The size of the bus is not the end all determining factor in a chip's performance. 8800GTX had a 512-bit bus. Based on that, Nvidia has done nothing except gone backwards since then.
Certainly not.
But the 680 looks to actually be their first step backwards with a flagship I have seen since, well, forever.
Compared to their past flagships: Small die limiting max performance potential, small bus, mediocre VRAM increase, lower TDP. Some might say lower TDP is good, but in respect to performance, it just says this card is not powerful enough to need more power.
Perf/mm2 improvements are impressive to engineers and nvidia's margins, but mean jack all to me as an end user who wants performance. Do. Not. Care.
I'll wager there is no way the 680 is going to offer the kind of increase nvidia has offered over their past flagship on the previous process as the same movement has in the past. Meaning the leaps seen from 8800GTX to GTX 280 and GTX285 to GTX480 are better improvements than we will see from GTX580 to GTX680. Really interested how they plan to market the 680 given their past flagships perf. improvements.
This is not nvidia style I am seeing, it's AMD style. But here it is; $550, X80 moniker. Could be the big die strategy bit nvidia on the ass even harder than it did on 40nm and GK110 is
way out, 2013 out. :thumbsdown: