HollyGamer said: Do you realize each of GCN architecture have their own number and code , GCN 5 is what Vega are using now, it might have some similarities, but as far as i know we haven't got any improvement since Nvidia and Radeon invent CUDA cores and GCN. |
I would never have guessed. You are just essentially rehashing my prior statement though.
And yes. There has been improvements.
HollyGamer said: Agree that's why i mentioned it on the OP. What they have now for SDK are probably Vega |
"Probably" is not something I can adhere to.
It is likely a semi-custom GCN base with forward-like navi features.
Maybe just maybe... AMD has sorted out their Primitive Shaders and Draw Stream Binning Rasterizer in Navi? :P
Graphics Core Next is modular remember... AMD can take a Graphics Core Next layout and just update certain blocks on the chip whilst leaving the rest identical.
HollyGamer said: Reducing the cost by making it it more smaller |
There is more to chip costs than just size.
Patterning, Leakage, Spare-Area... You name it.
And contrary to popular belief... Just shrinking a chip doesn't always mean it will be cheaper to produce, sometimes the contrary can remain true.
HollyGamer said: Try to look at the AMD GPU road map, it clearly said "scalability " in Navi |
"Availability" is one of the design principles of GCN to begin with.
HollyGamer said: That's obvious. Jaguar that used on PS4 and Xbox One are clearly not the whole chip they cut a corner to make it simple and reducing cost and TDP. |
Not only that... But in the consoles things like power delivery, power states, video decode/encode, caches, memory controllers and various other bits and pieces can be consolidated/reduced.
--::{PC Gaming Master Race}::--