Post #3647485
2026-04-20 17:01 UTC
Implicit in the comments is that a Vec4 is a glam struct backed by an __m128 SIMD value. The point of all of this is to do SIMD-optimized math while still having access to Vec4::w without paying the deref cost
Replies (1)
-
@esther_alter@mastodon.social 2026-04-20 18:12
And, this function is part of a fun-code project to see if I can accelerate color blending via SIMD. So far, derefing to access individual color channels seems to be the bottleneck