out which one was wrong.  I've now checked and compared the CPU utilization values in the Deepcool software for the cooler, my Lian Li L-Connect 3 software, CPUID HWMonitor, and Windows Task Manager Performance tab.  Even the WigiDash's specific software is reading the correct value.  All of them are showing the same values.  The only software not showing the same value is AIDA64's sensor panel.

I don't know if it's a known bug, or if it's an issue specific to the beta branch, but I thought I should report it.  Not sure if this is the correct place to do it nor what format you prefer for bug reports.  Hopefully I got it close enough.

Task Manager and HWMonitor measure CPU utilization using a different formula than the classic formula (that is used by AIDA64).  Task Manager and HWMonitor take Turbo Boost clock levels into account as well, and they can measure CPU utilization over 100% too -- which is quite "interesting" to say the least.

Try to disable Turbo and check if the measured CPU utilization values match that way.

[ValhallasAshes]

No that's ok.  They all hit 100% when all threads are loaded under Cinebench.  It's just the utilization numbers in between that seemed wrong.  Aida64 was the only app giving contradicting measurements from all the other programs I tested with and thought it was a bug.  If you say it's because of differences in the formula used to measure utilization in Aida64, but that measurement is accurate, that's good enough for me. 

Although there is something to be said for being the only person using Fahrenheit in a room full of Celsius users.

[Fiery]

53 minutes ago, ValhallasAshes said:

No that's ok.  They all hit 100% when all threads are loaded under Cinebench.  It's just the utilization numbers in between that seemed wrong.  Aida64 was the only app giving contradicting measurements from all the other programs I tested with and thought it was a bug.  If you say it's because of differences in the formula used to measure utilization in Aida64, but that measurement is accurate, that's good enough for me. 

Although there is something to be said for being the only person using Fahrenheit in a room full of Celsius users.

Our stance is that the other formula is counter-factual, especially when it yields to values crossing over 100%.  However, in case we receive a whole bunch of requests to apply the other formula, we may implement it as an optional method.

[ValhallasAshes]

I understand.  I wasn't saying you were wrong.  In fact, I would probably agree if wasn't for the fact that I don't know the formula for how you measure activity nor how that compares to the method everyone else uses.  Not that I would technically understand the formula even if I saw it.  But it also doesn't help that I can't corroborate your numbers because I don't know of a single other application that uses the same measurement method. 

So I can't understand because I don't know and I can't corroborate because nothing is shared/replicated between any third parties.  Again, you're probably right and your formula is probably better.  I have full faith and trust in that.  I simply have no way to verify that internally nor externally. 

In the end though, it's not that big of a deal for me.  Between my CPU cooler and the Aida64 sensor I can see both measurement methods anyway, but even so, I use it more for temps, RAM/VRAM usages, wattage monitoring and gaming frame rates.  The only thing I use CPU utilization for is to see if and when an application hangs on black screen, failing to start or progress (used to happen a lot, not so much these days). 

Thanks again for the info and especially for the quick responses.  It's really appreciated.