Yeah, its pretty easy to reproduce set the following in the BIOS to replicate what I have... 
 
All Core Multiplier - x44
AVX2 Offset - 0
AVX3 Offset - 1
CPU VIN - 1.800v
VCORE - 1.200v
 
Than I am running OCCT v7.1.0 selecting the Large Dataset with AVX512 only. Which should result in a core clock of 4.3GHz with the above offset however it will drop to 4.2GHz. If you dont want to use OCCT than you can use Prime95, just use current version and select large dataset. 

jasoncodispoti
Yeah, its pretty easy to reproduce set the following in the BIOS to replicate what I have... 
 
All Core Multiplier - x44
AVX2 Offset - 0
AVX3 Offset - 1
CPU VIN - 1.800v
VCORE - 1.200v
 
Than I am running OCCT v7.1.0 selecting the Large Dataset with AVX512 only. Which should result in a core clock of 4.3GHz with the above offset however it will drop to 4.2GHz. If you dont want to use OCCT than you can use Prime95, just use current version and select large dataset. 

OK, I will let you know what I found once I get to point of being able to test that (I'm at least few weeks away from it). In the meantime which build of Windows you are on?

Edition Windows 10 Home
Version 20H2
Installed on ‎7/‎9/‎2020
OS build 19042.685
Experience Windows Feature Experience Pack 120.2212.551.0 (whatever the hell this means LMAO) 

jasoncodispoti
Edition Windows 10 Home
Version 20H2
Installed on ‎7/‎9/‎2020
OS build 19042.685
Experience Windows Feature Experience Pack 120.2212.551.0 (whatever the hell this means LMAO) 

BTW, do you happen to know what are Intel's official AVX offsets for 10900X/10920X when running at official all-core turbo multiplier (x43) or where one could find that?
 
P.S. Windows Feature Experience Pack is IMO Microsoft's way of separating updates for features that are not an actual OS but an app (like screen snipping app ... err ... "feature").

i got the same problem. avx both at 0,  No thermal throttling or power, But the Cpu still throttle some how.   im on water cooling with decent rads++

Exnetic
i got the same problem. avx both at 0,  No thermal throttling or power, But the Cpu still throttle some how.   im on water cooling with decent rads++

I don't want to speak prematurely, before researching more and testing, but I was always under impression that AVX workloads (especially AVX-512) need an offset, that 0 offset at high frequencies will be pushing CPU too hard resulting in some kind of internal throttling. I know that when I was testing my i9-10900X zero offset for AVX3 was pushing temperatures practically immediately through the roof at 43x CPU multiplier, and even 3 was seeing them climb steadily.

ZoranC

Exnetic
i got the same problem. avx both at 0,  No thermal throttling or power, But the Cpu still throttle some how.   im on water cooling with decent rads++

I don't want to speak prematurely, before researching more and testing, but I was always under impression that AVX workloads (especially AVX-512) need an offset, that 0 offset at high frequencies will be pushing CPU too hard resulting in some kind of internal throttling. I know that when I was testing my i9-10900X zero offset for AVX3 was pushing temperatures practically immediately through the roof at 43x CPU multiplier, and even 3 was seeing them climb steadily.

Does not always need an offset... just depends if AVX has been implemented by the software correctly. Using something like handbrake for example that worked with Intel on the AVX-512 implementation for X264 you dont really need an offset if you have a proper cooling setup.

And this should now be fixed with this version of the BIOS:
https://forums.evga.com/BIOS-Updates-for-X299-Dark-125-FTW-K-125-Micro-ATX-2-118-Micro-ATX-126-m3196486.aspx

I have confirmed this morning that BIOS update v1.25 corrected the issue mentioned in this thread. 
 
Note: 
- If you are using AVX-512 than I would highly recommend updating to this BIOS version. 
- Depending on your OC and due to the Microcode issue in previous versions you may need to re-do your OC in regards to AVX-512 datasets. 
 
Thanks EVGA for getting this update out! 

jasoncodispoti
I have confirmed this morning that BIOS update v1.25 corrected the issue mentioned in this thread. 
 
Note: 
- If you are using AVX-512 than I would highly recommend updating to this BIOS version. 
- Depending on your OC and due to the Microcode issue in previous versions you may need to re-do your OC in regards to AVX-512 datasets. 
 
Thanks EVGA for getting this update out! 

That is good news!