On Thu Sep 28 14:19:14 2023 +0000, Giovanni Mascellani wrote:
Approved, though I guess that eventually we'd like to have just one convention on how to store swizzles (i.e., one component per byte or four components packed into a single byte). I suppose there is no fundamental reason why the HLSL compiler must be different from the rest of vkd3d-shader.
Yep, eventually we would like the HLSL compiler to match VSIR in this regard, albeit we may need to use an uint64 to also have enough space matrix swizzles.