not a significant ammount - basically the size of another kernel + some non shared resources so I'd guess around 200 MB - 2 GB
the bigger problem is how much cache this thing will hog because cache is very valuable and degraded cache performance can mean order of magnitude slowdowns in high access code.
actually yes and no
from what I understand, DX12 is more of a minor house cleaning in terms of structure and an attempt to catch up/leapfrog PSGL/OpenGL on the control side of things (CPU) since previously OpenGL and DirectX were pretty bad
IMO DirectX is way way uglier if you want to drop down and do some lower level stuff at a critical bottleneck so this seems promising
another key observation to be made is that in accordance with Amdahl...
> You do know PS4 runs on OpenGL... right?
OpenGL already has a far better immediate mode than D3D
DX12 is a catch-up/house cleaning to bring it up to speed with OpenGL/Mantle and PSGL which is based upon OpenGL/Mantle
it's a big change for direct x but more of a catchup move relative to OpenGL than anything
this won't help the XB1 much other than making the code nicer to deal with - less antiquated/messy api names/parameters and such
the XBO also does not use DX 11.2 - it uses DX 11.X which is a fork of 11.2 that removes some abstraction/compatibility scaffolding unnecessary for consoles (due 1 SKU spec)
I do
DX12 may be able to bring the XB up a few % by reducing overhead and giving better lower level access which thus far has not been very nice on with any DX device/version but very nice on PS but it is PHYSICALLY IMPOSSIBLE for the hardware's ALU to perform better than it already is.
Clock-speed and memory properties govern the maximum computational throughput of a GPU - the XB1 has 40% lower maximum computational throughput end of story.
remove the parity clause hopefully
I've been developing games for about a year now. What he said is correct.