runmat-accelerate 0.4.4

Pluggable GPU acceleration layer for RunMat (CUDA, ROCm, Metal, Vulkan/Spir-V)
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
155

pub const DIAG_FROM_VECTOR_SHADER_F64: &str = r#"
struct Vector {
    data: array<f64>,
};

struct Matrix {
    data: array<f64>,
};

struct DiagVecParams {
    len: u32,
    size: u32,
    offset: i32,
    _pad: u32,
};

@group(0) @binding(0) var<storage, read> Input: Vector;
@group(0) @binding(1) var<storage, read_write> Output: Matrix;
@group(0) @binding(2) var<uniform> params: DiagVecParams;

@compute @workgroup_size(@WG@)
fn main(@builtin(global_invocation_id) gid: vec3<u32>) {
    let idx = gid.x;
    if idx >= params.len {
        return;
    }
    var row = idx;
    var col = idx;
    if params.offset >= 0 {
        col = idx + u32(params.offset);
    } else {
        let shift = u32(-params.offset);
        row = idx + shift;
    }
    let base = row + col * params.size;
    Output.data[base] = Input.data[idx];
}
"#;

pub const DIAG_FROM_VECTOR_SHADER_F32: &str = r#"
struct Vector {
    data: array<f32>,
};

struct Matrix {
    data: array<f32>,
};

struct DiagVecParams {
    len: u32,
    size: u32,
    offset: i32,
    _pad: u32,
};

@group(0) @binding(0) var<storage, read> Input: Vector;
@group(0) @binding(1) var<storage, read_write> Output: Matrix;
@group(0) @binding(2) var<uniform> params: DiagVecParams;

@compute @workgroup_size(@WG@)
fn main(@builtin(global_invocation_id) gid: vec3<u32>) {
    let idx = gid.x;
    if idx >= params.len {
        return;
    }
    var row = idx;
    var col = idx;
    if params.offset >= 0 {
        col = idx + u32(params.offset);
    } else {
        let shift = u32(-params.offset);
        row = idx + shift;
    }
    let base = row + col * params.size;
    Output.data[base] = Input.data[idx];
}
"#;

pub const DIAG_EXTRACT_SHADER_F64: &str = r#"
struct Matrix {
    data: array<f64>,
};

struct Vector {
    data: array<f64>,
};

struct DiagExtractParams {
    rows: u32,
    cols: u32,
    offset: i32,
    diag_len: u32,
};

@group(0) @binding(0) var<storage, read> Input: Matrix;
@group(0) @binding(1) var<storage, read_write> Output: Vector;
@group(0) @binding(2) var<uniform> params: DiagExtractParams;

@compute @workgroup_size(@WG@)
fn main(@builtin(global_invocation_id) gid: vec3<u32>) {
    let idx = gid.x;
    if idx >= params.diag_len {
        return;
    }
    var row = idx;
    var col = idx;
    if params.offset >= 0 {
        col = idx + u32(params.offset);
    } else {
        let shift = u32(-params.offset);
        row = idx + shift;
    }
    let base = row + col * params.rows;
    Output.data[idx] = Input.data[base];
}
"#;

pub const DIAG_EXTRACT_SHADER_F32: &str = r#"
struct Matrix {
    data: array<f32>,
};

struct Vector {
    data: array<f32>,
};

struct DiagExtractParams {
    rows: u32,
    cols: u32,
    offset: i32,
    diag_len: u32,
};

@group(0) @binding(0) var<storage, read> Input: Matrix;
@group(0) @binding(1) var<storage, read_write> Output: Vector;
@group(0) @binding(2) var<uniform> params: DiagExtractParams;

@compute @workgroup_size(@WG@)
fn main(@builtin(global_invocation_id) gid: vec3<u32>) {
    let idx = gid.x;
    if idx >= params.diag_len {
        return;
    }
    var row = idx;
    var col = idx;
    if params.offset >= 0 {
        col = idx + u32(params.offset);
    } else {
        let shift = u32(-params.offset);
        row = idx + shift;
    }
    let base = row + col * params.rows;
    Output.data[idx] = Input.data[base];
}
"#;