trueno 0.16.4

High-performance SIMD compute library with GPU support for matrix operations
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
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72

//! WebAssembly SIMD128 arithmetic operations (add, sub, mul, div).

#[cfg(target_arch = "wasm32")]
use std::arch::wasm32::*;

/// WASM SIMD128 vector addition.
#[target_feature(enable = "simd128")]
// SAFETY: caller ensures preconditions are met for this unsafe function
pub unsafe fn add(a: &[f32], b: &[f32], result: &mut [f32]) {
    let len = a.len();
    let mut i = 0;
    while i + 4 <= len {
        let va = v128_load(a.as_ptr().add(i) as *const v128);
        let vb = v128_load(b.as_ptr().add(i) as *const v128);
        v128_store(result.as_mut_ptr().add(i) as *mut v128, f32x4_add(va, vb));
        i += 4;
    }
    for j in i..len {
        result[j] = a[j] + b[j];
    }
}

/// WASM SIMD128 vector subtraction.
#[target_feature(enable = "simd128")]
// SAFETY: caller ensures preconditions are met for this unsafe function
pub unsafe fn sub(a: &[f32], b: &[f32], result: &mut [f32]) {
    let len = a.len();
    let mut i = 0;
    while i + 4 <= len {
        let va = v128_load(a.as_ptr().add(i) as *const v128);
        let vb = v128_load(b.as_ptr().add(i) as *const v128);
        v128_store(result.as_mut_ptr().add(i) as *mut v128, f32x4_sub(va, vb));
        i += 4;
    }
    for j in i..len {
        result[j] = a[j] - b[j];
    }
}

/// WASM SIMD128 vector multiplication.
#[target_feature(enable = "simd128")]
// SAFETY: caller ensures preconditions are met for this unsafe function
pub unsafe fn mul(a: &[f32], b: &[f32], result: &mut [f32]) {
    let len = a.len();
    let mut i = 0;
    while i + 4 <= len {
        let va = v128_load(a.as_ptr().add(i) as *const v128);
        let vb = v128_load(b.as_ptr().add(i) as *const v128);
        v128_store(result.as_mut_ptr().add(i) as *mut v128, f32x4_mul(va, vb));
        i += 4;
    }
    for j in i..len {
        result[j] = a[j] * b[j];
    }
}

/// WASM SIMD128 vector division.
#[target_feature(enable = "simd128")]
// SAFETY: caller ensures preconditions are met for this unsafe function
pub unsafe fn div(a: &[f32], b: &[f32], result: &mut [f32]) {
    let len = a.len();
    let mut i = 0;
    while i + 4 <= len {
        let va = v128_load(a.as_ptr().add(i) as *const v128);
        let vb = v128_load(b.as_ptr().add(i) as *const v128);
        v128_store(result.as_mut_ptr().add(i) as *mut v128, f32x4_div(va, vb));
        i += 4;
    }
    for j in i..len {
        result[j] = a[j] / b[j];
    }
}