Very comprehensive and useful test. Congrats for your hard work!
It would be great if possible to test some more with the CPU's at the same clock...
The reasonable difference in clock speeds surely affects the framerates.
Yeah. It would be nice if Xbits could do test with i5 overclocked to i7's speed so we can see the benefits of LGA 1366, extra memory bandwith and Hyper-Threading. If I'm not mistaken, Hyper-Threading slows some games down actually consumes more power when enabled.
Not necessarily overclocking. The EE could be underclocked just by multiplier to the frequency of i5 750 I think. It's common sense that a higher clocked CPU will perform better than a lower clocked one, even on the same architecture, more so if the architecture is considered superior.
Your Quake 1 glide screenshots are taken at much higher resolution then the software screenshots, so it's not a fair comparison. I think it's 640x480 vs 320x200, or something like that. Pentium pro 200's were available before Quake 1 was released and they could run Quake 1 in software at 640x480, so there's no reason not to compare software vs glide at the same resolution.
I have to disagree with the conclusions. the tests showed a difference in only a few games, but:
1. the CPU tested were both very high end, most would attempt with lower end CPUs. Where CPU bottlenecking is more pronounced.
2. even though games where the CPU is a problem are rare, they do exist. And lowering your graphic settings has significant impact on the GPU utilization, but very little on the CPU.
Your message is dont buy overprice gulftown (1000 $)no need for game,other dilemma is which GPU for gaming ? grill named Fermi or Ati HD5xxx series, in fact that is not dilemma ,because Nvidia have problem with Windows 7 driver.