Your faith in him is admirable but his strength is in rumour mongering, not in understanding the fine details of GPU architecture.
What reasons? First of all there's is zero evidence that the fixed function tessellation stage is performed on the shader core. Secondly, Nvidia has specifically indicated that there is fixed function hardware for that stage in Fermi. You're essentially ignoring all the real data and saying it must be so because you read it on a blog and it's clear you haven't evaluated the claim on its own merits because you would've realized it's nonsense.As far as tessellation, the reasons given make sense.
Bookmarks