Cool slideshow, though I will say, I'm a tad disappointed at the quality of images that nVidia has kept; seems that a lot aren't as high-contrast as you see in most CPU shots (as well as some of ATI/AMD's GPU shots) and I have my doubts that the grid pattern on some of the earlier ones is actually part of the GPU. Still, it all looks nice; especially once we get to the G80/G92/GT200 dies.
And I also agree; having a reference to the fabrication process would've been nice as well. For those who can't find them, I'm providing what specs I have on the GPUs themselves that isn't listed. I'm missing the die area for a number of the GPUs, which is unfortunate. Hard to find that tidbit of data a lot.
* NV4 - 8 million transistors, 350nm process.
* NV5 - 15 million transistors, 250nm process.
* NV10 - 22 million transistors, 220nm process.
* NV11 - 19 million transistors, 180nm process.
* NV15 - 25 million transistors, 180nm process.
* NV17 - 29 million transistors, 150nm process.
* NV20 - 57 million transistors, 150nm process.
* NV25 - 63 million transistors, 150nm process.
* NV40 - 222 million transistors, 130nm process.
* NV43 - 143 million transistors, 130nm process, 150 mm² die area. (mistakenly labeled in-article as NV46)
* G70 - 302 million transistors, 110nm process, 334 mm² die area.
* G80 - 681 million transistors, 90nm process, 484 mm² die area.
* G92 - 754 million transistors, 65nm process, 324 mm² die area.
* GT200 - 1,400 million transistors, 65nm process, 576 mm² die area.
Interesting to note is the upward slope in die area between G70, G80, and GT200. Of course, this can be seen as a direct consequence of more than doubling the transistor count each step, while not advancing more than a full fabrication node.