valadaptive

Even if platform-specific SIMD implementations are absent, shouldn't the code fall back to the scalar implementation rather than failing to compile?

I'm playing around with this API a bit and I see that the `Parameters` API uses 32-bit MurmurHash to determine parameter IDs. While the probability of a collision between IDs...

Ah, I didn't realize it was possible to pass a custom ID. That'll definitely work for my use case.

@AdrianEddy Would you mind enabling the Issues tab on your fork?

If you're willing to help with making a macOS build, that'd be great! I haven't been able to do so myself because I don't have any Apple hardware. What kinds...

That simdnoise issue is annoying. It looks like the last version was published 3 years ago and the port to a version of simdeez which supports arm64 is a WIP...

I tried using [FastNoiseLite](https://crates.io/crates/fastnoise-lite) but it was substantially slower--10.2ms/frame vs 6.8ms/frame on my built-in benchmark. You can see my code on the [`noise` branch](https://github.com/valadaptive/ntsc-rs/tree/noise). I did a quick benchmark of...

I'm working on an simdnoise port that uses [a different SIMD library](https://crates.io/crates/wide) that supports arm64 and properly falls back to scalar operations. I'll let you know when that's done.

Try pulling the latest main branch; I've updated it with my own fork of simdnoise. Let me know if it works, and how well the benchmarks run if so.

I missed some incorrect casts in the original code (they were float-to-int conversions but should've been float-to-int bitcasts) which seem to produce NaNs only on ARM. Try again with the...