2 * Copyright (c) 2017 ARM Limited.
4 * SPDX-License-Identifier: MIT
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:
13 * The above copyright notice and this permission notice shall be included in all
14 * copies or substantial portions of the Software.
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
26 #define SWAP_ROW(u0, l0) \
33 #define SWAP_4x4(u0, u1, u2, u3, l0, l1, l2, l3) \
35 VEC_DATA_TYPE(DATA_TYPE, 4) \
43 #define SWAP_8x8(u0, u1, u2, u3, u4, u5, u6, u7, l0, l1, l2, l3, l4, l5, l6, l7) \
45 VEC_DATA_TYPE(DATA_TYPE, 8) \
57 #define TRANSPOSE_4x4(u0, u1, u2, u3) \
59 VEC_DATA_TYPE(DATA_TYPE, 4) \
80 #define TRANSPOSE_8x8(u0, u1, u2, u3, u4, u5, u6, u7) \
82 TRANSPOSE_4x4(u0.s0123, u1.s0123, u2.s0123, u3.s0123); \
83 TRANSPOSE_4x4(u0.s4567, u1.s4567, u2.s4567, u3.s4567); \
84 TRANSPOSE_4x4(u4.s0123, u5.s0123, u6.s0123, u7.s0123); \
85 TRANSPOSE_4x4(u4.s4567, u5.s4567, u6.s4567, u7.s4567); \
86 SWAP_4x4(u0.s4567, u1.s4567, u2.s4567, u3.s4567, u4.s0123, u5.s0123, u6.s0123, u7.s0123); \
89 #define TRANSPOSE_16x16(u0, u1, u2, u3, u4, u5, u6, u7, u8, u9, u10, u11, u12, u13, u14, u15) \
91 TRANSPOSE_8x8(u0.s01234567, u1.s01234567, u2.s01234567, u3.s01234567, u4.s01234567, u5.s01234567, u6.s01234567, u7.s01234567); \
92 TRANSPOSE_8x8(u0.s89ABCDEF, u1.s89ABCDEF, u2.s89ABCDEF, u3.s89ABCDEF, u4.s89ABCDEF, u5.s89ABCDEF, u6.s89ABCDEF, u7.s89ABCDEF); \
93 TRANSPOSE_8x8(u8.s01234567, u9.s01234567, u10.s01234567, u11.s01234567, u12.s01234567, u13.s01234567, u14.s01234567, u15.s01234567); \
94 TRANSPOSE_8x8(u8.s89ABCDEF, u9.s89ABCDEF, u10.s89ABCDEF, u11.s89ABCDEF, u12.s89ABCDEF, u13.s89ABCDEF, u14.s89ABCDEF, u15.s89ABCDEF); \
95 SWAP_8x8(u0.s89ABCDEF, u1.s89ABCDEF, u2.s89ABCDEF, u3.s89ABCDEF, u4.s89ABCDEF, u5.s89ABCDEF, u6.s89ABCDEF, u7.s89ABCDEF, \
96 u8.s01234567, u9.s01234567, u10.s01234567, u11.s01234567, u12.s01234567, u13.s01234567, u14.s01234567, u15.s01234567); \
99 #ifndef DATA_TYPE_IN_BYTES
100 #error DATA_TYPE_IN_BYTES not set for the transpose OpenCL kernel
101 #endif /* not DATA_TYPE_IN_BYTES */
106 #if DATA_TYPE_IN_BYTES == 4
107 #define DATA_TYPE uint
108 #define TRANSPOSE() TRANSPOSE_4x4(u0, u1, u2, u3)
109 #define VLOAD(x, y) vload4(x, y)
110 #define VSTORE(x, y, z) vstore4(x, y, z)
112 #elif DATA_TYPE_IN_BYTES == 2
113 #define DATA_TYPE ushort
114 #define TRANSPOSE() TRANSPOSE_8x8(u0, u1, u2, u3, u4, u5, u6, u7)
115 #define VLOAD(x, y) vload8(x, y)
116 #define VSTORE(x, y, z) vstore8(x, y, z)
118 #elif DATA_TYPE_IN_BYTES == 1
119 #define DATA_TYPE uchar
120 #define TRANSPOSE() TRANSPOSE_16x16(u0, u1, u2, u3, u4, u5, u6, u7, u8, u9, u10, u11, u12, u13, u14, u15)
121 #define VLOAD(x, y) vload16(x, y)
122 #define VSTORE(x, y, z) vstore16(x, y, z)
123 #define BLOCK_SIZE 16
124 #else /* switch DATA_TYPE_IN_BYTES */
125 #error DATA_TYPE_IN_BYTES not supported for transpose
126 #endif /* switch DATA_TYPE_IN_BYTES */
128 /** This OpenCL kernel computes the matrix transposition of input matrix
130 * @attention The number of bytes of the data type need to be passed at compile time using -DDATA_TYPE_IN_BYTES. DATA_TYPE_IN_BYTES can be:
131 * -# -DDATA_TYPE_IN_BYTES=1 for transposing U8 or S8 matrices
132 * -# -DDATA_TYPE_IN_BYTES=2 for transposing U16, S16 or FP16 matrices
133 * -# -DDATA_TYPE_IN_BYTES=4 for transposing U32, S32 or FP32 matrices
135 * @param[in] src_ptr Pointer to the source matrix. Supported data types: U8/S8/U16/S16/F16/U32/S32/F32
136 * @param[in] src_stride_x Stride of the source matrix in X dimension (in bytes)
137 * @param[in] src_step_x src_stride_x * number of elements along X processed per workitem(in bytes)
138 * @param[in] src_stride_y Stride of the source matrix in Y dimension (in bytes)
139 * @param[in] src_step_y src_stride_y * number of elements along Y processed per workitem(in bytes)
140 * @param[in] src_offset_first_element_in_bytes The offset of the first element in the source matrix
141 * @param[out] dst_ptr Pointer to the destination matrix Supported data type: same as src_ptr
142 * @param[in] dst_stride_x Stride of the destination matrix in X dimension (in bytes)
143 * @param[in] dst_step_x dst_gx_stride_x * number of elements along X processed per workitem(in bytes)
144 * @param[in] dst_stride_y Stride of the destination matrix in Y dimension (in bytes)
145 * @param[in] dst_step_y dst_gx_stride_y * number of elements along Y processed per workitem(in bytes)
146 * @param[in] dst_offset_first_element_in_bytes The offset of the first element in the destination matrix
148 __kernel void transpose(IMAGE_DECLARATION(src),
149 IMAGE_DECLARATION(dst))
151 uint x = get_global_id(0) * BLOCK_SIZE;
152 uint y = get_global_id(1) * BLOCK_SIZE;
154 // Compute source address
155 Image src = CONVERT_TO_IMAGE_STRUCT(src);
157 // Load the NxN block at (x, y)
158 VEC_DATA_TYPE(DATA_TYPE, BLOCK_SIZE)
159 u0 = VLOAD(0, (__global DATA_TYPE *)(offset(&src, 0, 0)));
160 VEC_DATA_TYPE(DATA_TYPE, BLOCK_SIZE)
161 u1 = VLOAD(0, (__global DATA_TYPE *)(offset(&src, 0, 1)));
162 VEC_DATA_TYPE(DATA_TYPE, BLOCK_SIZE)
163 u2 = VLOAD(0, (__global DATA_TYPE *)(offset(&src, 0, 2)));
164 VEC_DATA_TYPE(DATA_TYPE, BLOCK_SIZE)
165 u3 = VLOAD(0, (__global DATA_TYPE *)(offset(&src, 0, 3)));
167 VEC_DATA_TYPE(DATA_TYPE, BLOCK_SIZE)
168 u4 = VLOAD(0, (__global DATA_TYPE *)(offset(&src, 0, 4)));
169 VEC_DATA_TYPE(DATA_TYPE, BLOCK_SIZE)
170 u5 = VLOAD(0, (__global DATA_TYPE *)(offset(&src, 0, 5)));
171 VEC_DATA_TYPE(DATA_TYPE, BLOCK_SIZE)
172 u6 = VLOAD(0, (__global DATA_TYPE *)(offset(&src, 0, 6)));
173 VEC_DATA_TYPE(DATA_TYPE, BLOCK_SIZE)
174 u7 = VLOAD(0, (__global DATA_TYPE *)(offset(&src, 0, 7)));
176 VEC_DATA_TYPE(DATA_TYPE, BLOCK_SIZE)
177 u8 = VLOAD(0, (__global DATA_TYPE *)(offset(&src, 0, 8)));
178 VEC_DATA_TYPE(DATA_TYPE, BLOCK_SIZE)
179 u9 = VLOAD(0, (__global DATA_TYPE *)(offset(&src, 0, 9)));
180 VEC_DATA_TYPE(DATA_TYPE, BLOCK_SIZE)
181 u10 = VLOAD(0, (__global DATA_TYPE *)(offset(&src, 0, 10)));
182 VEC_DATA_TYPE(DATA_TYPE, BLOCK_SIZE)
183 u11 = VLOAD(0, (__global DATA_TYPE *)(offset(&src, 0, 11)));
184 VEC_DATA_TYPE(DATA_TYPE, BLOCK_SIZE)
185 u12 = VLOAD(0, (__global DATA_TYPE *)(offset(&src, 0, 12)));
186 VEC_DATA_TYPE(DATA_TYPE, BLOCK_SIZE)
187 u13 = VLOAD(0, (__global DATA_TYPE *)(offset(&src, 0, 13)));
188 VEC_DATA_TYPE(DATA_TYPE, BLOCK_SIZE)
189 u14 = VLOAD(0, (__global DATA_TYPE *)(offset(&src, 0, 14)));
190 VEC_DATA_TYPE(DATA_TYPE, BLOCK_SIZE)
191 u15 = VLOAD(0, (__global DATA_TYPE *)(offset(&src, 0, 15)));
192 #endif /* BLOCK_SIZE == 16 */
193 #endif /* BLOCK_SIZE > 4 */
195 // Transpose the block
198 // Store the block at (y, x)
199 uint dst_offset_in_bytes = y * DATA_TYPE_IN_BYTES + x * dst_stride_y + dst_offset_first_element_in_bytes;
200 VSTORE(u0, 0, (__global DATA_TYPE *)(dst_ptr + dst_offset_in_bytes + 0 * dst_stride_y));
201 VSTORE(u1, 0, (__global DATA_TYPE *)(dst_ptr + dst_offset_in_bytes + 1 * dst_stride_y));
202 VSTORE(u2, 0, (__global DATA_TYPE *)(dst_ptr + dst_offset_in_bytes + 2 * dst_stride_y));
203 VSTORE(u3, 0, (__global DATA_TYPE *)(dst_ptr + dst_offset_in_bytes + 3 * dst_stride_y));
205 VSTORE(u4, 0, (__global DATA_TYPE *)(dst_ptr + dst_offset_in_bytes + 4 * dst_stride_y));
206 VSTORE(u5, 0, (__global DATA_TYPE *)(dst_ptr + dst_offset_in_bytes + 5 * dst_stride_y));
207 VSTORE(u6, 0, (__global DATA_TYPE *)(dst_ptr + dst_offset_in_bytes + 6 * dst_stride_y));
208 VSTORE(u7, 0, (__global DATA_TYPE *)(dst_ptr + dst_offset_in_bytes + 7 * dst_stride_y));
210 VSTORE(u8, 0, (__global DATA_TYPE *)(dst_ptr + dst_offset_in_bytes + 8 * dst_stride_y));
211 VSTORE(u9, 0, (__global DATA_TYPE *)(dst_ptr + dst_offset_in_bytes + 9 * dst_stride_y));
212 VSTORE(u10, 0, (__global DATA_TYPE *)(dst_ptr + dst_offset_in_bytes + 10 * dst_stride_y));
213 VSTORE(u11, 0, (__global DATA_TYPE *)(dst_ptr + dst_offset_in_bytes + 11 * dst_stride_y));
214 VSTORE(u12, 0, (__global DATA_TYPE *)(dst_ptr + dst_offset_in_bytes + 12 * dst_stride_y));
215 VSTORE(u13, 0, (__global DATA_TYPE *)(dst_ptr + dst_offset_in_bytes + 13 * dst_stride_y));
216 VSTORE(u14, 0, (__global DATA_TYPE *)(dst_ptr + dst_offset_in_bytes + 14 * dst_stride_y));
217 VSTORE(u15, 0, (__global DATA_TYPE *)(dst_ptr + dst_offset_in_bytes + 15 * dst_stride_y));
218 #endif /* BLOCK_SIZE == 16 */
219 #endif /* BLOCK_SIZE > 4 */