Raytracing seems to be as much of a parallellable task as rasterization, so I'd bet that in the end what counts is throuhgput, or units x clock speed.Is there any educated guesses which is better from RT accelerations point of view - having more units or faster units, or are they just as good options respective to RT performance?
Is there any educated guesses which is better from RT accelerations point of view - having more units or faster units, or are they just as good options respective to RT performance?
Yeah, kinda.Wonder if dgpu’s will even clock that high
Yeah, kinda.
Think of Sapphire Toxic kind of SKU.
Obviously, but we're talking exotic SKUs, not exotic cooling.And even higher on LN2.
^^^^^^^ that would be wonderful!We may see AMD RT in action today here: https://mixer.com/DirectX
We may see AMD RT in action today here: https://mixer.com/DirectX
Just a catch phrase for hardware that supports: Texture Space Shading, VRS , Ray Tracing and Mesh Shaders. DirectML is omitted for now.New Directx12 Ultimate?
Quite a bad demo. Looks like they wanted to show off with all these reflections.Ha.
(very shiny, my poor eyes)
DirectStorage (the newest thing so far) is also not there.DirectML is omitted for now.
thatsthepoint.pngLooks like they wanted to show off with all these reflections.
Early Adopters asked for more
* Programmability inside AS Traversal
Seems AMD is interested in doing RT with simple shaders, preferably one single shader using DXR1.1, as opposed to the many shader approach in DXR1.0, AMD says single shader provides better performance. Microsoft says both approaches have their uses, advantages and disadvantages.Oh no... more shiny robots