The interesting bit is that the G80 is already huge, 480mm2 or so, a little bigger than the R600. Once you add in the necessary 50mm2 or so of the I/O chip, that brings the total to 530mm2, notably larger than the R600. For comparison the G71 is 197mm2 including I/O, so this one is almost triple the size.
It looks like Nvidia ran into problems getting the threading engine out on a reasonable size budget while ATI is on its third or fourth generation of the concept, if you could the 360. This has interesting implications for integrated pipelines, basically the first gen is big, then when you figure it out, it gets smaller and faster.
So Nvidia pulled out the end parts of the chip and made them external, not a huge deal in and of itself. While a 480mm2 + 50mm2 chips is certainly cheaper than a 530mm2 chip, it adds costs because in addition to fabbing a second part, you also have to manufacture a board with two chips instead of one. For the wild and wonderful world of halo parts, this is not a big deal, but it will be very interesting to see how far down the line this idea goes. µ
Sign up for INQbot – a weekly roundup of the best from the INQ