On Wed, 8 Dec 2021 at 16:00, Conor McCarthy cmccarthy@codeweavers.com wrote:
Saves a few CPU cycles on a potentially very hot code path. Based on a vkd3d-proton patch by Philip Rebohle.
I've signed of on this as the patch probably makes sense regardless of its performance implications, but could you quantify that claim above?
+#define MAKE_MAGIC(a,b,c,f) (((uint32_t)a) | (((uint32_t)b) << 8) | (((uint32_t)c) << 16) | f)
We'll probably want to add a common macro for this to vkd3d_common.h, and then use that both here and in vkd3d-shader, where we're current using the MAKE_TAG.macro.