So this non technically oriented person extrapolated to 12TF all by their lonesome, or their source expressed it in equivalent terms and completely neglected to provide the background explanation for their numbers? Either way, you’re still taking a leap of faith someone got something wrong here when the potential to be more clear was abundantly available. This is no longer an issue of dev kits where a Vega part at 12TF could be misinterpreted as representative of the final GPU. This is literally the end product, and you’re saying they’re misrepresenting it when that’s never happened before.
This talk of architecture efficiency and equivalent FLOPs is an argument that has been constructed on forums and chat rooms completely on the internet, and removed from the industry. We’ve grafted those terms into industry discussions because it’s become so pervasive and normalized that it feels consistent. We’ve added complexity and nuance to the discussion, and now we’re operating on a reverse Occam’s razor principle.
I think some people are rationalizing it this way because 12TF just doesn’t seem believable to them. However, we have no idea what Big Navi looks like yet, and Nvidia hasn’t even revealed their 7nm arch yet. We may well get 20TF flagship GPUs in 2020 and the comparative basis of that mindset should melt away. There’s so many assumed rules and boundaries people are just taking at face value that consoles can’t violate, and yet MS has plainly just revealed a SFF PC looking console that is markedly different than any other ‘console’ product the market has seen to date.
Look at the size of that thing. It’s clearly cooling a level of consumption heretofore unseen in consoles.