I'd say most of those optimizations are not needed (`@inline`, `@ntuple`, `@inbounds` can also often be dropped). Vectorization, in the sense of using `.+` etc

I'd say most of those optimizations are not needed (`@inline`, `@ntuple`, `@inbounds` can also often be dropped). Vectorization, in the sense of using `.+` etc, does not provide any speedup in Julia, it is only for notational convenience, and is at best the same speed as a loop, often slower.