Need versus want.
I built my P4 system starting in July '04, and went through about three processors in two years -- didn't burn-'em out, but just felt "the need for speed" which was really a "want."
Held off through Smithfield and Presler. Waited a full six or seven months after the release of C2D during summer '06.
One little glitch that a "clear CMOS" didn't fix added to the euphoria over the July 22 price-cuts, and I jumped on the Q6600 -- only to find that it was the B3 stepping. E6600 goes for mom's Xmas present. Apparently, removing it from the mobo and re-installing it . . . . "did something."
I'm looking at the trail of receipts on this "project." I do not want to discuss further just how much over budget this has taken me.
You have to balance certain reasonable restraints against your search for the "fastest, bestest most wonderful-est computer in all Geek-dom." Because -- once you find that "holy grail," a few months later, it won't be the fastest, bestest most wonderful-est.
Eventually, there will be eight-core processors, and I thought I read somewhere of a multi-core processor in the planning stages that makes eight-cores look primitive.
Think of it like surfing. You have all these factors to consider: wind, undertow, the occasional big wave. Similarly -- prices, technical improvements, budget, bragging rights, and various "personal satisfactions."
Try and catch the right wave, and then ride that sucker for a while . . . .