In both graphics and compute ?

Its ok for nvidia to release an architecture that does compute very well, but barely improves graphics, and vice-versa.

But I don't recall any compute generation where there was a better product from the competition.