numr 0.5.2

High-performance numerical computing with multi-backend GPU acceleration (CPU/CUDA/WebGPU)
Documentation 
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
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121

// Fused matmul+bias operations. F32 only.
// C = A @ B + bias (fused epilogue)
// Entry points: matmul_bias_f32, batched_matmul_bias_f32

const TILE_SIZE: u32 = 16u;

var<workgroup> tile_a: array<array<f32, 16>, 16>;
var<workgroup> tile_b: array<array<f32, 16>, 16>;

struct MatmulBiasParams {
    M: u32,
    K: u32,
    N: u32,
    batch_size: u32,
}

@group(0) @binding(0) var<storage, read_write> matmul_a: array<f32>;
@group(0) @binding(1) var<storage, read_write> matmul_b: array<f32>;
@group(0) @binding(2) var<storage, read_write> matmul_bias: array<f32>;
@group(0) @binding(3) var<storage, read_write> matmul_c: array<f32>;
@group(0) @binding(4) var<uniform> matmul_params: MatmulBiasParams;

@compute @workgroup_size(16, 16, 1)
fn matmul_bias_f32(@builtin(global_invocation_id) global_id: vec3<u32>,
                   @builtin(local_invocation_id) local_id: vec3<u32>,
                   @builtin(workgroup_id) group_id: vec3<u32>) {
    let M = matmul_params.M;
    let K = matmul_params.K;
    let N = matmul_params.N;

    let row = group_id.y * TILE_SIZE + local_id.y;
    let col = group_id.x * TILE_SIZE + local_id.x;

    var sum: f32 = 0.0;

    let num_tiles = (K + TILE_SIZE - 1u) / TILE_SIZE;

    for (var t: u32 = 0u; t < num_tiles; t = t + 1u) {
        let a_col = t * TILE_SIZE + local_id.x;
        if (row < M && a_col < K) {
            tile_a[local_id.y][local_id.x] = matmul_a[row * K + a_col];
        } else {
            tile_a[local_id.y][local_id.x] = 0.0;
        }

        let b_row = t * TILE_SIZE + local_id.y;
        if (b_row < K && col < N) {
            tile_b[local_id.y][local_id.x] = matmul_b[b_row * N + col];
        } else {
            tile_b[local_id.y][local_id.x] = 0.0;
        }

        workgroupBarrier();

        for (var k: u32 = 0u; k < TILE_SIZE; k = k + 1u) {
            sum = sum + tile_a[local_id.y][k] * tile_b[k][local_id.x];
        }

        workgroupBarrier();
    }

    // Fused epilogue: add bias and write result
    if (row < M && col < N) {
        matmul_c[row * N + col] = sum + matmul_bias[col];
    }
}

@compute @workgroup_size(16, 16, 1)
fn batched_matmul_bias_f32(@builtin(global_invocation_id) global_id: vec3<u32>,
                            @builtin(local_invocation_id) local_id: vec3<u32>,
                            @builtin(workgroup_id) group_id: vec3<u32>) {
    let M = matmul_params.M;
    let K = matmul_params.K;
    let N = matmul_params.N;
    let batch_size = matmul_params.batch_size;

    let batch = group_id.z;
    if (batch >= batch_size) {
        return;
    }

    let row = group_id.y * TILE_SIZE + local_id.y;
    let col = group_id.x * TILE_SIZE + local_id.x;

    let a_batch_offset = batch * M * K;
    let b_batch_offset = batch * K * N;
    let c_batch_offset = batch * M * N;

    var sum: f32 = 0.0;

    let num_tiles = (K + TILE_SIZE - 1u) / TILE_SIZE;

    for (var t: u32 = 0u; t < num_tiles; t = t + 1u) {
        let a_col = t * TILE_SIZE + local_id.x;
        if (row < M && a_col < K) {
            tile_a[local_id.y][local_id.x] = matmul_a[a_batch_offset + row * K + a_col];
        } else {
            tile_a[local_id.y][local_id.x] = 0.0;
        }

        let b_row = t * TILE_SIZE + local_id.y;
        if (b_row < K && col < N) {
            tile_b[local_id.y][local_id.x] = matmul_b[b_batch_offset + b_row * N + col];
        } else {
            tile_b[local_id.y][local_id.x] = 0.0;
        }

        workgroupBarrier();

        for (var k: u32 = 0u; k < TILE_SIZE; k = k + 1u) {
            sum = sum + tile_a[local_id.y][k] * tile_b[k][local_id.x];
        }

        workgroupBarrier();
    }

    // Fused epilogue: add bias (same bias for all batches) and write result
    if (row < M && col < N) {
        matmul_c[c_batch_offset + row * N + col] = sum + matmul_bias[col];
    }
}