rknpu2 0.0.9

Unofficial bindings to rknpu2 (librknnrt and librknnmrt)
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
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227

use {
    half::{bf16, f16},
    rknpu2_sys::{
        _rknn_tensor_format::{
            self, RKNN_TENSOR_FORMAT_MAX, RKNN_TENSOR_NC1HWC2, RKNN_TENSOR_NCHW, RKNN_TENSOR_NHWC,
            RKNN_TENSOR_UNDEFINED,
        },
        _rknn_tensor_qnt_type,
        _rknn_tensor_type::{
            self, RKNN_TENSOR_BFLOAT16, RKNN_TENSOR_BOOL, RKNN_TENSOR_FLOAT16, RKNN_TENSOR_FLOAT32,
            RKNN_TENSOR_INT4, RKNN_TENSOR_INT8, RKNN_TENSOR_INT16, RKNN_TENSOR_INT32,
            RKNN_TENSOR_INT64, RKNN_TENSOR_TYPE_MAX, RKNN_TENSOR_UINT8, RKNN_TENSOR_UINT16,
            RKNN_TENSOR_UINT32,
        },
    },
};

#[derive(Debug)]
pub struct TensorFormat;

impl TensorFormat {
    pub const NCHW: _rknn_tensor_format::Type = RKNN_TENSOR_NCHW;
    pub const NHWC: _rknn_tensor_format::Type = RKNN_TENSOR_NHWC;
    pub const NC1HWC2: _rknn_tensor_format::Type = RKNN_TENSOR_NC1HWC2;
    pub const UNDEFINED: _rknn_tensor_format::Type = RKNN_TENSOR_UNDEFINED;
    pub const MAX: _rknn_tensor_format::Type = RKNN_TENSOR_FORMAT_MAX;
}

#[derive(Debug, PartialEq, Eq, Clone, Copy)]
pub enum TensorFormatKind {
    NCHW(_rknn_tensor_format::Type),
    NHWC(_rknn_tensor_format::Type),
    NC1HWC2(_rknn_tensor_format::Type),
    UNDEFINED(_rknn_tensor_format::Type),
    Max(_rknn_tensor_format::Type),
    Other(_rknn_tensor_format::Type),
}

impl From<_rknn_tensor_format::Type> for TensorFormatKind {
    fn from(format: _rknn_tensor_format::Type) -> Self {
        match format {
            RKNN_TENSOR_NCHW => TensorFormatKind::NCHW(format),
            RKNN_TENSOR_NHWC => TensorFormatKind::NHWC(format),
            RKNN_TENSOR_NC1HWC2 => TensorFormatKind::NC1HWC2(format),
            RKNN_TENSOR_UNDEFINED => TensorFormatKind::UNDEFINED(format),
            RKNN_TENSOR_FORMAT_MAX => TensorFormatKind::Max(format),
            _ => TensorFormatKind::Other(format),
        }
    }
}

impl From<TensorFormatKind> for _rknn_tensor_format::Type {
    fn from(format: TensorFormatKind) -> Self {
        match format {
            TensorFormatKind::NCHW(_) => RKNN_TENSOR_NCHW,
            TensorFormatKind::NHWC(_) => RKNN_TENSOR_NHWC,
            TensorFormatKind::NC1HWC2(_) => RKNN_TENSOR_NC1HWC2,
            TensorFormatKind::UNDEFINED(_) => RKNN_TENSOR_UNDEFINED,
            TensorFormatKind::Max(_) => RKNN_TENSOR_FORMAT_MAX,
            TensorFormatKind::Other(a) => a,
        }
    }
}

pub struct DataType;

impl DataType {
    pub const FLOAT32: _rknn_tensor_type::Type = RKNN_TENSOR_FLOAT32;
    pub const FLOAT16: _rknn_tensor_type::Type = RKNN_TENSOR_FLOAT16;
    pub const BFLOAT16: _rknn_tensor_type::Type = RKNN_TENSOR_BFLOAT16;
    pub const INT4: _rknn_tensor_type::Type = RKNN_TENSOR_INT4;
    pub const INT8: _rknn_tensor_type::Type = RKNN_TENSOR_INT8;
    pub const UINT8: _rknn_tensor_type::Type = RKNN_TENSOR_UINT8;
    pub const INT16: _rknn_tensor_type::Type = RKNN_TENSOR_INT16;
    pub const UINT16: _rknn_tensor_type::Type = RKNN_TENSOR_UINT16;
    pub const INT32: _rknn_tensor_type::Type = RKNN_TENSOR_INT32;
    pub const UINT32: _rknn_tensor_type::Type = RKNN_TENSOR_UINT32;
    pub const INT64: _rknn_tensor_type::Type = RKNN_TENSOR_INT64;
    pub const BOOL: _rknn_tensor_type::Type = RKNN_TENSOR_BOOL;
    pub const MAX: _rknn_tensor_type::Type = RKNN_TENSOR_TYPE_MAX;
}

#[derive(Debug, PartialEq, Eq, Clone, Copy)]
pub enum DataTypeKind {
    Float32(_rknn_tensor_type::Type),
    Float16(_rknn_tensor_type::Type),
    BFloat16(_rknn_tensor_type::Type),
    Int4(_rknn_tensor_type::Type),
    Int8(_rknn_tensor_type::Type),
    UInt8(_rknn_tensor_type::Type),
    Int16(_rknn_tensor_type::Type),
    UInt16(_rknn_tensor_type::Type),
    Int32(_rknn_tensor_type::Type),
    UInt32(_rknn_tensor_type::Type),
    Int64(_rknn_tensor_type::Type),
    Bool(_rknn_tensor_type::Type),
    Max(_rknn_tensor_type::Type),
    Other(_rknn_tensor_type::Type),
}

impl From<_rknn_tensor_type::Type> for DataTypeKind {
    fn from(data_type: _rknn_tensor_type::Type) -> Self {
        match data_type {
            RKNN_TENSOR_FLOAT32 => DataTypeKind::Float32(RKNN_TENSOR_FLOAT32),
            RKNN_TENSOR_FLOAT16 => DataTypeKind::Float16(RKNN_TENSOR_FLOAT16),
            RKNN_TENSOR_BFLOAT16 => DataTypeKind::BFloat16(RKNN_TENSOR_BFLOAT16),
            RKNN_TENSOR_INT4 => DataTypeKind::Int4(RKNN_TENSOR_INT4),
            RKNN_TENSOR_INT8 => DataTypeKind::Int8(RKNN_TENSOR_INT8),
            RKNN_TENSOR_UINT8 => DataTypeKind::UInt8(RKNN_TENSOR_UINT8),
            RKNN_TENSOR_INT16 => DataTypeKind::Int16(RKNN_TENSOR_INT16),
            RKNN_TENSOR_UINT16 => DataTypeKind::UInt16(RKNN_TENSOR_UINT16),
            RKNN_TENSOR_INT32 => DataTypeKind::Int32(RKNN_TENSOR_INT32),
            RKNN_TENSOR_UINT32 => DataTypeKind::UInt32(RKNN_TENSOR_UINT32),
            RKNN_TENSOR_INT64 => DataTypeKind::Int64(RKNN_TENSOR_INT64),
            RKNN_TENSOR_BOOL => DataTypeKind::Bool(RKNN_TENSOR_BOOL),
            RKNN_TENSOR_TYPE_MAX => DataTypeKind::Max(RKNN_TENSOR_TYPE_MAX),
            _ => DataTypeKind::Other(data_type),
        }
    }
}

impl From<DataTypeKind> for _rknn_tensor_type::Type {
    fn from(data_type: DataTypeKind) -> Self {
        match data_type {
            DataTypeKind::Float32(_) => RKNN_TENSOR_FLOAT32,
            DataTypeKind::Float16(_) => RKNN_TENSOR_FLOAT16,
            DataTypeKind::BFloat16(_) => RKNN_TENSOR_BFLOAT16,
            DataTypeKind::Int4(_) => RKNN_TENSOR_INT4,
            DataTypeKind::Int8(_) => RKNN_TENSOR_INT8,
            DataTypeKind::UInt8(_) => RKNN_TENSOR_UINT8,
            DataTypeKind::Int16(_) => RKNN_TENSOR_INT16,
            DataTypeKind::UInt16(_) => RKNN_TENSOR_UINT16,
            DataTypeKind::Int32(_) => RKNN_TENSOR_INT32,
            DataTypeKind::UInt32(_) => RKNN_TENSOR_UINT32,
            DataTypeKind::Int64(_) => RKNN_TENSOR_INT64,
            DataTypeKind::Bool(_) => RKNN_TENSOR_BOOL,
            DataTypeKind::Max(_) => RKNN_TENSOR_TYPE_MAX,
            DataTypeKind::Other(a) => a,
        }
    }
}

#[derive(Debug)]
pub struct QuantType;

impl QuantType {
    pub const QNT_NONE: _rknn_tensor_qnt_type::Type = _rknn_tensor_qnt_type::RKNN_TENSOR_QNT_NONE;
    pub const QNT_DFP: _rknn_tensor_qnt_type::Type = _rknn_tensor_qnt_type::RKNN_TENSOR_QNT_DFP;
    pub const QNT_AFFINE_ASYMMETRIC: _rknn_tensor_qnt_type::Type =
        _rknn_tensor_qnt_type::RKNN_TENSOR_QNT_AFFINE_ASYMMETRIC;
}

#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub enum QuantTypeKind {
    None(_rknn_tensor_qnt_type::Type),
    Dfp(_rknn_tensor_qnt_type::Type),
    AffineAsymmetric(_rknn_tensor_qnt_type::Type),
    Other(_rknn_tensor_qnt_type::Type),
}

impl From<QuantTypeKind> for _rknn_tensor_qnt_type::Type {
    fn from(quant_type: QuantTypeKind) -> Self {
        match quant_type {
            QuantTypeKind::None(_) => QuantType::QNT_NONE,
            QuantTypeKind::Dfp(_) => QuantType::QNT_DFP,
            QuantTypeKind::AffineAsymmetric(_) => QuantType::QNT_AFFINE_ASYMMETRIC,
            QuantTypeKind::Other(a) => a,
        }
    }
}

impl From<_rknn_tensor_qnt_type::Type> for QuantTypeKind {
    fn from(quant_type: _rknn_tensor_qnt_type::Type) -> Self {
        match quant_type {
            QuantType::QNT_NONE => QuantTypeKind::None(QuantType::QNT_NONE),
            QuantType::QNT_DFP => QuantTypeKind::Dfp(QuantType::QNT_DFP),
            QuantType::QNT_AFFINE_ASYMMETRIC => {
                QuantTypeKind::AffineAsymmetric(QuantType::QNT_AFFINE_ASYMMETRIC)
            }
            _ => QuantTypeKind::Other(quant_type),
        }
    }
}

pub trait TensorType: Sized + Default {
    const TYPE: _rknn_tensor_type::Type;
}

impl TensorType for f32 {
    const TYPE: _rknn_tensor_type::Type = _rknn_tensor_type::RKNN_TENSOR_FLOAT32;
}

impl TensorType for f16 {
    const TYPE: _rknn_tensor_type::Type = _rknn_tensor_type::RKNN_TENSOR_FLOAT16;
}

impl TensorType for bf16 {
    const TYPE: _rknn_tensor_type::Type = _rknn_tensor_type::RKNN_TENSOR_BFLOAT16;
}

impl TensorType for u8 {
    const TYPE: _rknn_tensor_type::Type = _rknn_tensor_type::RKNN_TENSOR_UINT8;
}

impl TensorType for i8 {
    const TYPE: _rknn_tensor_type::Type = _rknn_tensor_type::RKNN_TENSOR_INT8;
}

impl TensorType for i32 {
    const TYPE: _rknn_tensor_type::Type = _rknn_tensor_type::RKNN_TENSOR_INT32;
}

impl TensorType for u32 {
    const TYPE: _rknn_tensor_type::Type = _rknn_tensor_type::RKNN_TENSOR_UINT32;
}

impl TensorType for i16 {
    const TYPE: _rknn_tensor_type::Type = _rknn_tensor_type::RKNN_TENSOR_INT16;
}

impl TensorType for u16 {
    const TYPE: _rknn_tensor_type::Type = _rknn_tensor_type::RKNN_TENSOR_UINT16;
}

impl TensorType for i64 {
    const TYPE: _rknn_tensor_type::Type = _rknn_tensor_type::RKNN_TENSOR_INT64;
}