that's a tough question, since a lot of this is subjective. A lot also depends on what you'll do with it. Some programs like memory bandwidth more than others. But I think most everyone (no matter how big of a power user) would agree that it won't "run like a dog". In most cases (if I remember correctly) the biggest difference you'll see is about 5%. In games, the effect is negligable. I believe encoding programs (mp3 rippers, dvd rippers, trans-coding software, etc.) are where you would see the biggest benefiet.
Something you might think about if you really want to run dual channel, is getting the same amount, just in two sticks instead of one. For example: the cheapest 512 stick of pc3200 I found on newegg was around $60, where as the cheapest set of 2x256 pc3200 sticks were around $75. So you only have to pay $15 extra bucks for dual channel. Just have to ask is %5 worth the $15 bucks for you. Maybe it is, maybe it isn't.
I'm the kind of person who would pay the extra $15 just to make sure I was going to be happy with what I got, instead of always wondering if I would have liked dual better than single channel. But it's equaly reasonable to think that you could save the $15 and run at a speed so close to dual, that only a benchmark can tell the difference.
n e way, that's enough babbling from me. I have been looking around trying to find benchmarks that support the "5%" thing, but for some dumb reason can't find any. I'll keep looking and if I find some, I'll post them. HOWEVER, it has been shown that senthetic benchmarks can sometimes exagerate the real world effects of better (ie. more expensive) ram, or any other component for that matter! To see a good example of this, look at Zebo's sticky in CPU/OC, here:
http://forums.anandtech.com/me...5190&enterthread=y