cubek-test-utils 0.2.0-pre.3

CubeK: Test Utils
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

use cubecl::{
    TestRuntime, calculate_cube_count_elemwise,
    prelude::*,
    std::tensor::{TensorHandle, ViewOperationsMut, ViewOperationsMutExpand},
    zspace::{Shape, Strides},
};

use crate::BaseInputSpec;

#[cube(launch)]
fn custom_data_launch<T: Numeric>(
    tensor: &mut Tensor<T>,
    contiguous_data: Array<f32>,
    #[define(T)] _types: StorageType,
) {
    let linear = ABSOLUTE_POS;

    if linear >= tensor.len() {
        terminate!();
    }

    let mut remaining = linear;
    let mut offset = 0;

    for d in 0..tensor.rank() {
        let dim = tensor.shape(tensor.rank() - 1 - d);
        let idx = remaining % dim;
        remaining /= dim;
        offset += idx * tensor.stride(tensor.rank() - 1 - d);
    }

    tensor.write_checked(offset, T::cast_from(contiguous_data[linear]));
}

fn new_custom_data(
    client: &ComputeClient<TestRuntime>,
    shape: Shape,
    strides: Strides,
    dtype: StorageType,
    contiguous_data: Vec<f32>,
) -> TensorHandle<TestRuntime> {
    let num_elems = shape.iter().product::<usize>();
    assert!(contiguous_data.len() == num_elems);

    // Performance is not important here and this simplifies greatly the problem
    let vector_size = 1;

    let working_units: u32 = num_elems as u32 / vector_size as u32;
    let cube_dim = CubeDim::new(client, working_units as usize);
    let cube_count = calculate_cube_count_elemwise(client, working_units as usize, cube_dim);

    let out = TensorHandle::new(
        client.empty(dtype.size() * num_elems),
        shape,
        strides,
        dtype,
    );

    let contiguous_handle = client.create_from_slice(f32::as_bytes(&contiguous_data));

    custom_data_launch::launch::<TestRuntime>(
        client,
        cube_count,
        cube_dim,
        out.clone().into_arg(),
        unsafe { ArrayArg::from_raw_parts(contiguous_handle, num_elems) },
        dtype,
    );

    out
}

pub(crate) fn build_custom(
    base_spec: BaseInputSpec,
    contiguous_data: Vec<f32>,
) -> TensorHandle<TestRuntime> {
    new_custom_data(
        &base_spec.client,
        base_spec.shape.clone(),
        base_spec.strides(),
        base_spec.dtype,
        contiguous_data,
    )
}