Hi,

I just switched my Unstable 9900KF to a stable 9900k

The KF used to run at 4.8 for 1.24 Vcore, the 9900K is running at 5Ghz for 1.25 stable

I don't understand why my firestrike score is so different though,
the same OC has been applied on the gpu
Gsync is off
Vsync is off
power maxed plan on the gpu

With the former KF Cpu at 4.8 : https://www.3dmark.com/fs/22647094
3DMark Score 29420

Graphics Score 39102

Physics Score 25038

Combined Score 11340

Graphics Test 1 193.97 fps

Graphics Test 2 151.32 fps

Physics Test 79.49 fps

Combined Test 52.75 fps
 

with the new K cpu at 5gh : https://www.3dmark.com/fs/22761024
3DMark Score 28308

Graphics Score 36641

Physics Score 24335

Combined Score 11504

Graphics Test 1 181.14 fps

Graphics Test 2 142.18 fps

Physics Test 77.26 fps

Combined Test 53.51 fps

Can someone explain me why this difference?
I've updated the driver and did the last windows update too.

Thanks

I have put a lower vcore that I had before actually and the chip passed cinebench and occt so I don't understand why the 1000K difference

Well a couple things note, is that you are using 2 different drivers to compare. That's is not possible to get a like for like when using different drivers. It also looks like your card in the was boosting higher in the score with the KF CPU. Not only is the listed boost clock higher in the KF score, but the graphics score is also higher by around 3k points. 
 
Now this could be a couple of things, firstly the driver revision as they aren't directly comparable between them. And secondly I suspect your CPU isn't really stable at 5Ghz with only 1.25v. That's an EXTREMELY low amount of voltage for that CPU. I would try again with 1.35v as your baseline and go from there. Running such a low voltage to your CPU may be causing it to actually throttle itself to prevent a crash. 

Asryan
I've ran 2h OCCT with that vcore without issue so i'm pretty sure it's stable actually :)

And yes, drivers are different and i've also updated windows to the last version

here is the last test i've did : https://www.3dmark.com/fs/22761483

You are creating your own problem. You cannot compare scores between driver revisions, OS updates and the like. They aren't directly comparable. 
 
As for your CPU being truly stable at 5Ghz with only 1.25v I have a hard time believing that when Silicon Lottery who tests thousands of them found only 30% to be stable at 5Ghz with 1.3v. And you're claming stability at 5Ghz with 1.25v? Not very likely at all. Try my suggestion just for giggles and up your vcore to 1.35v and test again. You just might surprise yourself. 
https://siliconlottery.com/pages/statistics
 
Am I saying your's isn't possible? No of course not, I'm saying it's extremely unlikely is all. 

Well, 30% is not little chance.
 
I am actually looking at this:
9900KS 5.20GHz (nonAVX) 5.0GHz (AVX) 1.325V Top 4%
... and my 9900KS actually runs at 5.2ghz and passing cinebench (avx) with 1.29v, results doesn't change (~5300) if I am increasing voltage up to 1.35v, so I just left it on 1.33 just to be sure.