Question i7-13700k runs at 4.91 GHz in Task Manager, and are my temps alright ?

[ludvigbuskthomasen]

I just Got an i7 13700k and i was wondering why it is showing as running 4.91 ghz in task manager and Corona benchmark when the p cores run at 5.3 to 5.4 and e cores at 4.2
I know they run at these speeds during the test because of hwmonitor, and they arent stuck because they can fluctuate when idle.. Is it because task manager and benchmarks takes e cores in to consideration and that slows the overall speed to 4.91 when using all cores? I don’t understand how it works completely.

Another thing is my temperatures. I use a noctua nh-d15 i Think it is called. At 100% load it goes to a Max of around 85-86 c with highest spinke i have seen at 92 c i Think. I dont Think it is thermal throttling at any point as my p cores stay at 5.3 GHz throughout benchmarks. Normally during the most intensive cpu gaming like cyberpunk 2077 it never goes above 68c and 60% usage at most.

Are these temps normal and is the speed normal and working like i Think it is?

Have a great day!

 

[Lafong]

Without knowing any more about your case, air flow, and case fans, I'd say those temps are fine.

Not sure about your ghz questions.

 

[ludvigbuskthomasen]

Without knowing any more about your case, air flow, and case fans, I'd say those temps are fine.

Not sure about your ghz questions.

I have a corsair 4000d airflow case, 3 case fans (1 exhaust at the back, 2 intake at the front)

 

[uWebb429]

why it is showing as running 4.91 GHz

Try using HWiNFO. Did you increase the turbo power limits in the BIOS? At default settings you might be power limit throttling. HWiNFO can report the Limit Reasons data within the CPU. This might show why your CPU is not reaching full speed.

 

[ludvigbuskthomasen]

Try using HWiNFO. Did you increase the turbo power limits in the BIOS? At default settings you might be power limit throttling. HWiNFO can report the Limit Reasons data within the CPU. This might show why your CPU is not reaching full speed.

Hmm, i installed it and could see the package power under 100% load was 210-220 and IA was the same. Maximum Voltage for the cores was 1.3. I did not touch any power limits in the BIOS. It showed an average clock frequency of 4.750 ghz when stressing it via Corona 10 ,which is mathematically right in the middle between the E cores 4.2ghz and P cores 5.3 Ghz.

 

[uWebb429]

average clock frequency of 4.750

HWiNFO reports the individual frequency of each core. Check to make sure the P and E cores are running at their proper speed. The average speed data is not very useful.

 

[ludvigbuskthomasen]

HWiNFO reports the individual frequency of each core. Check to make sure the P and E cores are running at their proper speed. The average speed data is not very useful.

It always show 5.3ghz on p cores and 4.2ghz on E cores. On both hwinfo and hwmonitor. And i dont Think it is showing wrong info since they Can fluctuate depending on pc power plan and overall cpu usage. The only Strange thing is the effective core speed during bench which sits at around 4.91 to 5.0

Btw when using userbenchmark, it States that my cpu is performance as expected. Dont know How reliable it is though. Are there other tests to compare my cpu under stress to others? Maybe that will tell if everything is normal?

 

[uWebb429]

effective core speed

The HWiNFO Core Effective Clocks will only be equal to the actual Core Clock speed when a CPU core is fully loaded. Use Cinebench R23 to test this. If the Core Effective Clocks do not reach maximum speed during R23 then try running,

ThrottleStop 9.6
https://www.techpowerup.com/download/techpowerup-throttlestop/

Open the Options window and check the Windows Defender Boost box. There used to be a Windows Defender / HWiNFO compatibility issue that could reduce the maximum Effective Clock speed reported by HWiNFO.

 

[ludvigbuskthomasen]

But what is maxi

The HWiNFO Core Effective Clocks will only be equal to the actual Core Clock speed when a CPU core is fully loaded. Use Cinebench R23 to test this. If the Core Effective Clocks do not reach maximum speed during R23 then try running,

ThrottleStop 9.6
https://www.techpowerup.com/download/techpowerup-throttlestop/

Open the Options window and check the Windows Defender Boost box. There used to be a Windows Defender / HWiNFO compatibility issue that could reduce the maximum Effective Clock speed reported by HWiNFO.

mum speed? Since some cores run at 5.3 turbo (P cores) and others 4.2. Turbo (E cores). I dont know what number to look for?

And what am i supposed to do in Windows Defender Boost Box to fix the issue that i dont even know if is an issue?

I Think my techical expertise is limited in this field. I just want to know that everything is working as it should

Thank you for taking the time to help, i appreciate it very much

 

[uWebb429]

what am i supposed to do

Run Cinebench. While Cinebench is running, look at the Effective Clocks and compare them to the Core Clocks. That will show if you have a problem. The Core Clocks for the P cores and the Effective Clocks for the P cores should be equal to each other. The Core Clocks for the E cores and the Effective Clocks for the E cores should also be equal to each other while Cinebench is running.

The Effective Clocks data is only meaningful when all cores of the CPU are fully loaded. Cinebench is a good test for this because while Cinebench is running, it should be fully loading all cores.

 

[ludvigbuskthomasen]

Run Cinebench. While Cinebench is running, look at the Effective Clocks and compare them to the Core Clocks. That will show if you have a problem. The Core Clocks for the P cores and the Effective Clocks for the P cores should be equal to each other. The Core Clocks for the E cores and the Effective Clocks for the E cores should also be equal to each other while Cinebench is running.

The Effective Clocks data is only meaningful when all cores of the CPU are fully loaded. Cinebench is a good test for this because while Cinebench is running, it should be fully loading all cores.

this is what I get. https://imagizer.imageshack.com/v2/614x821q70/922/7C7Sz5.png

While it doesn't keep it at 100% frequency isn't it basically good enough? I don't know why it doesn't though, as it does in your image?.

I still dont get the 4.9 to 5.0 overall effective core speed? is it because it factors in E cores?

 

[mdd1963]

4.9 GHz would seem a reasonable 'all-cores loaded' boost/turbo frequency, with a 5.3 GHz normal single-core turbo. (One might see 5.4 GHz for a second or two..) Run HWMonitor, and then CPU-Z/Bench/'stress'...

 

[uWebb429]

While it doesn't keep it at 100% frequency

Now try running ThrottleStop 9.6.

Check the Windows Defender Boost box in the Options window and see if that makes any difference.

 

[ludvigbuskthomasen]

Now try running ThrottleStop 9.6.

Check the Windows Defender Boost box in the Options window and see if that makes any difference.

Tried a tb bench 120m and it showed a 5.38ghz but keep in mind it diden't 100% load my cpu.

I don't know how to operate the program nor what to look for, since my knowledge is fairly limited.

Edit: i tried running 3dmark cpu profile and it showed my score was the same as other i7 13700k owners. That put me more at ease.

 

[uWebb429]

tb bench

I thought you were using Cinebench before. Use it again when testing. Cinebench is designed to fully load your CPU.

You only need to check the Windows Defender Boost box in ThrottleStop. That is all. After you do this watch the HWINFO Effective Clock data while Cinebench is running. Are the Effective Clocks higher now?

 

[ludvigbuskthomasen]

I thought you were using Cinebench before. Use it again when testing. Cinebench is designed to fully load your CPU.

You only need to check the Windows Defender Boost box in ThrottleStop. That is all. After you do this watch the HWINFO Effective Clock data while Cinebench is running. Are the Effective Clocks higher now?

I checked the box in throttlestop and got these results in hwinfo64, running cinebench 2024.

image 2.png - Click to see more photos

Store and share all of your images @ IMAGESHACK.com

imageshack.com

image.png - Click to see more photos

Store and share all of your images @ IMAGESHACK.com

imageshack.com

Watt stays the same, at around 210-220 which must indicate it isn't power throttling right? This time the all cores hit their frequency with only 1-5 mhz difference. I got a score of 1708 in cinebench, and the average score on their website for my cpu is 1602.

So now I am pretty confident it is all working as intended but I still don't understand the core effective clocks speed of 4.93 and average effective clock speed of 5.078.

 

[uWebb429]

This time the all cores hit their frequency with only 1-5 mhz difference.

The Effective Core clock speeds when a CPU is fully loaded are usually within +/- 2 MHz of where they should be. This is the typical amount of variation (error) that HWiNFO reports.

I cannot see your previous results because Image Shack has already deleted the data. Try using Imgur or some other image hosting site instead.

I believe your P Core effective clocks are now higher compared to when you were not running ThrottleStop. If the maximum effective clocks have increased, your CPU performance will also have increased.

I am guessing you are using Windows Defender like most other people use. Windows Defender is changing the Windows timers that HWiNFO is trying to use to calculate the Effective Clock speed. The result is HWiNFO does not always report Effective Clock data accurately. You can think of this problem as two different people sharing the same stop watch. If both individuals are trying to time two different events at the same time, the elapsed time and MHz results will not be very accurate.

ThrottleStop is able to fix this problem. Users often report a small boost in performance when using ThrottleStop. That might be one reason why your CPU is now outperforming similar CPUs in Cinebench 2024.

Averaging the P and E core Effective Clock Speed data will result in a fairly meaningless number. Only use Effective Clock data to compare P cores to P cores and E cores to E cores. The overall average is not important.

The other thing to keep in mind is that the Effective Clock data is only meaningful when all P and E cores are fully, 100% loaded. Some testing programs like Cinebench will drop the CPU load each time it completes an image. The core clocks should stay the same but the effective clocks will likely drop when this happens. Ignore effective clock data when all cores are not fully loaded.

 

[ludvigbuskthomasen]

The Effective Core clock speeds when a CPU is fully loaded are usually within +/- 2 MHz of where they should be. This is the typical amount of variation (error) that HWiNFO reports.

I cannot see your previous results because Image Shack has already deleted the data. Try using Imgur or some other image hosting site instead.

I believe your P Core effective clocks are now higher compared to when you were not running ThrottleStop. If the maximum effective clocks have increased, your CPU performance will also have increased.

I am guessing you are using Windows Defender like most other people use. Windows Defender is changing the Windows timers that HWiNFO is trying to use to calculate the Effective Clock speed. The result is HWiNFO does not always report Effective Clock data accurately. You can think of this problem as two different people sharing the same stop watch. If both individuals are trying to time two different events at the same time, the elapsed time and MHz results will not be very accurate.

ThrottleStop is able to fix this problem. Users often report a small boost in performance when using ThrottleStop. That might be one reason why your CPU is now outperforming similar CPUs in Cinebench 2024.

Averaging the P and E core Effective Clock Speed data will result in a fairly meaningless number. Only use Effective Clock data to compare P cores to P cores and E cores to E cores. The overall average is not important.

The other thing to keep in mind is that the Effective Clock data is only meaningful when all P and E cores are fully, 100% loaded. Some testing programs like Cinebench will drop the CPU load each time it completes an image. The core clocks should stay the same but the effective clocks will likely drop when this happens. Ignore effective clock data when all cores are not fully loaded.

Okay I think I will mark this thread as solved then. Thank you very much for your time and help
Last Question, Would I theoretically see a slight practical performance gain from running ThrottleStop in the background with windows defender boost checked? or is it only correcting an analytical issue between windows and hwinfo?

 

[uWebb429]

is it only correcting an analytical issue

I think running ThrottleStop in the background is doing a little of both. It fixes the timing issue and it improves performance at the same time. Do some testing to find out if the difference is significant to you or not.

Reboot and do not run ThrottleStop. After Windows settles down, run Cinebench and monitor the effective clocks. When the full load effective clocks are reduced, performance will be reduced. Sometimes things might be fine for the first 5 minutes after booting up so most people never notice this issue. The typical 50 MHz difference under estimates the problem. When Windows Defender starts messing with the internal clocks, performance can be reduced indefinitely after that.

After you get a baseline Cinebench score with the reduced effective clocks, try running ThrottleStop and see if your effective clocks and Cinebench score improves. How big of a difference do you see?

 

[ludvigbuskthomasen]

Okay. thank you guys! I think I understand everything a bit better now.