Closed spease closed 6 years ago
You should be able to do something like
vec.par_iter(|a| a.simd_iter().simd_map(|b| ...))
Faster is a library containing operations that can only be executed on a single thread, so it should be perfectly compatible with Rayon.
The above code won't compile on 0.2.1 because the API is unstable, but that should be the gist of it.
So I'd need to be using master? Any general idea of when you're planning on making an official version that is compatible?
0.3.0 is coming out at the end of December, and it breaks more than half the API. You can use 0.2.1 with Rayon as well - I think it's just .into_simd_iter().map(|a| {..})
instead. Just do whatever SIMD operations you want in your per-core closure.
Apologies for the confusion - I've forgotten what 0.2.1 does and doesn't contain at this point.
0.3.0 has been released, and you should be able to use this with Rayon without much issue. If you have any further questions, feel free to re-open or file another issue.
It wasn’t clear to me if faster can also work with rayon. I tried the simple example both with into_* and placing par_iter before and after simd_iter with no luck. So it looks to me like this isn’t a supported combination right now, but if it is, an example would be nice.
Thanks!