himada-dispatch 0.1.1

Adaptive SIMD dispatch for Himada — auto-selects fastest kernel at runtime
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33

pub mod conv1d;
pub mod cumsum;
pub mod dot;
pub mod fused;
pub mod gemv;
pub mod int;
pub mod layernorm;
pub mod matmul;
pub mod memchr;
pub mod ops;
pub mod reduce;
pub mod softmax;
pub mod sort;
pub mod sve;
pub mod trig;
pub mod wasm_simd;

pub use conv1d::*;
pub use cumsum::*;
pub use dot::*;
pub use fused::*;
pub use gemv::*;
pub use int::*;
pub use layernorm::*;
pub use matmul::*;
pub use memchr::*;
pub use ops::*;
pub use reduce::*;
pub use softmax::*;
pub use sort::*;
pub use sve::*;
pub use trig::*;
pub use wasm_simd::*;