4770?
My bad, 4750
I keep thinking it was 70/50, when it ended up being 50/30 in the end. Which brings me to another topic of discussion:
What's up with the decided 650/3200 clock btw? I thought they'd settled on 700mhz/3600. Judging by the TDP, I really don't understand it. If you're over 75W and need a 6-pin plug, why not go for broke in the desktop arena? 78W, amongst everything else (4ghz gddr5 @ 3200mhz), makes it appear more than a little held back...
Why not just release a 800/3900-4000ish part and be done with it? Aw, yes...that pesky 4850. They mustn't hurt it's sales.
By the time gt214 launches, rv770 parts should be fairly cleared from the market in preparation for rv870, in which a rv740 part with higher clockspeeds replacing the 4800 series would make quite a bit of sense.
Theory of the day:
There's GOT to be a reason for the 70 slot being left open, and why rv740's highest end part has such a weak clock speed. Maybe 4770/4790 parts launch with rv870 at similar clockspeeds (rumored to be 800/4000ish, 975/5000ish). These clockspeeds on a rv740 would effectively, almost perfectly, replace the 4850 and 4870 spec for spec, with the exception of course of the two parts being closer in bandwidth, and 4790 not being able to touch 4870's 256-bit bus...but would it really need to?
This theory would help explain how rv740 could bridge the midrange of two generations. Low clocks at first to be rv770's bitch, all the while higher parts are stock-piled and 40nm becomes more mature.