[dist1ll]

You can get pretty far with a branch per byte, as long as the bulk of the work is done w/ SIMD (like character classification). But yeah, LUT lookup per byte is not recommended.