at this point, when speed and price are a very good match and things seem to look right, i still look at the motherboard.
2 generations behind now in CPU architecture.
4th gen i5 and new mobo (bundle), can be had for a few dollars difference compared to older stuff people are getting rid of. suddenly, we're getting a new cpu, a new mobo and updated technology, regardless of speed.
Well, you could take an i5-2500K and drop it into a Z77 motherboard, so . . . one generation? Those and the skt-2011 boards are still being sold -- the whole spectrum of prices and features.
How many new instructions have been added to the IB and then Haswell instruction sets? How much computing power are you going to use at any given time? The trend is to reduce the lithography in nanometers, pack more transistors into a smaller package, reduce power consumption requirements. But the reduction in power requirements isn't compensating enough for cooling. They've pretty much edged up speed by increments -- the top-end "X" chip of the IB-E has a spec turbo speed of 4.0 Ghz.
The odd thing about building your own machines is the attachment. IF you sell a machine, your original outlay less what a buyer paid for it secondhand would tell you your annual computing cost. Or you could dismantle a perfectly good system you use daily for various mainstream or oddball purposes, sell certain parts and re-use the others.
Looking back, it seemed as though it took a lot longer to move from Pentium II to IV than it has going from SB to Haswell. I can't even keep up with the number of different chips they produce for any generation, when you throw in the mobile processors for laptops and . . . other . . . stuff . .
The technology has outstripped the needs of many mainstream uses, other than gaming. We should be getting rid of the C2Ds we have here, because they're all past 5 years old. But with SSDs and other things, nobody is unhappy yet.
I will build one machine this year. But I'll also have to retire one, or I won't be able to find a place to sit down . . .