Gitiles
Code Review Sign In
review.mlplatform.org / ml / ComputeLibrary / 17b0f8ba60ec9db4b96471f9406843bee6a43a4f / . / src / core / CL / cl_kernels / direct_convolution_1x1_3x3_5x5_quantized.cl
blob: 83da76785badccd9ebf59d675cd80eb477a0cc88 [file] [log] [blame]
Chunosovd621bca2017-11-03 17:33:15 +0700[diff] [blame]1/*
Giorgio Arenaa086a0a2018-02-16 12:42:16 +0000[diff] [blame]2 * Copyright (c) 2017-2018 ARM Limited.
Chunosovd621bca2017-11-03 17:33:15 +0700[diff] [blame]3 *
4 * SPDX-License-Identifier: MIT
5 *
6 * Permission is hereby granted, free of charge, to any person obtaining a copy
7 * of this software and associated documentation files (the "Software"), to
8 * deal in the Software without restriction, including without limitation the
9 * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
10 * sell copies of the Software, and to permit persons to whom the Software is
11 * furnished to do so, subject to the following conditions:
12 *
13 * The above copyright notice and this permission notice shall be included in all
14 * copies or substantial portions of the Software.
15 *
16 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
17 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
18 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
19 * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
20 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
21 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
22 * SOFTWARE.
23 */
24#include "helpers_asymm.h"
25
26#undef CONVERT_SAT
27
28#if defined(DATA_TYPE) && defined(STRIDE_X) && defined(WEIGHTS_DEPTH)
29
30#if KERNEL_SIZE == 5
31
32#if STRIDE_X == 1
33#define CONVOLUTION1x5(acc, src_row_ptr, weights_row_ptr) CONVOLUTION1x5_STRIDE1(acc, src_row_ptr, weights_row_ptr)
34#elif STRIDE_X == 2
35#define CONVOLUTION1x5(acc, src_row_ptr, weights_row_ptr) CONVOLUTION1x5_STRIDE2(acc, src_row_ptr, weights_row_ptr)
36#else /* STRIDE_X not equals 1 or 2 */
37#error "STRIDE_X larger than 2 is not supported"
38#endif /* STRIDE_X */
39
40#define CONVOLUTION1x5_STRIDE1(acc, src_row_ptr, weights_row_ptr) \
41 ({ \
42 int4 weights_values0 = convert_int4(vload4(0, weights_row_ptr)); \
43 int weights_value1 = convert_int(*(weights_row_ptr + 4)); \
44 int8 src0 = convert_int8(vload8(0, src_row_ptr)); \
45 int4 src1 = convert_int4(vload4(0, src_row_ptr + 8)); \
46 acc += (src0 + input_offset) * ((int8)weights_values0.s0 + weight_offset); \
47 acc += ((int8)(src0.s1234, src0.s567, src1.s0) + input_offset) * ((int8)weights_values0.s1 + weight_offset); \
48 acc += ((int8)(src0.s234, src0.s567, src1.s01) + input_offset) * ((int8)weights_values0.s2 + weight_offset); \
49 acc += ((int8)(src0.s345, src0.s67, src1.s012) + input_offset) * ((int8)weights_values0.s3 + weight_offset); \
50 acc += ((int8)(src0.s45, src0.s67, src1.s0123) + input_offset) * ((int8)weights_value1 + weight_offset); \
51 })
52
53#define CONVOLUTION1x5_STRIDE2(acc, src_row_ptr, weights_row_ptr) \
54 ({ \
55 int4 weights_values0 = convert_int4(vload4(0, weights_row_ptr)); \
56 int weights_value1 = convert_int(*(weights_row_ptr + 4)); \
57 int16 src0 = convert_int16(vload16(0, src_row_ptr)); \
58 int4 src1 = convert_int4(vload4(0, src_row_ptr + 16)); \
59 acc += (src0.even + input_offset) * ((int8)weights_values0.s0 + weight_offset); \
60 acc += ((int8)(src0.s1357, src0.s9BDF) + input_offset) * ((int8)weights_values0.s1 + weight_offset); \
61 acc += ((int8)(src0.s2468, src0.sACE, src1.s0) + input_offset) * ((int8)weights_values0.s2 + weight_offset); \
62 acc += ((int8)(src0.s3579, src0.sBDF, src1.s1) + input_offset) * ((int8)weights_values0.s3 + weight_offset); \
63 acc += ((int8)(src0.s468a, src0.sCE, src1.s02) + input_offset) * ((int8)weights_value1 + weight_offset); \
64 })
65
66#elif KERNEL_SIZE == 3
67
68#if STRIDE_X == 1
69#define CONVOLUTION1x3(acc, src_row_ptr, weights_row_ptr) CONVOLUTION1x3_STRIDE1(acc, src_row_ptr, weights_row_ptr)
70#elif STRIDE_X == 2
71#define CONVOLUTION1x3(acc, src_row_ptr, weights_row_ptr) CONVOLUTION1x3_STRIDE2(acc, src_row_ptr, weights_row_ptr)
72#else /* STRIDE_X not equals 1 or 2 */
73#error "STRIDE_X larger than 2 is not supported"
74#endif /* STRIDE_X */
75
76#define CONVOLUTION1x3_STRIDE1(acc, src_row_ptr, weights_row_ptr) \
77 ({ \
78 int3 weights_values0 = convert_int3(vload3(0, weights_row_ptr)); \
79 int8 src0 = convert_int8(vload8(0, src_row_ptr)); \
80 int2 src1 = convert_int2(vload2(0, src_row_ptr + 8)); \
81 acc += (src0 + input_offset) * ((int8)weights_values0.s0 + weight_offset); \
82 acc += ((int8)(src0.s1234, src0.s567, src1.s0) + input_offset) * ((int8)weights_values0.s1 + weight_offset); \
83 acc += ((int8)(src0.s234, src0.s567, src1.s01) + input_offset) * ((int8)weights_values0.s2 + weight_offset); \
84 })
85
86#define CONVOLUTION1x3_STRIDE2(acc, src_row_ptr, weights_row_ptr) \
87 ({ \
88 int3 weights_values0 = convert_int3(vload3(0, weights_row_ptr)); \
89 int16 src0 = convert_int16(vload16(0, src_row_ptr)); \
90 int src1 = convert_int(*(src_row_ptr + 16)); \
91 acc += (src0.even + input_offset) * ((int8)weights_values0.s0 + weight_offset); \
92 acc += ((int8)(src0.s1357, src0.s9BDF) + input_offset) * ((int8)weights_values0.s1 + weight_offset); \
93 acc += ((int8)(src0.s2468, src0.sACE, src1) + input_offset) * ((int8)weights_values0.s2 + weight_offset); \
94 })
95
96#elif KERNEL_SIZE == 1
97
98#if STRIDE_X == 3
99#define INPUT_PIXEL extract_input_stride3
100#elif STRIDE_X == 2
101#define INPUT_PIXEL extract_input_stride2
102#elif STRIDE_X == 1
103#define INPUT_PIXEL extract_input_stride1
104
105#else /* STRIDE_X not equals 1, 2 or 3 */
106#error "Only support strides 1, 2 and 3"
107#endif /* STRIDE_X */
108
109/** Extracts a 1D horizontal vector from the input tensor with stride as 1.
110 *
111 * @param[in] input_pixel Pointer to the first pixel.
112 *
113 * @return extracted input pixels.
114 */
115inline uchar8 extract_input_stride1(__global const uchar *input_pixel)
116{
117 return vload8(0, input_pixel);
118}
119
120/** Extracts a 1D horizontal vector from the input tensor with stride as 2.
121 *
122 * @param[in] input_pixel Pointer to the first pixel.
123 *
124 * @return extracted input pixels.
125 */
126inline uchar8 extract_input_stride2(__global const uchar *input_pixel)
127{
128 uchar16 temp = vload16(0, input_pixel);
129 return temp.s02468ace;
130}
131
132/** Extracts a 1D horizontal vector from the input tensor with stride as 3 and 8-bit data size.
133 *
134 * @param[in] input_pixel Pointer to the first pixel.
135 *
136 * @return extracted input pixels.
137 */
138inline uchar8 extract_input_stride3(__global const uchar *input_pixel)
139{
140 uchar16 temp1 = vload16(0, input_pixel);
141 uchar16 temp2 = vload16(0, input_pixel + 12);
142 return (uchar8)(temp1.s0369, temp2.s0369);
143}
144
145#else /* KERNEL_SIZE not equals 1, 3 or 5 */
146#error "Only kernel sizes 1, 3 and 5 are supported"
147#endif /* KERNEL_SIZE */
148
149/** This kernel performs a direct convolution to convolve the low three dimensions.
150 *
151 * @note The convolution stride x must be passed at compile time using -DSTRIDE_X e.g. -DSTRIDE_X=1
152 * @note The third dimensions of the weights tensors must be passed at compile time using -DWEIGHTS_DEPTH
153 * @note If biases are used then -DHAS_BIAS has to be passed at compile time
154 *
155 * @param[in] src_ptr Pointer to the source tensor. Supported data types: QASYMM8
156 * @param[in] src_stride_x Stride of the source tensor in X dimension (in bytes)
157 * @param[in] src_step_x src_stride_x * number of elements along X processed per workitem(in bytes)
158 * @param[in] src_stride_y Stride of the source tensor in Y dimension (in bytes)
159 * @param[in] src_step_y src_stride_y * number of elements along Y processed per workitem(in bytes)
160 * @param[in] src_stride_z Stride of the source tensor in Z dimension (in bytes)
161 * @param[in] src_step_z src_stride_z * number of elements along Z processed per workitem(in bytes)
162 * @param[in] src_offset_first_element_in_bytes The offset of the first element in the source tensor
163 * @param[out] dst_ptr Pointer to the destination tensor. Supported data types: same as @p src_ptr
164 * @param[in] dst_stride_x Stride of the destination tensor in X dimension (in bytes)
165 * @param[in] dst_step_x dst_stride_x * number of elements along X processed per workitem(in bytes)
166 * @param[in] dst_stride_y Stride of the destination tensor in Y dimension (in bytes)
167 * @param[in] dst_step_y dst_stride_y * number of elements along Z processed per workitem(in bytes)
168 * @param[in] dst_stride_z Stride of the destination tensor in Z dimension (in bytes)
169 * @param[in] dst_step_z dst_stride_z * number of elements along Z processed per workitem(in bytes)
170 * @param[in] dst_offset_first_element_in_bytes The offset of the first element in the destination tensor
Joel Liangf1f3ebd2017-11-10 09:59:19 +0800[diff] [blame]171 * @param[in] weights_ptr Pointer to the weights tensor. Supported data types: same as @p weights_ptr
Chunosovd621bca2017-11-03 17:33:15 +0700[diff] [blame]172 * @param[in] weights_stride_x Stride of the weights tensor in X dimension (in bytes)
173 * @param[in] weights_step_x weights_stride_x * number of elements along X processed per workitem(in bytes)
174 * @param[in] weights_stride_y Stride of the weights tensor in Y dimension (in bytes)
175 * @param[in] weights_step_y weights_stride_y * number of elements along y processed per workitem(in bytes)
176 * @param[in] weights_stride_z Stride of the weights tensor in Z dimension (in bytes)
177 * @param[in] weights_step_z weights_stride_z * number of elements along Z processed per workitem(in bytes)
178 * @param[in] weights_offset_first_element_in_bytes The offset of the first element in the weights tensor
Georgios Pinitas540d0082017-11-17 10:55:00 +0000[diff] [blame]179 * @param[in] biases_ptr Pointer to the biases tensor. Supported data types: S32
Chunosovd621bca2017-11-03 17:33:15 +0700[diff] [blame]180 * @param[in] biases_stride_x Stride of the biases tensor in X dimension (in bytes)
181 * @param[in] biases_step_x biases_stride_x * number of elements along X processed per workitem(in bytes)
182 * @param[in] biases_offset_first_element_in_bytes The offset of the first element in the biases tensor
183 * @param[in] weights_stride_w Stride of the weights tensor in the 4th dimension
184 * @param[in] input_offset Input offset quantization parameter
185 * @param[in] weight_offset Weights offset quantization parameter
186 * @param[in] output_offset Output offset quantization parameter
187 * @param[in] output_multiplier Output integer multiplier quantization parameter
188 * @param[in] output_shift Output integer shift quantization parameter
189 */
190__kernel void direct_convolution_1x1_3x3_5x5_quantized(
191 TENSOR3D_DECLARATION(src),
192 TENSOR3D_DECLARATION(dst),
193 TENSOR3D_DECLARATION(weights),
194#ifdef HAS_BIAS
195 VECTOR_DECLARATION(biases),
196#endif /* defined(HAS_BIAS) */
197 unsigned int weights_stride_w,
198 int input_offset,
199 int weight_offset,
200 int output_offset,
201 int output_multiplier,
202 int output_shift)
203{
204 Image src = CONVERT_TO_IMAGE_STRUCT(src);
205 Tensor3D weights = CONVERT_TO_TENSOR3D_STRUCT_NO_STEP(weights);
206 Tensor3D dst = CONVERT_TO_TENSOR3D_STRUCT(dst);
207
208 int8 pixels0 = 0;
209
210 __global uchar *weights_addr = (__global uchar *)tensor3D_offset(&weights, 0, 0, 0);
211 __global uchar *src_addr = (__global uchar *)offset(&src, 0, 0);
212
213 const int kernel_index = get_global_id(2);
214 weights_addr += kernel_index * weights_stride_w;
215
216 for(volatile int d = 0; d < WEIGHTS_DEPTH; ++d)
217 {
218#if KERNEL_SIZE == 5
219 CONVOLUTION1x5(pixels0, (__global uchar *)src_addr, (__global uchar *)weights_addr);
220 CONVOLUTION1x5(pixels0, (__global uchar *)(src_addr + 1 * src_stride_y), (__global uchar *)(weights_addr + 1 * weights_stride_y));
221 CONVOLUTION1x5(pixels0, (__global uchar *)(src_addr + 2 * src_stride_y), (__global uchar *)(weights_addr + 2 * weights_stride_y));
222 CONVOLUTION1x5(pixels0, (__global uchar *)(src_addr + 3 * src_stride_y), (__global uchar *)(weights_addr + 3 * weights_stride_y));
223 CONVOLUTION1x5(pixels0, (__global uchar *)(src_addr + 4 * src_stride_y), (__global uchar *)(weights_addr + 4 * weights_stride_y));
224#elif KERNEL_SIZE == 3
225 CONVOLUTION1x3(pixels0, (__global uchar *)(src_addr + 0 * src_stride_y), (__global uchar *)(weights_addr + 0 * weights_stride_y));
226 CONVOLUTION1x3(pixels0, (__global uchar *)(src_addr + 1 * src_stride_y), (__global uchar *)(weights_addr + 1 * weights_stride_y));
227 CONVOLUTION1x3(pixels0, (__global uchar *)(src_addr + 2 * src_stride_y), (__global uchar *)(weights_addr + 2 * weights_stride_y));
228#elif KERNEL_SIZE == 1
229 int weight = convert_int(*(__global uchar *)weights_addr);
230 int8 input_pixel = convert_int8(INPUT_PIXEL((__global uchar *)src_addr));
231 pixels0 += (input_pixel + input_offset) * ((int8)weight + weight_offset);
232#endif /* (KERNEL_SIZE == 1) || (KERNEL_SIZE == 3) || (KERNEL_SIZE == 5) */
233
234 src_addr += src_stride_z;
235 weights_addr += weights_stride_z;
236 }
237
238#ifdef HAS_BIAS
Georgios Pinitas540d0082017-11-17 10:55:00 +0000[diff] [blame]239 Vector biases = CONVERT_TO_VECTOR_STRUCT_NO_STEP(biases);
240 __global int *bias_addr = ((__global int *)(vector_offset(&biases, kernel_index)));
241 pixels0 += (int8)(*bias_addr);
Chunosovd621bca2017-11-03 17:33:15 +0700[diff] [blame]242#endif /* defined(HAS_BIAS) */
243
244 pixels0 = ASYMM_MULT_BY_QUANT_MULTIPLIER_LESS_THAN_ONE(pixels0, output_multiplier, output_shift, 8);
245 pixels0 = pixels0 + output_offset;
Chunosovd621bca2017-11-03 17:33:15 +0700[diff] [blame]246
Georgios Pinitas6fdfaa82017-11-29 14:27:24 +0000[diff] [blame]247 vstore8(convert_uchar8_sat(pixels0), 0, (__global uchar *)dst.ptr);
Chunosovd621bca2017-11-03 17:33:15 +0700[diff] [blame]248}
249#endif // defined(DATA_TYPE) && defined(STRIDE_X) && defined(WEIGHTS_DEPTH)
Giorgio Arenaa086a0a2018-02-16 12:42:16 +0000[diff] [blame]250
Georgios Pinitas5b52fe32018-07-12 12:42:35 +0100[diff] [blame]251#if defined(VEC_SIZE)
252
253#define VEC_INT VEC_DATA_TYPE(int, VEC_SIZE)
254#define CONVERT_SAT_UCHAR_STR(x, size) (convert_uchar##size##_sat((x)))
255#define CONVERT_SAT_UCHAR(x, size) CONVERT_SAT_UCHAR_STR(x, size)
256
Giorgio Arenaa086a0a2018-02-16 12:42:16 +0000[diff] [blame]257/** This function computes the output stage of a depthwise convolution.
258 *
259 * @param[in] src_ptr Pointer to the source image. Supported data types: QASYMM8
260 * @param[in] src_stride_x Stride of the source image in X dimension (in bytes)
261 * @param[in] src_step_x src_stride_x * number of elements along X processed per workitem(in bytes)
262 * @param[in] src_stride_y Stride of the source image in Y dimension (in bytes)
263 * @param[in] src_step_y src_stride_y * number of elements along Y processed per workitem(in bytes)
264 * @param[in] src_offset_first_element_in_bytes The offset of the first element in the source image
265 * @param[in] src_stride_z Stride of the source tensor in Z dimension (in bytes)
266 * @param[in] src_step_z src_stride_z * number of elements along Y processed per workitem(in bytes)
267 * @param[in] dst_ptr Pointer to the destination tensor. Supported data types: QASYMM8
268 * @param[in] dst_stride_x Stride of the destination tensor in X dimension (in bytes)
269 * @param[in] dst_step_x dst_stride_x * number of elements along X processed per workitem(in bytes)
270 * @param[in] dst_stride_y Stride of the destination tensor in Y dimension (in bytes)
271 * @param[in] dst_step_y dst_stride_y * number of elements along Y processed per workitem(in bytes)
272 * @param[in] dst_stride_z Stride of the destination tensor in Z dimension (in bytes)
273 * @param[in] dst_step_z dst_stride_z * number of elements along Y processed per workitem(in bytes)
274 * @param[in] dst_offset_first_element_in_bytes The offset of the first element in the destination tensor
275 * @param[in] bias_ptr (Optional) Pointer to the biases vector. Supported data types: S32
276 * @param[in] bias_stride_x (Optional) Stride of the biases vector in X dimension (in bytes)
277 * @param[in] bias_step_x (Optional) bias_stride_x * number of elements along X processed per workitem(in bytes)
278 * @param[in] bias_offset_first_element_in_bytes (Optional) The offset of the first element in the biases vector
279 * @param[in] output_offset Quantized offset of zero point of the output tensor data range
280 * @param[in] output_multiplier Output scale multiplier
281 * @param[in] output_shift Output scale divisor exponent
282 */
Giorgio Arenaa086a0a2018-02-16 12:42:16 +0000[diff] [blame]283__kernel void output_stage_quantized(
284 TENSOR3D_DECLARATION(src),
285 TENSOR3D_DECLARATION(dst),
286#if defined(HAS_BIAS)
287 VECTOR_DECLARATION(bias),
288#endif //defined(HAS_BIAS)
289 int output_offset,
290 int output_multiplier,
291 int output_shift)
292{
293 Image src = CONVERT_TENSOR3D_TO_IMAGE_STRUCT(src);
294 Image dst = CONVERT_TENSOR3D_TO_IMAGE_STRUCT(dst);
295#if defined(HAS_BIAS)
296 Vector bias = CONVERT_TO_VECTOR_STRUCT_NO_STEP(bias);
297#endif //defined(HAS_BIAS)
298
299 // Load input
Georgios Pinitas5b52fe32018-07-12 12:42:35 +0100[diff] [blame]300 VEC_INT vals = VLOAD(VEC_SIZE)(0, (__global int *)(src.ptr));
Giorgio Arenaa086a0a2018-02-16 12:42:16 +0000[diff] [blame]301
302#if defined(HAS_BIAS)
303 // Load and add bias
Giorgio Arenad051e972018-06-20 11:46:42 +0100[diff] [blame]304#if defined(NCHW)
Giorgio Arenaa086a0a2018-02-16 12:42:16 +0000[diff] [blame]305 int bias_value = *((__global int *)(vector_offset(&bias, get_global_id(2))));
Giorgio Arenad051e972018-06-20 11:46:42 +0100[diff] [blame]306#else // defined(NCHW)
Georgios Pinitas5b52fe32018-07-12 12:42:35 +0100[diff] [blame]307 VEC_INT bias_value = VLOAD(VEC_SIZE)(0, ((__global int *)(vector_offset(&bias, get_global_id(0) * VEC_SIZE))));
Giorgio Arenad051e972018-06-20 11:46:42 +0100[diff] [blame]308#endif // defined(NCHW)
309
Georgios Pinitas5b52fe32018-07-12 12:42:35 +0100[diff] [blame]310 vals += (VEC_INT)(bias_value);
Giorgio Arenaa086a0a2018-02-16 12:42:16 +0000[diff] [blame]311#endif //defined(HAS_BIAS)
312
Georgios Pinitas5b52fe32018-07-12 12:42:35 +0100[diff] [blame]313 vals = ASYMM_MULT_BY_QUANT_MULTIPLIER_LESS_THAN_ONE(vals, output_multiplier, output_shift, VEC_SIZE);
Giorgio Arenaa086a0a2018-02-16 12:42:16 +0000[diff] [blame]314 vals = vals + output_offset;
Giorgio Arenaa086a0a2018-02-16 12:42:16 +0000[diff] [blame]315
316 // Store result in dst
Georgios Pinitas5b52fe32018-07-12 12:42:35 +0100[diff] [blame]317 VSTORE(VEC_SIZE)
318 (CONVERT_SAT_UCHAR(vals, VEC_SIZE), 0, (__global uchar *)dst.ptr);
Giorgio Arenaa086a0a2018-02-16 12:42:16 +0000[diff] [blame]319}
Georgios Pinitas5b52fe32018-07-12 12:42:35 +0100[diff] [blame]320
321#undef VEC_INT
322#undef CONVERT_SAT_UCHAR_STR
323#undef CONVERT_SAT_UCHAR
324
325#endif // defined(VEC_SIZE)