EricHiggin said:
Assuming there's no legal concern, I wouldn't be surprised to see GPU specs laid out much like how they are now, but with a 3 letter addition possibly. GPU : 1.84 TFLOPS, AMD Radeon™ based graphics engine GPU : 4.20 TFLOPS, AMD Radeon™ based graphics engine GPU : 12.? GCN TFLOPS, AMD Radeon™ based graphics engine With the way the GPU specs are laid out and worded, they may not even need to use "GCN" since they don't specify the exact model or architecture. Not sure what the PS5 TF will actually work out to depending on what version of the Navi RDNA 5000 series they use. By doing it this way they can match up the PS5 TF to be usefully comparable for the average consumer, who won't know or care what GCN means, if PS even needs to state that at all. The numbers themselves will be what matter most to the majority. Also depends on what the competition does. Whoever uses a calculation that get's them a larger number, the other will follow suit, even if that meant FP16 was used like you mentioned. |
Should also mention that... RDNA, or more specifically the implementation of RDNA in Navi is built on GCN's design principles anyway, just with a bigger focus on gaming and less on single precision compute (Aka Teraflops), hence why Navi has so much in common with older GCN GPU's.
It's like comparing AMD's Zen and Zen 2, Zen 2 is a massive deviation from Zen's initial design principles, but it's still built upon the same building blocks.
--::{PC Gaming Master Race}::--