Quote Originally Posted by hurleybird View Post
He was right about both SP number and relative performance. He predicted 480SP a long time ago, that was right. Not too long ago a couple of his sources got their hands on 512SP cards, but only clocked at 600 and 625MHz (as stated in the article), and that's where the 5% faster claim came from. Obviously the final 480 shader parts clocked at 700Mhz (with the rest of the core clocked at 700MHz as well) perform a little bit better. Charlie actually posted the facts this time, albeit with a bit too much 'dear leader'-esque Nvidia. Nvidia taking a long time to decide the final specs in no way invalidates any of the facts charlie gleaned.

As far as tessellation, he claimed Fermi was better in heaven benchmark but suffered when other work had to be done on the shaders. At face value this looks true -- Fermi performs a lot better in Heaven than actual DX11 games -- but is that due to Heaven using more tessellation, or actual DX11 games needing more shading power? We probably won't know that for a very long time, if ever.
Charlie never said 480SP. He first said 512, then said it was "castrated" to 448, then said it would be 512 again. He was never accurate on SP count, and he was never accurate on clocks. He predicted 600-625 mhz all the time, whereas even the 512SP part was to have 650mhz clocks.

His "no tessellator, it's going to be emulated, will suck" claim was 180 degrees wrong. Maybe tess. performance isn't as good as Nvidia claimed but it's definitely no worse than ATI.