Improve move_mask and boolean ops for i64x2 i32x4 i16x8 i8x16

Lokathor / wide

A crate to help you go wide. By which I mean use SIMD stuff.

https://docs.rs/wide

zlib License

288 stars 24 forks source link

Closed mcroomp closed 4 months ago

mcroomp commented 4 months ago

Improve the efficiency of move_mask and boolean operations, especially on Aarch64 Neon
Add implementation of boolean ops to i64x2 and i64x4
Add randomized testing for boolean ops
Improve testing by making randomized testing more random (use random library for tests only)
Fix missing inline for shr/shl ops

Lokathor commented 4 months ago

released wide-0.7.24