Fully agree with OP.
My own results absolutely murder even Skaterbenchers OC guide results using my daily config.
All the #1 spots on OCCT benchmark for 9950x are mine.
Look at 2 channel DDR5 on OCCT, you’ll find the same user very high up.
9950X has the best PBO overclocking user experience ever. PBO behaves intuitively with no “technically functional but practically useless” items e.g. scalar on 3000 series. Yes it had an effect but it was not useful. 9950x responds well to everything.
FMAX +200…yep, effective clocks now peg 5921mhz.
Scalar at 10…yep, VID now + 40(ish) mv.
PPT/TDC/EDC …yep, all work to contain various loads.
CO…yep, massively boosts single and multithreaded loads.
Shaper…yep, allows for big negative CO for all core loads and boosts voltage for single dual threaded work to stabilize.
Asus added per core limiting to the list of knobs to tweak, giving use to async clock for those who know how to use it to gain even more multithreaded speed without losing single/dual threaded stability.
I don’t use static OC aka precise and direct. All scores are PBO tuned. No Dynamic OC switcher, no Core Flex. They’re gimmicks, nothing more.
My daily 310 PPT/225 TDC/225EDC config results in win 11 24H2 with core isolation OFF:
CPUZ:
941 st
19200 mt
R23:
2370 st
48600 mt
R24:
148 st
2740 mt
OCCT
118.x
19xx
244.x
38xx
At 200 PPT it still puts out ~46k in R23.
In gaming, once again, complete non issue. CPU jumps to 5850+ on CCD0 and 5650 on CCD1 and stays there. Frame rates are essentially identical to my delidded and OCd 7800X3D. Why? Well, simple, I don’t play at potato settings on a 2007 monitor. I didn’t hesitate for a second replacing the 7800 with the 9950x.
9950x is snappier in windows, crushes all prior Zen in actual programming workloads like compiling big projects, etc. Windows can finally keep up with my input speed, an absolute joy for a programmer working quickly jumping across many tasks at once.
When recording BM:Wukong using OBS and CPU encoded h264 with some very beefy settings, I pay a whopping 1 FPS. ONE. A Single frame is lost to encoding at near lossless quality.
There’s no config that allowed any 8 core X3D to do the same using an AMD GPU. You absolutely cannot do it, I know, I own them all 5700, 5800, and 7800 X3Ds. I hold or held the #1 spots on basically every benchmark on hwbot for the 5800 and 7800 at one point, some were superceded by LN2 testing etc.
The tech media completely dropped the ball on the 9950x. They’re lazy, ignorant, and have always failed to extract anywhere near the limits of these CPUs (exception being GN when they play with LN2 while guest OC gurus are present).
