That is kind of my point. Why buy 1GHz RAM if you can't even realize all of its potential? Good DDR-533 or DDR-667 is all you need IMO, AND you will have better timings. No point in running RAM beyond 1:1, but a high FSB will open up all of the other busses. Raising JUST the RAM Bandwidth is not doing much good IMO (E.G: DDR800 RAM on a "Stock" 200FSB - why would we do that?). Dual-Core Intels will still be using the SAME FSB (one FSB, 2 Cores) to talk to the RAM (and they use the FSB to talk to each other IIRC), so opening the FSB up would make more sense than simply increasing the RAM bandwidth IMO (again, high FSB at 1:1)...
But $300 for 1 Gig? That is a bit above what I would consider spending (I believe 1GB of 1GHz DDR-2 is like $700 IIRC). I could see maybe $400 for 2 Gigs in a high-performance system.
I run a Gig of Corsair XMS 4400C25 (DDR550) in my PC#1 rig, and it does well, but costs ~$200 per gig. I am swapping to a 2 Gig set of OCZ Gold GX DDR-500 that was like $260 or so ($140/GB). Still a bit expensive, but within my budget, and has EXCELLENT performance to boot...
I guess if you are going to spend $1000 on a CPU, you must have money to burn, and all of this is irrelevant
(just get the most expensive RAM you can find) . I, like most mortals, have a budget, and want the best performance possible. Running high RAM Dividers is NOT the way to build a cost-to-performance rig ("Return on Investment"). A cost-no-object rig? Sure. Knock yourself out