What is the real core frequency in HWiNFO64 -- Core Clock or Core Effevtive Clock?

[kingsong]

I'm glad to share with you what I found by chance. I bought a 5950x and just used it for two days. Suddenly, I found that the CPU speed of the task manager has changed to 6.2GHz, but I set the full core lock of 4.6ghz and the core voltage of 1.2625v in the BIOS. Of course, I thought this was a system error, so I tested the cpuz and the displayed core frequency is still 4.6ghz, When I was ready to ignore this error, something surprised me happened. Except that the cpuz running points had not changed, including cinebench R15, R20, R23, chess test and super PI test, the performance was improved by 20% - 40% compared with before. Of course, the power consumption was also increased to around 330W, which was incredible, Because I didn't believe that such a low voltage could stably operate at such a high frequency, I conducted repeated tests. I was almost sure that the performance was improved a lot when no detection software could normally display the frequency of 6.2GHz. I told several leaders in the science and technology circle about this, They also said that it is almost impossible to increase the core frequency to 6GHz in the case of air cooling. The only thing I found to prove is that under the sensor stacus of hwinfo64 v7.11-4545, the Core Effective clock can accelerate from the previous 4.6Ghz to 6.2GHz, and all 32 lines. Usually, HWiNFO64 monitors the core real-time frequency according to the core clock, However, on my computer, the Core Clock is invariable and stable at 4.6ghz. I can't see any frequency fluctuation, but I can observe the dynamic change of the Core Effective Clock accelerating from 10MHz to 6000Mhz, which is consistent with the CPU load. This is what I have found so far.

 

[Martin]

FEATURE - Effective clock vs instant (discrete) clock

It has become a common practice for several years to report instant (discrete) clock values for CPUs. This method is based on knowledge of the actual bus clock (BCLK) and sampling of core ratios at specific time points. The resulting clock is then a simple result of ratio * BCLK. Such approach...

www.hwinfo.com

The Effective clock in this case better reflects the quick maximum turbo fluctuations. You might also check in the main HWiNFO window what your turbo limits are set to.

 

[kingsong]

FEATURE - Effective clock vs instant (discrete) clock

It has become a common practice for several years to report instant (discrete) clock values for CPUs. This method is based on knowledge of the actual bus clock (BCLK) and sampling of core ratios at specific time points. The resulting clock is then a simple result of ratio * BCLK. Such approach...

www.hwinfo.com

The Effective clock in this case better reflects the quick maximum turbo fluctuations. You might also check in the main HWiNFO window what your turbo limits are set to.

But I still don’t understand why the extreme value of the effective frequency exceeds the core frequency. If it is a phase sampling algorithm, it can only be lower than the maximum value of the core frequency, but the result I see is much higher than the maximum value of the core frequency, is it caused by the different signal sources detected? What's interesting is that in addition to the effecvtive clock of HWiNFO, which can reflect the true frequency I think, Windows Task Manager and Geekbench 5 can also display the maximum acceleration frequency of 6G. Is there any connection between the three? Can I check the real-time frequency through CMD?

 

[Martin]

The classic clock sampling method might not catch the highest single-core turbo values due to the observer effect - it might trigger reduction of max clock due to utilization of multiple cores.

 

[kingsong]

The classic clock sampling method might not catch the highest single-core turbo values due to the observer effect - it might trigger reduction of max clock due to utilization of multiple cores.

Your professional reply has benefited me a lot, and I also have new understanding and inspiration. I am very grateful

 

[Martin]

You're welcome