The title says fast, but no benches, but the README puts more emphasis on that it’s a more usable solution than LLVM-PTX + Rust.

I mean, what’s the point of “fast”, if LLVM-PTX is clunky with rust in the first place?