#simd #sse #avx #x86intrin

nightly x86intrin

x86intrin implementation

11 releases

0.4.5 Mar 11, 2018
0.4.3 Jul 25, 2017
0.4.1 Oct 19, 2016

#51 in Hardware support

Download history 53/week @ 2019-03-24 5/week @ 2019-03-31 15/week @ 2019-04-07 6/week @ 2019-04-14 14/week @ 2019-04-21 15/week @ 2019-04-28 14/week @ 2019-05-05 23/week @ 2019-05-12 16/week @ 2019-05-19 11/week @ 2019-05-26 12/week @ 2019-06-02 45/week @ 2019-06-09 34/week @ 2019-06-16 176/week @ 2019-06-23 45/week @ 2019-06-30

63 downloads per month
Used in 4 crates (3 directly)

MIT/Apache

550KB
10K SLoC

x86intrin

x86 intrinsics for rust

Crates.io Status Build Status

This crate implements C-like x86 intrinsics. The function and type names are the same as what Intel uses (removing prefix underscore; e.g. __m128i -> m128i, _mm_set_epi32 -> mm_set_epi32).

I actually need various integer SIMD arithmetics, so such functions will be implemeneted with priority.

Current Status

Currently most of SSE, SSE2, SSE3, SSSE3, SSE4.1, SSE4.2, AVX, and AVX2 are implemented. Some of the functions cannot be implemented since rust is not exposing necessary functions.

After all done, I'd like to contribute to rust libraries to support missing functions.

Note

You need nightly channel, since this crate uses unstable features.

To build with cargo, you need to set target-cpu or target-feature in RUSTFLAGS.

For example:

$ RUSTFLAGS="-C target-cpu=native" cargo build
$ RUSTFLAGS="-C target-feature=+sse3" cargo build

No runtime deps