HWiNFO® 64 - sensor clocks extremely low/extremely high beyond actual range

[Almighty1]

I noticed that with HWiNFO® 64 recently, I did not check before with older versions that with my Dell XPS 15 9570 notebook on a Intel Core i7-8750H, the minimum and maximum clocks appear out of range as the CPU is 2.2GHz-4.1GHz but when looking at HWiNFO® 64, the following seems out of range:
1. Core Clock(s) appears to have a 0.7MHz minimum and a 10,603.7MHz maximum
2. Bus Clock appears to have a 0.7MHz minimum and a 303.0MHz maximum
3. Ring/LLC Clock appears to have a 10.0MHz minimum and a 9,694.MHz maximum
the above all seem to be significantly lower and significantly higher than the minimum and maximums in the specs.


Is this actually a bug? Thanks!

 

[Martin]

This all is the result of wrong bus clock measured.
The problem is that older CPUs didn't provide reliable methods to measure bus clock and HVCI blocked access to certain interfaces. Disabling HVCI might help to mitigate this problem.

 

[Almighty1]

Thanks for the answer. How recent does the CPU has to be atleast for the bus clock to be reliably measured as I was just reading this thread and 13th generation Intel is recent and still has the issue? Wouldn't disabling HVCI be bad as having less security though? Is it actually HVCI or is it VBS as mentioned at https://www.hwinfo.com/forum/threads/bus-clock-incorrect.6552/post-26864 because if it is related to MSR, then it's VBS since that's the same thing that prevents ThrottleStop and Intel XTU from working as VBS is what blocks access to the MSR.

 

[Martin]

HVCI and VBS are the same thing - "Core isolation/Memory integrity" in Windows settings.
Precise BCLK works since Intel 6th generation (Skylake) but due to HVCI/VBS it's not accessible. On 10th generation (Ice Lake) there's also an alternate method available which is not blocked by HVCI/VBS.

 

[Almighty1]

Thanks, but referencing the thread at https://www.hwinfo.com/forum/thread...hreads/core-clocks-accuracy-reliability.8554/ - wouldn't 13th generation CPUs not have the problem since it's newer than 10th generation or is the alternate method nt available?

 

[Martin]

That thread discusses a different issue - slight variation in BCLK measurement which is in fact correct due to SSC.
13th Gen has the alternate method.