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
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154

//! Binary and scalar operation implementations for WebGPU.

use super::helpers::*;
use crate::dtype::DType;
use crate::error::Result;
use crate::runtime::wgpu::shaders::elementwise;
use crate::runtime::wgpu::{WgpuClient, WgpuRuntime};
use crate::runtime::{compute_broadcast_shape, ensure_contiguous, validate_binary_dtypes};
use crate::tensor::Tensor;

pub(crate) fn native_binary_op(
    client: &WgpuClient,
    op: &'static str,
    a: &Tensor<WgpuRuntime>,
    b: &Tensor<WgpuRuntime>,
) -> Result<Tensor<WgpuRuntime>> {
    // Use shared helpers for validation (same as CPU and CUDA backends)
    let dtype = validate_binary_dtypes(a, b)?;
    let out_shape = compute_broadcast_shape(a, b)?;

    let a_contig = ensure_contiguous(a);
    let b_contig = ensure_contiguous(b);

    let numel: usize = out_shape.iter().product();
    let out = alloc_output(client, &out_shape, dtype);

    let a_buf = get_tensor_buffer(&a_contig)?;
    let b_buf = get_tensor_buffer(&b_contig)?;
    let out_buf = get_tensor_buffer(&out)?;

    // Use broadcast kernel if shapes differ, element-wise kernel otherwise
    if a.shape() != b.shape() {
        launch_broadcast_binary(
            client,
            op,
            &a_buf,
            &b_buf,
            &out_buf,
            a.shape(),
            b.shape(),
            &out_shape,
            numel,
            dtype,
        )?;
    } else {
        let params = BinaryParams {
            numel: numel as u32,
        };
        let params_buf = create_params_buffer(client, &params);

        elementwise::launch_binary_op(
            client.pipeline_cache(),
            client.wgpu_queue(),
            op,
            &a_buf,
            &b_buf,
            &out_buf,
            &params_buf,
            numel,
            dtype,
        )?;
    }

    Ok(out)
}

/// Launch broadcast binary op with stride buffers.
#[allow(clippy::too_many_arguments)]
fn launch_broadcast_binary(
    client: &WgpuClient,
    op: &'static str,
    a_buf: &wgpu::Buffer,
    b_buf: &wgpu::Buffer,
    out_buf: &wgpu::Buffer,
    a_shape: &[usize],
    b_shape: &[usize],
    out_shape: &[usize],
    numel: usize,
    dtype: DType,
) -> Result<()> {
    let ndim = out_shape.len();

    // Compute broadcast strides (0 for broadcast dimensions)
    let a_strides = compute_broadcast_strides(a_shape, out_shape);
    let b_strides = compute_broadcast_strides(b_shape, out_shape);

    // Compute output strides (row-major)
    let mut out_strides = vec![1u32; ndim];
    for i in (0..ndim.saturating_sub(1)).rev() {
        out_strides[i] = out_strides[i + 1] * out_shape[i + 1] as u32;
    }

    // Create stride buffers
    let a_strides_buf = create_storage_buffer(client, &a_strides);
    let b_strides_buf = create_storage_buffer(client, &b_strides);
    let out_strides_buf = create_storage_buffer(client, &out_strides);

    // Create params: [numel, ndim]
    let params = BroadcastBinaryParams {
        numel: numel as u32,
        ndim: ndim as u32,
    };
    let params_buf = create_params_buffer(client, &params);

    elementwise::launch_broadcast_binary_op(
        client.pipeline_cache(),
        client.wgpu_queue(),
        op,
        a_buf,
        b_buf,
        out_buf,
        &a_strides_buf,
        &b_strides_buf,
        &out_strides_buf,
        &params_buf,
        numel,
        dtype,
    )
}

pub(crate) fn native_scalar_op(
    client: &WgpuClient,
    op: &'static str,
    a: &Tensor<WgpuRuntime>,
    scalar: f64,
) -> Result<Tensor<WgpuRuntime>> {
    let dtype = a.dtype();
    let a_contig = ensure_contiguous(a);
    let numel = a.numel();

    let out = alloc_output(client, a.shape(), dtype);

    let a_buf = get_tensor_buffer(&a_contig)?;
    let out_buf = get_tensor_buffer(&out)?;

    let params = ScalarParams {
        numel: numel as u32,
        scalar: scalar as f32,
    };
    let params_buf = create_params_buffer(client, &params);

    elementwise::launch_scalar_op(
        client.pipeline_cache(),
        client.wgpu_queue(),
        op,
        &a_buf,
        &out_buf,
        &params_buf,
        numel,
        dtype,
    )?;

    Ok(out)
}