cubecl-convolution 0.7.0

CubeCL Convolution Kernels Engine
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

use cubecl_core::prelude::TensorHandleRef;
use cubecl_core::{Runtime, client::ComputeClient};
use cubecl_matmul::{
    MatmulInputHandleRef,
    components::{
        InputArg, InputRuntimeArg, MatmulLineSizes, MatmulSelection, MatmulSpec, OutputArg,
        OutputRuntimeArg, global::GlobalConfig as _,
    },
};

use crate::{
    components::{
        ConvSetupError, ConvolutionProblem,
        global::{
            args::{ConcreteInputsFactory, ConcreteOutputFactory},
            entry_point::ConvolutionLaunch,
        },
    },
    kernels::layered::algorithm::Algorithm,
};

/// Select which kernel to launch for the given Algorithm.
///
/// Only works for concrete tensor inputs and output.
#[allow(clippy::result_large_err, clippy::too_many_arguments)]
pub fn launch_kernel_concrete<MS: MatmulSpec, R: Runtime, A: Algorithm>(
    client: &ComputeClient<R::Server>,
    input: &MatmulInputHandleRef<'_, R>,
    weight: &MatmulInputHandleRef<'_, R>,
    bias: &Option<TensorHandleRef<'_, R>>,
    out: &TensorHandleRef<'_, R>,
    problem: ConvolutionProblem,
    line_sizes: MatmulLineSizes,
    selection: MatmulSelection,
) -> Result<(), ConvSetupError>
where
    InputArg<MS>: ConcreteInputsFactory,
    OutputArg<MS>: ConcreteOutputFactory,
{
    let config = A::setup::<R, MS::Precision>(client, &problem, &selection, &line_sizes)?;

    let input = <InputArg<MS> as ConcreteInputsFactory>::create(
        client,
        input,
        weight,
        bias.as_ref(),
        &selection,
        &problem,
        &line_sizes,
        config,
    );
    let output = <OutputArg<MS> as ConcreteOutputFactory>::create(
        client,
        out,
        &selection,
        &problem,
        &line_sizes,
        config,
    );

    unsafe {
        A::GlobalConvolution::launch_unchecked::<MS, R>(
            client,
            config.cube_dim(),
            A::cube_count(&selection, &problem),
            input,
            output,
            &problem,
            config,
        );
    }

    Ok(())
}

/// Select which kernel to launch for the given Algorithm.
pub fn launch_kernel_virtual<'a, MS: MatmulSpec, R: Runtime, A: Algorithm>(
    client: &ComputeClient<R::Server>,
    input: InputRuntimeArg<'a, MS, R>,
    output: OutputRuntimeArg<'a, MS, R>,
    problem: ConvolutionProblem,
    line_sizes: MatmulLineSizes,
    selection: MatmulSelection,
) -> Result<(), ConvSetupError> {
    let config = A::setup::<R, MS::Precision>(client, &problem, &selection, &line_sizes)?;

    unsafe {
        A::GlobalConvolution::launch_unchecked::<MS, R>(
            client,
            config.cube_dim(),
            A::cube_count(&selection, &problem),
            input,
            output,
            &problem,
            config,
        );
    }

    Ok(())
}