Quote Originally Posted by Jacky View Post
Can anyone elaborate?

I see it is 32 bit, so now that macro-ops fusion is enable in 64bit this should give a boost of a couple percent to nehalem in contrast to C2D in realworld 64bit code. Thus they'd be at least eqal in FPU/ALU tests I guess.
Why is nehalem faster in the GPU tests, does anyone know? (it seems they also make us of the CPU?)

Are there any 3dmark experts here?
Could someone compare nehalem with penryn: http://ocxtreme.org/forumenus/showth...?t=3978&page=7 (link used to work)
Nehalem 3,2ghz, 9800 GTX+ default
memory IIRC: tripple channel 800mhz (7-7-7-21)
scores 62781 in 3dmark01.
I can venture a guess as to 3D performance. Latency and bandwidth from CPU to PCIe are both greatly improved through using a direct QPI link to the PCI controller on the X58 chipset. There might also be several improvements on the controller to make use of these new found boosts.