Boring from a flat out performance improvement? Yes. But it's anything but in all other respects. Innovation reigned supreme in the face of technological stagnation and cost barriers. GM204 destroyed Tahiti in performance (55-60%) at lower power consumption, lower memory bandwidth, and only a 13% increase in power consumption. GM204 also beats the two bigger chips in all the same metrics with a noticeably smaller die size. On top of all that, GM204 has incredible overclocking headroom which actually expands on it's lead over the Hawaii and GK110.
It's a taste of what is to come with finfets.
It has failed from absolute performance given the timeframe, relative performance jump vs. previous 780Ti/290X, and on price/performance - so all 3 most valuable metrics for 20 years of desktop GPUs until performance/watt marketing took over. 980 is basically a 6600GT/460 1GB, nowhere near worthy the x80 designation, far worse than 580--> 680.
In your comparison to Tahiti, You literally ignored one the most important aspects -- Time. GM204 may be faster than Tahiti, but since it's been 3 years since Tahiti came out and GPUs
at least double every 3 years (already way off historical 2x faster every 2 years), GM204 badly misses the mark as "Tahiti's" 3-year successor if you want to look at it that way. It's not NV's fault high performance 20nm node is dead or alive, don't get me wrong, but strictly as both a 3 year newer GPU and as an x80 series part bringing just 7-10% more performance over 780Ti, 980 is way off the mark. A $550 card 3 years after Tahiti launched should be at least 2X faster stock:
http://www.computerbase.de/2013-12/grafikkarten-2013-vergleich/10/
Secondly, it happens to be the most overpriced x80 part relative to x70 part in probably a decade, maybe ever, actually more
than twice as expensive in terms of $ spent for every 1% gained over the x70 part compared to say 470 vs. 480.
Thirdly, it changes very little in terms of moving 4K performance over 1 year old 290X/780Ti. If it wasn't for 970's great price/performance at launch, GM204 on the desktop would be a major failure in terms of GPU history. Now, I have doubts that even 390X will double Tahiti's performance but honestly I expect nothing less given the timeframe. If not, GM200 will do it :biggrin:
NV is just doing an exact repeat of 680-->780Ti, bifurcating the generation into 2, maybe even 3 parts (possible shrink of GM200 in 2016). If AMD doesn't launch 390X as a 520mm2 28nm part or a 20nm part, I have little hope for that card either.
At the end I guess it comes down to the games. PC software in 2014 has been way more disappointing from a technical perspective in pushing the PC than I imagined. I expected next gen PC games to start trickling but besides Ryse Son of Rome, none lived up to the hype so far from a graphical/technical point of view. Maybe Dragon Age Inquisition can still surprise. I have little hope that AC Unity or FC4 will change my opinion. Watch Dogs failed.
Unless 4K IPS monitors drop in price a lot more or there will be more games like Witcher 3 and Project CARS pushing the PC envelope, we might be in a serious drought of technical advancement until BF5.
Also, based on NV's bifurcating of a GPU generation, I am putting flagship GP200 Pascal at Q1 2017 at best even though NV has penciled Pascal for 2016. That will probably be low end 750Ti style card and mid-range GP204 parts.