I've seen that as well.. but, the numbers don't seem to add up to me... at a high-level here... we have a 250W 2080 Ti that is (in FE form) ~45% faster than a 5700XT (225W reference). AMD is bringing an updated/new arch to the table on the same node, but perhaps a tweaked process. To reach flagship levels of NV from 2018, they are already 45% in the hole. If Ampre is just 30-35% faster (similar to 1080Ti vs 2080 Ti), that means the new card needs to be at least ~75% faster just to match Ampre. I don't recall seeing such a significant jump from flagship to flagship in one generation...and I think a 30% increase is tiny for what they are saying power use is going to be. Where does big Navi perform with this in mind? To reach ampre, is it going to be 300W? Does it need to run closer to the limit like the 5700 XT does?
That said, on the NV side... we're already at 250W with 2080 Ti. Ampre is a node shrink AND brand new architecture. With this comes performance /W improvements out of the box. So if Ampre is to use 20% more power even after a node shrink and have IPC/architectural improvements... where will its performance land? Do we expect 30%? 50%? More?
That's how I'm thinking of things.