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
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
|
// Copyright (c) 2022, ARM Limited.
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
// THIS FILE IS GENERATED. DO NOT EDIT!
// See scripts/operator_api/generate_api.py
#ifndef OPERATORS_H_
#define OPERATORS_H_
#include <stddef.h>
#include <stdint.h>
#ifdef __cplusplus
extern "C"
{
#endif /* __cplusplus */
// Note status needs to be aligned with graph_status
enum tosa_status_t
{
tosa_status_valid = 0,
tosa_status_unpredictable = 1,
tosa_status_error = 2
};
enum tosa_mode_t
{
tosa_mode_unknown = 0,
tosa_mode_nearest = 1,
tosa_mode_bilinear = 2,
tosa_mode_min = 3,
tosa_mode_max = 4
};
enum tosa_datatype_t
{
tosa_datatype_bf16_t = 0,
tosa_datatype_bool_t = 1,
tosa_datatype_fp16_t = 2,
tosa_datatype_fp32_t = 3,
tosa_datatype_int16_t = 4,
tosa_datatype_int32_t = 5,
tosa_datatype_int48_t = 6,
tosa_datatype_int4_t = 7,
tosa_datatype_int8_t = 8,
tosa_datatype_uint16_t = 9,
tosa_datatype_uint8_t = 10,
};
struct tosa_tensor_t
{
int32_t* shape;
int32_t num_dims;
tosa_datatype_t data_type;
uint8_t* data;
size_t size;
};
tosa_status_t tosa_run_argmax(tosa_tensor_t client_input, const int32_t client_axis, tosa_tensor_t client_output);
tosa_status_t tosa_run_avg_pool2d(tosa_tensor_t client_input,
const int32_t client_kernel[2],
const int32_t client_stride[2],
const int32_t client_pad[4],
const int32_t client_input_zp,
const int32_t client_output_zp,
tosa_tensor_t client_output);
tosa_status_t tosa_run_conv2d(tosa_tensor_t client_input,
tosa_tensor_t client_weight,
tosa_tensor_t client_bias,
const int32_t client_pad[4],
const int32_t client_stride[2],
const int32_t client_dilation[2],
const int32_t client_input_zp,
const int32_t client_weight_zp,
tosa_tensor_t client_output);
tosa_status_t tosa_run_conv3d(tosa_tensor_t client_input,
tosa_tensor_t client_weight,
tosa_tensor_t client_bias,
const int32_t client_pad[6],
const int32_t client_stride[3],
const int32_t client_dilation[3],
const int32_t client_input_zp,
const int32_t client_weight_zp,
tosa_tensor_t client_output);
tosa_status_t tosa_run_depthwise_conv2d(tosa_tensor_t client_input,
tosa_tensor_t client_weight,
tosa_tensor_t client_bias,
const int32_t client_pad[4],
const int32_t client_stride[2],
const int32_t client_dilation[2],
const int32_t client_input_zp,
const int32_t client_weight_zp,
tosa_tensor_t client_output);
tosa_status_t tosa_run_fully_connected(tosa_tensor_t client_input,
const int32_t client_input_zp,
const int32_t client_weight_zp,
tosa_tensor_t client_output);
tosa_status_t tosa_run_matmul(tosa_tensor_t client_a,
tosa_tensor_t client_b,
const int32_t client_a_zp,
const int32_t client_b_zp,
tosa_tensor_t client_output);
tosa_status_t tosa_run_max_pool2d(tosa_tensor_t client_input,
const int32_t client_kernel[2],
const int32_t client_stride[2],
const int32_t client_pad[4],
const int32_t client_input_zp,
const int32_t client_output_zp,
tosa_tensor_t client_output);
tosa_status_t tosa_run_transpose_conv2d(tosa_tensor_t client_input,
tosa_tensor_t client_weight,
tosa_tensor_t client_bias,
const int32_t client_out_pad[4],
const int32_t client_stride[2],
const int32_t client_out_shape[4],
const int32_t client_input_zp,
const int32_t client_weight_zp,
const int32_t client_pad_len,
const int32_t client_pad[],
const int32_t client_dilation_len,
const int32_t client_dilation[],
tosa_tensor_t client_output);
tosa_status_t tosa_run_clamp(tosa_tensor_t client_input,
const int32_t client_min_int,
const int32_t client_max_int,
const float client_min_fp,
const float client_max_fp,
tosa_tensor_t client_output);
tosa_status_t tosa_run_sigmoid(tosa_tensor_t client_input, tosa_tensor_t client_output);
tosa_status_t tosa_run_tanh(tosa_tensor_t client_input, tosa_tensor_t client_output);
tosa_status_t tosa_run_add(tosa_tensor_t client_input1, tosa_tensor_t client_input2, tosa_tensor_t client_output);
tosa_status_t tosa_run_arithmetic_right_shift(tosa_tensor_t client_input1,
tosa_tensor_t client_input2,
const bool client_round,
tosa_tensor_t client_output);
tosa_status_t
tosa_run_bitwise_and(tosa_tensor_t client_input1, tosa_tensor_t client_input2, tosa_tensor_t client_output);
tosa_status_t
tosa_run_bitwise_or(tosa_tensor_t client_input1, tosa_tensor_t client_input2, tosa_tensor_t client_output);
tosa_status_t
tosa_run_bitwise_xor(tosa_tensor_t client_input1, tosa_tensor_t client_input2, tosa_tensor_t client_output);
tosa_status_t
tosa_run_intdiv(tosa_tensor_t client_input1, tosa_tensor_t client_input2, tosa_tensor_t client_output);
tosa_status_t
tosa_run_logical_and(tosa_tensor_t client_input1, tosa_tensor_t client_input2, tosa_tensor_t client_output);
tosa_status_t tosa_run_logical_left_shift(tosa_tensor_t client_input1,
tosa_tensor_t client_input2,
tosa_tensor_t client_output);
tosa_status_t tosa_run_logical_right_shift(tosa_tensor_t client_input1,
tosa_tensor_t client_input2,
tosa_tensor_t client_output);
tosa_status_t
tosa_run_logical_or(tosa_tensor_t client_input1, tosa_tensor_t client_input2, tosa_tensor_t client_output);
tosa_status_t
tosa_run_logical_xor(tosa_tensor_t client_input1, tosa_tensor_t client_input2, tosa_tensor_t client_output);
tosa_status_t
tosa_run_maximum(tosa_tensor_t client_input1, tosa_tensor_t client_input2, tosa_tensor_t client_output);
tosa_status_t
tosa_run_minimum(tosa_tensor_t client_input1, tosa_tensor_t client_input2, tosa_tensor_t client_output);
tosa_status_t tosa_run_mul(tosa_tensor_t client_input1,
tosa_tensor_t client_input2,
const uint8_t client_shift,
tosa_tensor_t client_output);
tosa_status_t tosa_run_pow(tosa_tensor_t client_input1, tosa_tensor_t client_input2, tosa_tensor_t client_output);
tosa_status_t tosa_run_sub(tosa_tensor_t client_input1, tosa_tensor_t client_input2, tosa_tensor_t client_output);
tosa_status_t tosa_run_table(tosa_tensor_t client_input,
const int32_t client_table_len,
const int16_t client_table[],
tosa_tensor_t client_output);
tosa_status_t tosa_run_abs(tosa_tensor_t client_input1, tosa_tensor_t client_output);
tosa_status_t tosa_run_bitwise_not(tosa_tensor_t client_input1, tosa_tensor_t client_output);
tosa_status_t tosa_run_ceil(tosa_tensor_t client_input1, tosa_tensor_t client_output);
tosa_status_t tosa_run_clz(tosa_tensor_t client_input1, tosa_tensor_t client_output);
tosa_status_t tosa_run_exp(tosa_tensor_t client_input1, tosa_tensor_t client_output);
tosa_status_t tosa_run_floor(tosa_tensor_t client_input1, tosa_tensor_t client_output);
tosa_status_t tosa_run_log(tosa_tensor_t client_input1, tosa_tensor_t client_output);
tosa_status_t tosa_run_logical_not(tosa_tensor_t client_input1, tosa_tensor_t client_output);
tosa_status_t tosa_run_negate(tosa_tensor_t client_input1,
const int32_t client_input1_zp,
const int32_t client_output_zp,
tosa_tensor_t client_output);
tosa_status_t tosa_run_reciprocal(tosa_tensor_t client_input1, tosa_tensor_t client_output);
tosa_status_t tosa_run_rsqrt(tosa_tensor_t client_input1, tosa_tensor_t client_output);
tosa_status_t tosa_run_select(tosa_tensor_t client_input1,
tosa_tensor_t client_input2,
tosa_tensor_t client_input3,
tosa_tensor_t client_output);
tosa_status_t tosa_run_equal(tosa_tensor_t client_input1, tosa_tensor_t client_input2, tosa_tensor_t client_output);
tosa_status_t
tosa_run_greater(tosa_tensor_t client_input1, tosa_tensor_t client_input2, tosa_tensor_t client_output);
tosa_status_t
tosa_run_greater_equal(tosa_tensor_t client_input1, tosa_tensor_t client_input2, tosa_tensor_t client_output);
tosa_status_t
tosa_run_reduce_all(tosa_tensor_t client_input, const int32_t client_axis, tosa_tensor_t client_output);
tosa_status_t
tosa_run_reduce_any(tosa_tensor_t client_input, const int32_t client_axis, tosa_tensor_t client_output);
tosa_status_t
tosa_run_reduce_max(tosa_tensor_t client_input, const int32_t client_axis, tosa_tensor_t client_output);
tosa_status_t
tosa_run_reduce_min(tosa_tensor_t client_input, const int32_t client_axis, tosa_tensor_t client_output);
tosa_status_t
tosa_run_reduce_product(tosa_tensor_t client_input, const int32_t client_axis, tosa_tensor_t client_output);
tosa_status_t
tosa_run_reduce_sum(tosa_tensor_t client_input, const int32_t client_axis, tosa_tensor_t client_output);
tosa_status_t tosa_run_concat(tosa_tensor_t client_input1, const int32_t client_axis, tosa_tensor_t client_output);
tosa_status_t tosa_run_pad(tosa_tensor_t client_input1,
const int32_t client_padding_len,
const int32_t client_padding[],
const int32_t client_pad_const_int,
const float client_pad_const_fp,
tosa_tensor_t client_output);
tosa_status_t tosa_run_reshape(tosa_tensor_t client_input1,
const int32_t client_new_shape_len,
const int32_t client_new_shape[],
tosa_tensor_t client_output);
tosa_status_t tosa_run_reverse(tosa_tensor_t client_input, const int32_t client_axis, tosa_tensor_t client_output);
tosa_status_t tosa_run_slice(tosa_tensor_t client_input1,
const int32_t client_start_len,
const int32_t client_start[],
const int32_t client_size_len,
const int32_t client_size[],
tosa_tensor_t client_output);
tosa_status_t tosa_run_tile(tosa_tensor_t client_input1,
const int32_t client_multiplies_len,
const int32_t client_multiplies[],
const int32_t client_multiples_len,
const int32_t client_multiples[],
tosa_tensor_t client_output);
tosa_status_t tosa_run_transpose(tosa_tensor_t client_input1,
const int32_t client_perms_len,
const int32_t client_perms[],
tosa_tensor_t client_output);
tosa_status_t
tosa_run_gather(tosa_tensor_t client_values, tosa_tensor_t client_indices, tosa_tensor_t client_output);
tosa_status_t tosa_run_scatter(tosa_tensor_t client_values_in,
tosa_tensor_t client_indices,
tosa_tensor_t client_input,
tosa_tensor_t client_values_out);
tosa_status_t tosa_run_resize(tosa_tensor_t client_input,
const int16_t client_scale[4],
const int16_t client_offset[2],
const int16_t client_border[2],
const tosa_mode_t client_mode,
tosa_tensor_t client_output);
tosa_status_t tosa_run_cast(tosa_tensor_t client_input, tosa_tensor_t client_output);
tosa_status_t tosa_run_rescale(tosa_tensor_t client_input,
tosa_tensor_t client_output,
const int32_t client_input_zp,
const int32_t client_output_zp,
const int32_t client_multiplier_len,
const int32_t client_multiplier[],
const int32_t client_shift_len,
const uint8_t client_shift[],
const bool client_scale32,
const bool client_double_round,
const bool client_per_channel);
tosa_status_t tosa_run_identity(tosa_tensor_t client_input1, tosa_tensor_t client_output);
#ifdef __cplusplus
}
#endif /* __cplusplus */
#endif // OPERATORS_H_
|
