pulling the multiplier above 10x for 2 cores of the Kentsfield, I did the same test on the D975XBX2 Bad Axe 2.
That’s great, I found this problem to be non-existant on the Bad Axe 2, all
4 cores are in sync with their multipliers even above 10x!
Now, this is what I found. CPUZ on the ASUS P5W DH Deluxe detects
the 4 cores in this sequence as you scroll through: Core 1, Core 2, Core 3,
Core 4. However, the Bad Axe 2 detects in this sequence: Core 1, Core 3, Core
2, Core 4. Thiis is probably due to the layer sequence differential of the 4
data groups (probably 1 group per core) as stated on the first page of this article.
Using just 1.55v vcore, I ran the 4 cores over 3D Mark 06 CPU
Test and got a great score of 6,300+!
Just for comparision, a Core 2 Duo @ 4.6GHz gets 4,000 points,
so that’s close to a 60% boost with 100+MHz less on the Kentsfield.