Quote Originally Posted by Bojamijams View Post
Can anyone test something that I've witnessed with LinX thats quite interesting.

I'm stable at +100mv VTT in LinX. the GFLOPS is 45.

If I go to +125mv VTT, my GFLOPS increases to 49.

If I goto +150mv VTT, gflops goes to 52.

So I'm getting better floating point calculation by giving the VTT more voltage, even though for stability, I don't need it. It looks like it scales its performance based on VTT voltage.

I'm curious if anyone else is seeing the same things and if they see the performance transfer anywhere else?
Interesting finds. If VVT is affecting the performance in the way you are discovering, then it is really bad news for my W3520. It runs stable with VVT=1.21/1.26v(BIOS/Load) at almost any x8 RAM speed and x18 Uncore.

I couldn't track the effect of VVT on CineBench 10 score yet, do you get different results on CineBench 10 (or other CPU/system-performance monitoring SW)?