1 /* 2 * Copyright 2012 The LibYuv Project Authors. All rights reserved. 3 * 4 * Use of this source code is governed by a BSD-style license 5 * that can be found in the LICENSE file in the root of the source 6 * tree. An additional intellectual property rights grant can be found 7 * in the file PATENTS. All contributing project authors may 8 * be found in the AUTHORS file in the root of the source tree. 9 */ 10 11 #include "libyuv/row.h" 12 13 #include <string.h> // For memset. 14 15 #include "libyuv/basic_types.h" 16 17 #ifdef __cplusplus 18 namespace libyuv { 19 extern "C" { 20 #endif 21 22 // memset for temp is meant to clear the source buffer (not dest) so that 23 // SIMD that reads full multiple of 16 bytes will not trigger msan errors. 24 // memset is not needed for production, as the garbage values are processed but 25 // not used, although there may be edge cases for subsampling. 26 // The size of the buffer is based on the largest read, which can be inferred 27 // by the source type (e.g. ARGB) and the mask (last parameter), or by examining 28 // the source code for how much the source pointers are advanced. 29 30 // Subsampled source needs to be increase by 1 of not even. 31 #define SS(width, shift) (((width) + (1 << (shift)) - 1) >> (shift)) 32 33 // Any 4 planes to 1 with yuvconstants 34 #define ANY41C(NAMEANY, ANY_SIMD, UVSHIFT, DUVSHIFT, BPP, MASK) \ 35 void NAMEANY(const uint8_t* y_buf, const uint8_t* u_buf, \ 36 const uint8_t* v_buf, const uint8_t* a_buf, uint8_t* dst_ptr, \ 37 const struct YuvConstants* yuvconstants, int width) { \ 38 SIMD_ALIGNED(uint8_t temp[64 * 5]); \ 39 memset(temp, 0, 64 * 4); /* for msan */ \ 40 int r = width & MASK; \ 41 int n = width & ~MASK; \ 42 if (n > 0) { \ 43 ANY_SIMD(y_buf, u_buf, v_buf, a_buf, dst_ptr, yuvconstants, n); \ 44 } \ 45 memcpy(temp, y_buf + n, r); \ 46 memcpy(temp + 64, u_buf + (n >> UVSHIFT), SS(r, UVSHIFT)); \ 47 memcpy(temp + 128, v_buf + (n >> UVSHIFT), SS(r, UVSHIFT)); \ 48 memcpy(temp + 192, a_buf + n, r); \ 49 ANY_SIMD(temp, temp + 64, temp + 128, temp + 192, temp + 256, \ 50 yuvconstants, MASK + 1); \ 51 memcpy(dst_ptr + (n >> DUVSHIFT) * BPP, temp + 256, \ 52 SS(r, DUVSHIFT) * BPP); \ 53 } 54 55 #ifdef HAS_I422ALPHATOARGBROW_SSSE3 56 ANY41C(I422AlphaToARGBRow_Any_SSSE3, I422AlphaToARGBRow_SSSE3, 1, 0, 4, 7) 57 #endif 58 #ifdef HAS_I422ALPHATOARGBROW_AVX2 59 ANY41C(I422AlphaToARGBRow_Any_AVX2, I422AlphaToARGBRow_AVX2, 1, 0, 4, 15) 60 #endif 61 #ifdef HAS_I422ALPHATOARGBROW_NEON 62 ANY41C(I422AlphaToARGBRow_Any_NEON, I422AlphaToARGBRow_NEON, 1, 0, 4, 7) 63 #endif 64 #ifdef HAS_I422ALPHATOARGBROW_MSA 65 ANY41C(I422AlphaToARGBRow_Any_MSA, I422AlphaToARGBRow_MSA, 1, 0, 4, 7) 66 #endif 67 #undef ANY41C 68 69 // Any 3 planes to 1. 70 #define ANY31(NAMEANY, ANY_SIMD, UVSHIFT, DUVSHIFT, BPP, MASK) \ 71 void NAMEANY(const uint8_t* y_buf, const uint8_t* u_buf, \ 72 const uint8_t* v_buf, uint8_t* dst_ptr, int width) { \ 73 SIMD_ALIGNED(uint8_t temp[64 * 4]); \ 74 memset(temp, 0, 64 * 3); /* for YUY2 and msan */ \ 75 int r = width & MASK; \ 76 int n = width & ~MASK; \ 77 if (n > 0) { \ 78 ANY_SIMD(y_buf, u_buf, v_buf, dst_ptr, n); \ 79 } \ 80 memcpy(temp, y_buf + n, r); \ 81 memcpy(temp + 64, u_buf + (n >> UVSHIFT), SS(r, UVSHIFT)); \ 82 memcpy(temp + 128, v_buf + (n >> UVSHIFT), SS(r, UVSHIFT)); \ 83 ANY_SIMD(temp, temp + 64, temp + 128, temp + 192, MASK + 1); \ 84 memcpy(dst_ptr + (n >> DUVSHIFT) * BPP, temp + 192, \ 85 SS(r, DUVSHIFT) * BPP); \ 86 } 87 88 // Merge functions. 89 #ifdef HAS_MERGERGBROW_SSSE3 90 ANY31(MergeRGBRow_Any_SSSE3, MergeRGBRow_SSSE3, 0, 0, 3, 15) 91 #endif 92 #ifdef HAS_MERGERGBROW_NEON 93 ANY31(MergeRGBRow_Any_NEON, MergeRGBRow_NEON, 0, 0, 3, 15) 94 #endif 95 #ifdef HAS_I422TOYUY2ROW_SSE2 96 ANY31(I422ToYUY2Row_Any_SSE2, I422ToYUY2Row_SSE2, 1, 1, 4, 15) 97 ANY31(I422ToUYVYRow_Any_SSE2, I422ToUYVYRow_SSE2, 1, 1, 4, 15) 98 #endif 99 #ifdef HAS_I422TOYUY2ROW_AVX2 100 ANY31(I422ToYUY2Row_Any_AVX2, I422ToYUY2Row_AVX2, 1, 1, 4, 31) 101 ANY31(I422ToUYVYRow_Any_AVX2, I422ToUYVYRow_AVX2, 1, 1, 4, 31) 102 #endif 103 #ifdef HAS_I422TOYUY2ROW_NEON 104 ANY31(I422ToYUY2Row_Any_NEON, I422ToYUY2Row_NEON, 1, 1, 4, 15) 105 #endif 106 #ifdef HAS_I422TOYUY2ROW_MSA 107 ANY31(I422ToYUY2Row_Any_MSA, I422ToYUY2Row_MSA, 1, 1, 4, 31) 108 #endif 109 #ifdef HAS_I422TOUYVYROW_NEON 110 ANY31(I422ToUYVYRow_Any_NEON, I422ToUYVYRow_NEON, 1, 1, 4, 15) 111 #endif 112 #ifdef HAS_I422TOUYVYROW_MSA 113 ANY31(I422ToUYVYRow_Any_MSA, I422ToUYVYRow_MSA, 1, 1, 4, 31) 114 #endif 115 #ifdef HAS_BLENDPLANEROW_AVX2 116 ANY31(BlendPlaneRow_Any_AVX2, BlendPlaneRow_AVX2, 0, 0, 1, 31) 117 #endif 118 #ifdef HAS_BLENDPLANEROW_SSSE3 119 ANY31(BlendPlaneRow_Any_SSSE3, BlendPlaneRow_SSSE3, 0, 0, 1, 7) 120 #endif 121 #undef ANY31 122 123 // Note that odd width replication includes 444 due to implementation 124 // on arm that subsamples 444 to 422 internally. 125 // Any 3 planes to 1 with yuvconstants 126 #define ANY31C(NAMEANY, ANY_SIMD, UVSHIFT, DUVSHIFT, BPP, MASK) \ 127 void NAMEANY(const uint8_t* y_buf, const uint8_t* u_buf, \ 128 const uint8_t* v_buf, uint8_t* dst_ptr, \ 129 const struct YuvConstants* yuvconstants, int width) { \ 130 SIMD_ALIGNED(uint8_t temp[64 * 4]); \ 131 memset(temp, 0, 64 * 3); /* for YUY2 and msan */ \ 132 int r = width & MASK; \ 133 int n = width & ~MASK; \ 134 if (n > 0) { \ 135 ANY_SIMD(y_buf, u_buf, v_buf, dst_ptr, yuvconstants, n); \ 136 } \ 137 memcpy(temp, y_buf + n, r); \ 138 memcpy(temp + 64, u_buf + (n >> UVSHIFT), SS(r, UVSHIFT)); \ 139 memcpy(temp + 128, v_buf + (n >> UVSHIFT), SS(r, UVSHIFT)); \ 140 if (width & 1) { \ 141 temp[64 + SS(r, UVSHIFT)] = temp[64 + SS(r, UVSHIFT) - 1]; \ 142 temp[128 + SS(r, UVSHIFT)] = temp[128 + SS(r, UVSHIFT) - 1]; \ 143 } \ 144 ANY_SIMD(temp, temp + 64, temp + 128, temp + 192, yuvconstants, MASK + 1); \ 145 memcpy(dst_ptr + (n >> DUVSHIFT) * BPP, temp + 192, \ 146 SS(r, DUVSHIFT) * BPP); \ 147 } 148 149 #ifdef HAS_I422TOARGBROW_SSSE3 150 ANY31C(I422ToARGBRow_Any_SSSE3, I422ToARGBRow_SSSE3, 1, 0, 4, 7) 151 #endif 152 #ifdef HAS_I422TOAR30ROW_SSSE3 153 ANY31C(I422ToAR30Row_Any_SSSE3, I422ToAR30Row_SSSE3, 1, 0, 4, 7) 154 #endif 155 #ifdef HAS_I422TOAR30ROW_AVX2 156 ANY31C(I422ToAR30Row_Any_AVX2, I422ToAR30Row_AVX2, 1, 0, 4, 15) 157 #endif 158 #ifdef HAS_I444TOARGBROW_SSSE3 159 ANY31C(I444ToARGBRow_Any_SSSE3, I444ToARGBRow_SSSE3, 0, 0, 4, 7) 160 ANY31C(I422ToRGBARow_Any_SSSE3, I422ToRGBARow_SSSE3, 1, 0, 4, 7) 161 ANY31C(I422ToARGB4444Row_Any_SSSE3, I422ToARGB4444Row_SSSE3, 1, 0, 2, 7) 162 ANY31C(I422ToARGB1555Row_Any_SSSE3, I422ToARGB1555Row_SSSE3, 1, 0, 2, 7) 163 ANY31C(I422ToRGB565Row_Any_SSSE3, I422ToRGB565Row_SSSE3, 1, 0, 2, 7) 164 ANY31C(I422ToRGB24Row_Any_SSSE3, I422ToRGB24Row_SSSE3, 1, 0, 3, 7) 165 #endif // HAS_I444TOARGBROW_SSSE3 166 #ifdef HAS_I422TORGB24ROW_AVX2 167 ANY31C(I422ToRGB24Row_Any_AVX2, I422ToRGB24Row_AVX2, 1, 0, 3, 15) 168 #endif 169 #ifdef HAS_I422TOARGBROW_AVX2 170 ANY31C(I422ToARGBRow_Any_AVX2, I422ToARGBRow_AVX2, 1, 0, 4, 15) 171 #endif 172 #ifdef HAS_I422TORGBAROW_AVX2 173 ANY31C(I422ToRGBARow_Any_AVX2, I422ToRGBARow_AVX2, 1, 0, 4, 15) 174 #endif 175 #ifdef HAS_I444TOARGBROW_AVX2 176 ANY31C(I444ToARGBRow_Any_AVX2, I444ToARGBRow_AVX2, 0, 0, 4, 15) 177 #endif 178 #ifdef HAS_I422TOARGB4444ROW_AVX2 179 ANY31C(I422ToARGB4444Row_Any_AVX2, I422ToARGB4444Row_AVX2, 1, 0, 2, 15) 180 #endif 181 #ifdef HAS_I422TOARGB1555ROW_AVX2 182 ANY31C(I422ToARGB1555Row_Any_AVX2, I422ToARGB1555Row_AVX2, 1, 0, 2, 15) 183 #endif 184 #ifdef HAS_I422TORGB565ROW_AVX2 185 ANY31C(I422ToRGB565Row_Any_AVX2, I422ToRGB565Row_AVX2, 1, 0, 2, 15) 186 #endif 187 #ifdef HAS_I422TOARGBROW_NEON 188 ANY31C(I444ToARGBRow_Any_NEON, I444ToARGBRow_NEON, 0, 0, 4, 7) 189 ANY31C(I422ToARGBRow_Any_NEON, I422ToARGBRow_NEON, 1, 0, 4, 7) 190 ANY31C(I422ToRGBARow_Any_NEON, I422ToRGBARow_NEON, 1, 0, 4, 7) 191 ANY31C(I422ToRGB24Row_Any_NEON, I422ToRGB24Row_NEON, 1, 0, 3, 7) 192 ANY31C(I422ToARGB4444Row_Any_NEON, I422ToARGB4444Row_NEON, 1, 0, 2, 7) 193 ANY31C(I422ToARGB1555Row_Any_NEON, I422ToARGB1555Row_NEON, 1, 0, 2, 7) 194 ANY31C(I422ToRGB565Row_Any_NEON, I422ToRGB565Row_NEON, 1, 0, 2, 7) 195 #endif 196 #ifdef HAS_I422TOARGBROW_MSA 197 ANY31C(I444ToARGBRow_Any_MSA, I444ToARGBRow_MSA, 0, 0, 4, 7) 198 ANY31C(I422ToARGBRow_Any_MSA, I422ToARGBRow_MSA, 1, 0, 4, 7) 199 ANY31C(I422ToRGBARow_Any_MSA, I422ToRGBARow_MSA, 1, 0, 4, 7) 200 ANY31C(I422ToRGB24Row_Any_MSA, I422ToRGB24Row_MSA, 1, 0, 3, 15) 201 ANY31C(I422ToARGB4444Row_Any_MSA, I422ToARGB4444Row_MSA, 1, 0, 2, 7) 202 ANY31C(I422ToARGB1555Row_Any_MSA, I422ToARGB1555Row_MSA, 1, 0, 2, 7) 203 ANY31C(I422ToRGB565Row_Any_MSA, I422ToRGB565Row_MSA, 1, 0, 2, 7) 204 #endif 205 #undef ANY31C 206 207 // Any 3 planes of 16 bit to 1 with yuvconstants 208 // TODO(fbarchard): consider sharing this code with ANY31C 209 #define ANY31CT(NAMEANY, ANY_SIMD, UVSHIFT, DUVSHIFT, T, SBPP, BPP, MASK) \ 210 void NAMEANY(const T* y_buf, const T* u_buf, const T* v_buf, \ 211 uint8_t* dst_ptr, const struct YuvConstants* yuvconstants, \ 212 int width) { \ 213 SIMD_ALIGNED(T temp[16 * 3]); \ 214 SIMD_ALIGNED(uint8_t out[64]); \ 215 memset(temp, 0, 16 * 3 * SBPP); /* for YUY2 and msan */ \ 216 int r = width & MASK; \ 217 int n = width & ~MASK; \ 218 if (n > 0) { \ 219 ANY_SIMD(y_buf, u_buf, v_buf, dst_ptr, yuvconstants, n); \ 220 } \ 221 memcpy(temp, y_buf + n, r * SBPP); \ 222 memcpy(temp + 16, u_buf + (n >> UVSHIFT), SS(r, UVSHIFT) * SBPP); \ 223 memcpy(temp + 32, v_buf + (n >> UVSHIFT), SS(r, UVSHIFT) * SBPP); \ 224 ANY_SIMD(temp, temp + 16, temp + 32, out, yuvconstants, MASK + 1); \ 225 memcpy(dst_ptr + (n >> DUVSHIFT) * BPP, out, SS(r, DUVSHIFT) * BPP); \ 226 } 227 228 #ifdef HAS_I210TOAR30ROW_SSSE3 229 ANY31CT(I210ToAR30Row_Any_SSSE3, I210ToAR30Row_SSSE3, 1, 0, uint16_t, 2, 4, 7) 230 #endif 231 #ifdef HAS_I210TOARGBROW_SSSE3 232 ANY31CT(I210ToARGBRow_Any_SSSE3, I210ToARGBRow_SSSE3, 1, 0, uint16_t, 2, 4, 7) 233 #endif 234 #ifdef HAS_I210TOARGBROW_AVX2 235 ANY31CT(I210ToARGBRow_Any_AVX2, I210ToARGBRow_AVX2, 1, 0, uint16_t, 2, 4, 15) 236 #endif 237 #ifdef HAS_I210TOAR30ROW_AVX2 238 ANY31CT(I210ToAR30Row_Any_AVX2, I210ToAR30Row_AVX2, 1, 0, uint16_t, 2, 4, 15) 239 #endif 240 #undef ANY31CT 241 242 // Any 2 planes to 1. 243 #define ANY21(NAMEANY, ANY_SIMD, UVSHIFT, SBPP, SBPP2, BPP, MASK) \ 244 void NAMEANY(const uint8_t* y_buf, const uint8_t* uv_buf, uint8_t* dst_ptr, \ 245 int width) { \ 246 SIMD_ALIGNED(uint8_t temp[64 * 3]); \ 247 memset(temp, 0, 64 * 2); /* for msan */ \ 248 int r = width & MASK; \ 249 int n = width & ~MASK; \ 250 if (n > 0) { \ 251 ANY_SIMD(y_buf, uv_buf, dst_ptr, n); \ 252 } \ 253 memcpy(temp, y_buf + n * SBPP, r * SBPP); \ 254 memcpy(temp + 64, uv_buf + (n >> UVSHIFT) * SBPP2, \ 255 SS(r, UVSHIFT) * SBPP2); \ 256 ANY_SIMD(temp, temp + 64, temp + 128, MASK + 1); \ 257 memcpy(dst_ptr + n * BPP, temp + 128, r * BPP); \ 258 } 259 260 // Merge functions. 261 #ifdef HAS_MERGEUVROW_SSE2 262 ANY21(MergeUVRow_Any_SSE2, MergeUVRow_SSE2, 0, 1, 1, 2, 15) 263 #endif 264 #ifdef HAS_MERGEUVROW_AVX2 265 ANY21(MergeUVRow_Any_AVX2, MergeUVRow_AVX2, 0, 1, 1, 2, 31) 266 #endif 267 #ifdef HAS_MERGEUVROW_NEON 268 ANY21(MergeUVRow_Any_NEON, MergeUVRow_NEON, 0, 1, 1, 2, 15) 269 #endif 270 #ifdef HAS_MERGEUVROW_MSA 271 ANY21(MergeUVRow_Any_MSA, MergeUVRow_MSA, 0, 1, 1, 2, 15) 272 #endif 273 274 // Math functions. 275 #ifdef HAS_ARGBMULTIPLYROW_SSE2 276 ANY21(ARGBMultiplyRow_Any_SSE2, ARGBMultiplyRow_SSE2, 0, 4, 4, 4, 3) 277 #endif 278 #ifdef HAS_ARGBADDROW_SSE2 279 ANY21(ARGBAddRow_Any_SSE2, ARGBAddRow_SSE2, 0, 4, 4, 4, 3) 280 #endif 281 #ifdef HAS_ARGBSUBTRACTROW_SSE2 282 ANY21(ARGBSubtractRow_Any_SSE2, ARGBSubtractRow_SSE2, 0, 4, 4, 4, 3) 283 #endif 284 #ifdef HAS_ARGBMULTIPLYROW_AVX2 285 ANY21(ARGBMultiplyRow_Any_AVX2, ARGBMultiplyRow_AVX2, 0, 4, 4, 4, 7) 286 #endif 287 #ifdef HAS_ARGBADDROW_AVX2 288 ANY21(ARGBAddRow_Any_AVX2, ARGBAddRow_AVX2, 0, 4, 4, 4, 7) 289 #endif 290 #ifdef HAS_ARGBSUBTRACTROW_AVX2 291 ANY21(ARGBSubtractRow_Any_AVX2, ARGBSubtractRow_AVX2, 0, 4, 4, 4, 7) 292 #endif 293 #ifdef HAS_ARGBMULTIPLYROW_NEON 294 ANY21(ARGBMultiplyRow_Any_NEON, ARGBMultiplyRow_NEON, 0, 4, 4, 4, 7) 295 #endif 296 #ifdef HAS_ARGBADDROW_NEON 297 ANY21(ARGBAddRow_Any_NEON, ARGBAddRow_NEON, 0, 4, 4, 4, 7) 298 #endif 299 #ifdef HAS_ARGBSUBTRACTROW_NEON 300 ANY21(ARGBSubtractRow_Any_NEON, ARGBSubtractRow_NEON, 0, 4, 4, 4, 7) 301 #endif 302 #ifdef HAS_ARGBMULTIPLYROW_MSA 303 ANY21(ARGBMultiplyRow_Any_MSA, ARGBMultiplyRow_MSA, 0, 4, 4, 4, 3) 304 #endif 305 #ifdef HAS_ARGBADDROW_MSA 306 ANY21(ARGBAddRow_Any_MSA, ARGBAddRow_MSA, 0, 4, 4, 4, 7) 307 #endif 308 #ifdef HAS_ARGBSUBTRACTROW_MSA 309 ANY21(ARGBSubtractRow_Any_MSA, ARGBSubtractRow_MSA, 0, 4, 4, 4, 7) 310 #endif 311 #ifdef HAS_SOBELROW_SSE2 312 ANY21(SobelRow_Any_SSE2, SobelRow_SSE2, 0, 1, 1, 4, 15) 313 #endif 314 #ifdef HAS_SOBELROW_NEON 315 ANY21(SobelRow_Any_NEON, SobelRow_NEON, 0, 1, 1, 4, 7) 316 #endif 317 #ifdef HAS_SOBELROW_MSA 318 ANY21(SobelRow_Any_MSA, SobelRow_MSA, 0, 1, 1, 4, 15) 319 #endif 320 #ifdef HAS_SOBELTOPLANEROW_SSE2 321 ANY21(SobelToPlaneRow_Any_SSE2, SobelToPlaneRow_SSE2, 0, 1, 1, 1, 15) 322 #endif 323 #ifdef HAS_SOBELTOPLANEROW_NEON 324 ANY21(SobelToPlaneRow_Any_NEON, SobelToPlaneRow_NEON, 0, 1, 1, 1, 15) 325 #endif 326 #ifdef HAS_SOBELTOPLANEROW_MSA 327 ANY21(SobelToPlaneRow_Any_MSA, SobelToPlaneRow_MSA, 0, 1, 1, 1, 31) 328 #endif 329 #ifdef HAS_SOBELXYROW_SSE2 330 ANY21(SobelXYRow_Any_SSE2, SobelXYRow_SSE2, 0, 1, 1, 4, 15) 331 #endif 332 #ifdef HAS_SOBELXYROW_NEON 333 ANY21(SobelXYRow_Any_NEON, SobelXYRow_NEON, 0, 1, 1, 4, 7) 334 #endif 335 #ifdef HAS_SOBELXYROW_MSA 336 ANY21(SobelXYRow_Any_MSA, SobelXYRow_MSA, 0, 1, 1, 4, 15) 337 #endif 338 #undef ANY21 339 340 // Any 2 planes to 1 with yuvconstants 341 #define ANY21C(NAMEANY, ANY_SIMD, UVSHIFT, SBPP, SBPP2, BPP, MASK) \ 342 void NAMEANY(const uint8_t* y_buf, const uint8_t* uv_buf, uint8_t* dst_ptr, \ 343 const struct YuvConstants* yuvconstants, int width) { \ 344 SIMD_ALIGNED(uint8_t temp[64 * 3]); \ 345 memset(temp, 0, 64 * 2); /* for msan */ \ 346 int r = width & MASK; \ 347 int n = width & ~MASK; \ 348 if (n > 0) { \ 349 ANY_SIMD(y_buf, uv_buf, dst_ptr, yuvconstants, n); \ 350 } \ 351 memcpy(temp, y_buf + n * SBPP, r * SBPP); \ 352 memcpy(temp + 64, uv_buf + (n >> UVSHIFT) * SBPP2, \ 353 SS(r, UVSHIFT) * SBPP2); \ 354 ANY_SIMD(temp, temp + 64, temp + 128, yuvconstants, MASK + 1); \ 355 memcpy(dst_ptr + n * BPP, temp + 128, r * BPP); \ 356 } 357 358 // Biplanar to RGB. 359 #ifdef HAS_NV12TOARGBROW_SSSE3 360 ANY21C(NV12ToARGBRow_Any_SSSE3, NV12ToARGBRow_SSSE3, 1, 1, 2, 4, 7) 361 #endif 362 #ifdef HAS_NV12TOARGBROW_AVX2 363 ANY21C(NV12ToARGBRow_Any_AVX2, NV12ToARGBRow_AVX2, 1, 1, 2, 4, 15) 364 #endif 365 #ifdef HAS_NV12TOARGBROW_NEON 366 ANY21C(NV12ToARGBRow_Any_NEON, NV12ToARGBRow_NEON, 1, 1, 2, 4, 7) 367 #endif 368 #ifdef HAS_NV12TOARGBROW_MSA 369 ANY21C(NV12ToARGBRow_Any_MSA, NV12ToARGBRow_MSA, 1, 1, 2, 4, 7) 370 #endif 371 #ifdef HAS_NV21TOARGBROW_SSSE3 372 ANY21C(NV21ToARGBRow_Any_SSSE3, NV21ToARGBRow_SSSE3, 1, 1, 2, 4, 7) 373 #endif 374 #ifdef HAS_NV21TOARGBROW_AVX2 375 ANY21C(NV21ToARGBRow_Any_AVX2, NV21ToARGBRow_AVX2, 1, 1, 2, 4, 15) 376 #endif 377 #ifdef HAS_NV21TOARGBROW_NEON 378 ANY21C(NV21ToARGBRow_Any_NEON, NV21ToARGBRow_NEON, 1, 1, 2, 4, 7) 379 #endif 380 #ifdef HAS_NV21TOARGBROW_MSA 381 ANY21C(NV21ToARGBRow_Any_MSA, NV21ToARGBRow_MSA, 1, 1, 2, 4, 7) 382 #endif 383 #ifdef HAS_NV12TORGB24ROW_NEON 384 ANY21C(NV12ToRGB24Row_Any_NEON, NV12ToRGB24Row_NEON, 1, 1, 2, 3, 7) 385 #endif 386 #ifdef HAS_NV21TORGB24ROW_NEON 387 ANY21C(NV21ToRGB24Row_Any_NEON, NV21ToRGB24Row_NEON, 1, 1, 2, 3, 7) 388 #endif 389 #ifdef HAS_NV12TORGB565ROW_SSSE3 390 ANY21C(NV12ToRGB565Row_Any_SSSE3, NV12ToRGB565Row_SSSE3, 1, 1, 2, 2, 7) 391 #endif 392 #ifdef HAS_NV12TORGB565ROW_AVX2 393 ANY21C(NV12ToRGB565Row_Any_AVX2, NV12ToRGB565Row_AVX2, 1, 1, 2, 2, 15) 394 #endif 395 #ifdef HAS_NV12TORGB565ROW_NEON 396 ANY21C(NV12ToRGB565Row_Any_NEON, NV12ToRGB565Row_NEON, 1, 1, 2, 2, 7) 397 #endif 398 #ifdef HAS_NV12TORGB565ROW_MSA 399 ANY21C(NV12ToRGB565Row_Any_MSA, NV12ToRGB565Row_MSA, 1, 1, 2, 2, 7) 400 #endif 401 #undef ANY21C 402 403 // Any 1 to 1. 404 #define ANY11(NAMEANY, ANY_SIMD, UVSHIFT, SBPP, BPP, MASK) \ 405 void NAMEANY(const uint8_t* src_ptr, uint8_t* dst_ptr, int width) { \ 406 SIMD_ALIGNED(uint8_t temp[128 * 2]); \ 407 memset(temp, 0, 128); /* for YUY2 and msan */ \ 408 int r = width & MASK; \ 409 int n = width & ~MASK; \ 410 if (n > 0) { \ 411 ANY_SIMD(src_ptr, dst_ptr, n); \ 412 } \ 413 memcpy(temp, src_ptr + (n >> UVSHIFT) * SBPP, SS(r, UVSHIFT) * SBPP); \ 414 ANY_SIMD(temp, temp + 128, MASK + 1); \ 415 memcpy(dst_ptr + n * BPP, temp + 128, r * BPP); \ 416 } 417 418 #ifdef HAS_COPYROW_AVX 419 ANY11(CopyRow_Any_AVX, CopyRow_AVX, 0, 1, 1, 63) 420 #endif 421 #ifdef HAS_COPYROW_SSE2 422 ANY11(CopyRow_Any_SSE2, CopyRow_SSE2, 0, 1, 1, 31) 423 #endif 424 #ifdef HAS_COPYROW_NEON 425 ANY11(CopyRow_Any_NEON, CopyRow_NEON, 0, 1, 1, 31) 426 #endif 427 #if defined(HAS_ARGBTORGB24ROW_SSSE3) 428 ANY11(ARGBToRGB24Row_Any_SSSE3, ARGBToRGB24Row_SSSE3, 0, 4, 3, 15) 429 ANY11(ARGBToRAWRow_Any_SSSE3, ARGBToRAWRow_SSSE3, 0, 4, 3, 15) 430 ANY11(ARGBToRGB565Row_Any_SSE2, ARGBToRGB565Row_SSE2, 0, 4, 2, 3) 431 ANY11(ARGBToARGB1555Row_Any_SSE2, ARGBToARGB1555Row_SSE2, 0, 4, 2, 3) 432 ANY11(ARGBToARGB4444Row_Any_SSE2, ARGBToARGB4444Row_SSE2, 0, 4, 2, 3) 433 #endif 434 #if defined(HAS_ARGBTORGB565ROW_AVX2) 435 ANY11(ARGBToRGB565Row_Any_AVX2, ARGBToRGB565Row_AVX2, 0, 4, 2, 7) 436 #endif 437 #if defined(HAS_ARGBTOARGB4444ROW_AVX2) 438 ANY11(ARGBToARGB1555Row_Any_AVX2, ARGBToARGB1555Row_AVX2, 0, 4, 2, 7) 439 ANY11(ARGBToARGB4444Row_Any_AVX2, ARGBToARGB4444Row_AVX2, 0, 4, 2, 7) 440 #endif 441 #if defined(HAS_ABGRTOAR30ROW_SSSE3) 442 ANY11(ABGRToAR30Row_Any_SSSE3, ABGRToAR30Row_SSSE3, 0, 4, 4, 3) 443 #endif 444 #if defined(HAS_ARGBTOAR30ROW_SSSE3) 445 ANY11(ARGBToAR30Row_Any_SSSE3, ARGBToAR30Row_SSSE3, 0, 4, 4, 3) 446 #endif 447 #if defined(HAS_ABGRTOAR30ROW_AVX2) 448 ANY11(ABGRToAR30Row_Any_AVX2, ABGRToAR30Row_AVX2, 0, 4, 4, 7) 449 #endif 450 #if defined(HAS_ARGBTOAR30ROW_AVX2) 451 ANY11(ARGBToAR30Row_Any_AVX2, ARGBToAR30Row_AVX2, 0, 4, 4, 7) 452 #endif 453 #if defined(HAS_J400TOARGBROW_SSE2) 454 ANY11(J400ToARGBRow_Any_SSE2, J400ToARGBRow_SSE2, 0, 1, 4, 7) 455 #endif 456 #if defined(HAS_J400TOARGBROW_AVX2) 457 ANY11(J400ToARGBRow_Any_AVX2, J400ToARGBRow_AVX2, 0, 1, 4, 15) 458 #endif 459 #if defined(HAS_I400TOARGBROW_SSE2) 460 ANY11(I400ToARGBRow_Any_SSE2, I400ToARGBRow_SSE2, 0, 1, 4, 7) 461 #endif 462 #if defined(HAS_I400TOARGBROW_AVX2) 463 ANY11(I400ToARGBRow_Any_AVX2, I400ToARGBRow_AVX2, 0, 1, 4, 15) 464 #endif 465 #if defined(HAS_RGB24TOARGBROW_SSSE3) 466 ANY11(RGB24ToARGBRow_Any_SSSE3, RGB24ToARGBRow_SSSE3, 0, 3, 4, 15) 467 ANY11(RAWToARGBRow_Any_SSSE3, RAWToARGBRow_SSSE3, 0, 3, 4, 15) 468 ANY11(RGB565ToARGBRow_Any_SSE2, RGB565ToARGBRow_SSE2, 0, 2, 4, 7) 469 ANY11(ARGB1555ToARGBRow_Any_SSE2, ARGB1555ToARGBRow_SSE2, 0, 2, 4, 7) 470 ANY11(ARGB4444ToARGBRow_Any_SSE2, ARGB4444ToARGBRow_SSE2, 0, 2, 4, 7) 471 #endif 472 #if defined(HAS_RAWTORGB24ROW_SSSE3) 473 ANY11(RAWToRGB24Row_Any_SSSE3, RAWToRGB24Row_SSSE3, 0, 3, 3, 7) 474 #endif 475 #if defined(HAS_RGB565TOARGBROW_AVX2) 476 ANY11(RGB565ToARGBRow_Any_AVX2, RGB565ToARGBRow_AVX2, 0, 2, 4, 15) 477 #endif 478 #if defined(HAS_ARGB1555TOARGBROW_AVX2) 479 ANY11(ARGB1555ToARGBRow_Any_AVX2, ARGB1555ToARGBRow_AVX2, 0, 2, 4, 15) 480 #endif 481 #if defined(HAS_ARGB4444TOARGBROW_AVX2) 482 ANY11(ARGB4444ToARGBRow_Any_AVX2, ARGB4444ToARGBRow_AVX2, 0, 2, 4, 15) 483 #endif 484 #if defined(HAS_ARGBTORGB24ROW_NEON) 485 ANY11(ARGBToRGB24Row_Any_NEON, ARGBToRGB24Row_NEON, 0, 4, 3, 7) 486 ANY11(ARGBToRAWRow_Any_NEON, ARGBToRAWRow_NEON, 0, 4, 3, 7) 487 ANY11(ARGBToRGB565Row_Any_NEON, ARGBToRGB565Row_NEON, 0, 4, 2, 7) 488 ANY11(ARGBToARGB1555Row_Any_NEON, ARGBToARGB1555Row_NEON, 0, 4, 2, 7) 489 ANY11(ARGBToARGB4444Row_Any_NEON, ARGBToARGB4444Row_NEON, 0, 4, 2, 7) 490 ANY11(J400ToARGBRow_Any_NEON, J400ToARGBRow_NEON, 0, 1, 4, 7) 491 ANY11(I400ToARGBRow_Any_NEON, I400ToARGBRow_NEON, 0, 1, 4, 7) 492 #endif 493 #if defined(HAS_ARGBTORGB24ROW_MSA) 494 ANY11(ARGBToRGB24Row_Any_MSA, ARGBToRGB24Row_MSA, 0, 4, 3, 15) 495 ANY11(ARGBToRAWRow_Any_MSA, ARGBToRAWRow_MSA, 0, 4, 3, 15) 496 ANY11(ARGBToRGB565Row_Any_MSA, ARGBToRGB565Row_MSA, 0, 4, 2, 7) 497 ANY11(ARGBToARGB1555Row_Any_MSA, ARGBToARGB1555Row_MSA, 0, 4, 2, 7) 498 ANY11(ARGBToARGB4444Row_Any_MSA, ARGBToARGB4444Row_MSA, 0, 4, 2, 7) 499 ANY11(J400ToARGBRow_Any_MSA, J400ToARGBRow_MSA, 0, 1, 4, 15) 500 ANY11(I400ToARGBRow_Any_MSA, I400ToARGBRow_MSA, 0, 1, 4, 15) 501 #endif 502 #if defined(HAS_RAWTORGB24ROW_NEON) 503 ANY11(RAWToRGB24Row_Any_NEON, RAWToRGB24Row_NEON, 0, 3, 3, 7) 504 #endif 505 #if defined(HAS_RAWTORGB24ROW_MSA) 506 ANY11(RAWToRGB24Row_Any_MSA, RAWToRGB24Row_MSA, 0, 3, 3, 15) 507 #endif 508 #ifdef HAS_ARGBTOYROW_AVX2 509 ANY11(ARGBToYRow_Any_AVX2, ARGBToYRow_AVX2, 0, 4, 1, 31) 510 #endif 511 #ifdef HAS_ARGBTOYJROW_AVX2 512 ANY11(ARGBToYJRow_Any_AVX2, ARGBToYJRow_AVX2, 0, 4, 1, 31) 513 #endif 514 #ifdef HAS_UYVYTOYROW_AVX2 515 ANY11(UYVYToYRow_Any_AVX2, UYVYToYRow_AVX2, 0, 2, 1, 31) 516 #endif 517 #ifdef HAS_YUY2TOYROW_AVX2 518 ANY11(YUY2ToYRow_Any_AVX2, YUY2ToYRow_AVX2, 1, 4, 1, 31) 519 #endif 520 #ifdef HAS_ARGBTOYROW_SSSE3 521 ANY11(ARGBToYRow_Any_SSSE3, ARGBToYRow_SSSE3, 0, 4, 1, 15) 522 #endif 523 #ifdef HAS_BGRATOYROW_SSSE3 524 ANY11(BGRAToYRow_Any_SSSE3, BGRAToYRow_SSSE3, 0, 4, 1, 15) 525 ANY11(ABGRToYRow_Any_SSSE3, ABGRToYRow_SSSE3, 0, 4, 1, 15) 526 ANY11(RGBAToYRow_Any_SSSE3, RGBAToYRow_SSSE3, 0, 4, 1, 15) 527 ANY11(YUY2ToYRow_Any_SSE2, YUY2ToYRow_SSE2, 1, 4, 1, 15) 528 ANY11(UYVYToYRow_Any_SSE2, UYVYToYRow_SSE2, 1, 4, 1, 15) 529 #endif 530 #ifdef HAS_ARGBTOYJROW_SSSE3 531 ANY11(ARGBToYJRow_Any_SSSE3, ARGBToYJRow_SSSE3, 0, 4, 1, 15) 532 #endif 533 #ifdef HAS_ARGBTOYROW_NEON 534 ANY11(ARGBToYRow_Any_NEON, ARGBToYRow_NEON, 0, 4, 1, 7) 535 #endif 536 #ifdef HAS_ARGBTOYROW_MSA 537 ANY11(ARGBToYRow_Any_MSA, ARGBToYRow_MSA, 0, 4, 1, 15) 538 #endif 539 #ifdef HAS_ARGBTOYJROW_NEON 540 ANY11(ARGBToYJRow_Any_NEON, ARGBToYJRow_NEON, 0, 4, 1, 7) 541 #endif 542 #ifdef HAS_ARGBTOYJROW_MSA 543 ANY11(ARGBToYJRow_Any_MSA, ARGBToYJRow_MSA, 0, 4, 1, 15) 544 #endif 545 #ifdef HAS_BGRATOYROW_NEON 546 ANY11(BGRAToYRow_Any_NEON, BGRAToYRow_NEON, 0, 4, 1, 7) 547 #endif 548 #ifdef HAS_BGRATOYROW_MSA 549 ANY11(BGRAToYRow_Any_MSA, BGRAToYRow_MSA, 0, 4, 1, 15) 550 #endif 551 #ifdef HAS_ABGRTOYROW_NEON 552 ANY11(ABGRToYRow_Any_NEON, ABGRToYRow_NEON, 0, 4, 1, 7) 553 #endif 554 #ifdef HAS_ABGRTOYROW_MSA 555 ANY11(ABGRToYRow_Any_MSA, ABGRToYRow_MSA, 0, 4, 1, 7) 556 #endif 557 #ifdef HAS_RGBATOYROW_NEON 558 ANY11(RGBAToYRow_Any_NEON, RGBAToYRow_NEON, 0, 4, 1, 7) 559 #endif 560 #ifdef HAS_RGBATOYROW_MSA 561 ANY11(RGBAToYRow_Any_MSA, RGBAToYRow_MSA, 0, 4, 1, 15) 562 #endif 563 #ifdef HAS_RGB24TOYROW_NEON 564 ANY11(RGB24ToYRow_Any_NEON, RGB24ToYRow_NEON, 0, 3, 1, 7) 565 #endif 566 #ifdef HAS_RGB24TOYROW_MSA 567 ANY11(RGB24ToYRow_Any_MSA, RGB24ToYRow_MSA, 0, 3, 1, 15) 568 #endif 569 #ifdef HAS_RAWTOYROW_NEON 570 ANY11(RAWToYRow_Any_NEON, RAWToYRow_NEON, 0, 3, 1, 7) 571 #endif 572 #ifdef HAS_RAWTOYROW_MSA 573 ANY11(RAWToYRow_Any_MSA, RAWToYRow_MSA, 0, 3, 1, 15) 574 #endif 575 #ifdef HAS_RGB565TOYROW_NEON 576 ANY11(RGB565ToYRow_Any_NEON, RGB565ToYRow_NEON, 0, 2, 1, 7) 577 #endif 578 #ifdef HAS_RGB565TOYROW_MSA 579 ANY11(RGB565ToYRow_Any_MSA, RGB565ToYRow_MSA, 0, 2, 1, 15) 580 #endif 581 #ifdef HAS_ARGB1555TOYROW_NEON 582 ANY11(ARGB1555ToYRow_Any_NEON, ARGB1555ToYRow_NEON, 0, 2, 1, 7) 583 #endif 584 #ifdef HAS_ARGB1555TOYROW_MSA 585 ANY11(ARGB1555ToYRow_Any_MSA, ARGB1555ToYRow_MSA, 0, 2, 1, 15) 586 #endif 587 #ifdef HAS_ARGB4444TOYROW_NEON 588 ANY11(ARGB4444ToYRow_Any_NEON, ARGB4444ToYRow_NEON, 0, 2, 1, 7) 589 #endif 590 #ifdef HAS_YUY2TOYROW_NEON 591 ANY11(YUY2ToYRow_Any_NEON, YUY2ToYRow_NEON, 1, 4, 1, 15) 592 #endif 593 #ifdef HAS_UYVYTOYROW_NEON 594 ANY11(UYVYToYRow_Any_NEON, UYVYToYRow_NEON, 1, 4, 1, 15) 595 #endif 596 #ifdef HAS_YUY2TOYROW_MSA 597 ANY11(YUY2ToYRow_Any_MSA, YUY2ToYRow_MSA, 1, 4, 1, 31) 598 #endif 599 #ifdef HAS_UYVYTOYROW_MSA 600 ANY11(UYVYToYRow_Any_MSA, UYVYToYRow_MSA, 1, 4, 1, 31) 601 #endif 602 #ifdef HAS_RGB24TOARGBROW_NEON 603 ANY11(RGB24ToARGBRow_Any_NEON, RGB24ToARGBRow_NEON, 0, 3, 4, 7) 604 #endif 605 #ifdef HAS_RGB24TOARGBROW_MSA 606 ANY11(RGB24ToARGBRow_Any_MSA, RGB24ToARGBRow_MSA, 0, 3, 4, 15) 607 #endif 608 #ifdef HAS_RAWTOARGBROW_NEON 609 ANY11(RAWToARGBRow_Any_NEON, RAWToARGBRow_NEON, 0, 3, 4, 7) 610 #endif 611 #ifdef HAS_RAWTOARGBROW_MSA 612 ANY11(RAWToARGBRow_Any_MSA, RAWToARGBRow_MSA, 0, 3, 4, 15) 613 #endif 614 #ifdef HAS_RGB565TOARGBROW_NEON 615 ANY11(RGB565ToARGBRow_Any_NEON, RGB565ToARGBRow_NEON, 0, 2, 4, 7) 616 #endif 617 #ifdef HAS_RGB565TOARGBROW_MSA 618 ANY11(RGB565ToARGBRow_Any_MSA, RGB565ToARGBRow_MSA, 0, 2, 4, 15) 619 #endif 620 #ifdef HAS_ARGB1555TOARGBROW_NEON 621 ANY11(ARGB1555ToARGBRow_Any_NEON, ARGB1555ToARGBRow_NEON, 0, 2, 4, 7) 622 #endif 623 #ifdef HAS_ARGB1555TOARGBROW_MSA 624 ANY11(ARGB1555ToARGBRow_Any_MSA, ARGB1555ToARGBRow_MSA, 0, 2, 4, 15) 625 #endif 626 #ifdef HAS_ARGB4444TOARGBROW_NEON 627 ANY11(ARGB4444ToARGBRow_Any_NEON, ARGB4444ToARGBRow_NEON, 0, 2, 4, 7) 628 #endif 629 #ifdef HAS_ARGB4444TOARGBROW_MSA 630 ANY11(ARGB4444ToARGBRow_Any_MSA, ARGB4444ToARGBRow_MSA, 0, 2, 4, 15) 631 #endif 632 #ifdef HAS_ARGBATTENUATEROW_SSSE3 633 ANY11(ARGBAttenuateRow_Any_SSSE3, ARGBAttenuateRow_SSSE3, 0, 4, 4, 3) 634 #endif 635 #ifdef HAS_ARGBUNATTENUATEROW_SSE2 636 ANY11(ARGBUnattenuateRow_Any_SSE2, ARGBUnattenuateRow_SSE2, 0, 4, 4, 3) 637 #endif 638 #ifdef HAS_ARGBATTENUATEROW_AVX2 639 ANY11(ARGBAttenuateRow_Any_AVX2, ARGBAttenuateRow_AVX2, 0, 4, 4, 7) 640 #endif 641 #ifdef HAS_ARGBUNATTENUATEROW_AVX2 642 ANY11(ARGBUnattenuateRow_Any_AVX2, ARGBUnattenuateRow_AVX2, 0, 4, 4, 7) 643 #endif 644 #ifdef HAS_ARGBATTENUATEROW_NEON 645 ANY11(ARGBAttenuateRow_Any_NEON, ARGBAttenuateRow_NEON, 0, 4, 4, 7) 646 #endif 647 #ifdef HAS_ARGBATTENUATEROW_MSA 648 ANY11(ARGBAttenuateRow_Any_MSA, ARGBAttenuateRow_MSA, 0, 4, 4, 7) 649 #endif 650 #ifdef HAS_ARGBEXTRACTALPHAROW_SSE2 651 ANY11(ARGBExtractAlphaRow_Any_SSE2, ARGBExtractAlphaRow_SSE2, 0, 4, 1, 7) 652 #endif 653 #ifdef HAS_ARGBEXTRACTALPHAROW_AVX2 654 ANY11(ARGBExtractAlphaRow_Any_AVX2, ARGBExtractAlphaRow_AVX2, 0, 4, 1, 32) 655 #endif 656 #ifdef HAS_ARGBEXTRACTALPHAROW_NEON 657 ANY11(ARGBExtractAlphaRow_Any_NEON, ARGBExtractAlphaRow_NEON, 0, 4, 1, 15) 658 #endif 659 #ifdef HAS_ARGBEXTRACTALPHAROW_MSA 660 ANY11(ARGBExtractAlphaRow_Any_MSA, ARGBExtractAlphaRow_MSA, 0, 4, 1, 15) 661 #endif 662 #undef ANY11 663 664 // Any 1 to 1 blended. Destination is read, modify, write. 665 #define ANY11B(NAMEANY, ANY_SIMD, UVSHIFT, SBPP, BPP, MASK) \ 666 void NAMEANY(const uint8_t* src_ptr, uint8_t* dst_ptr, int width) { \ 667 SIMD_ALIGNED(uint8_t temp[64 * 2]); \ 668 memset(temp, 0, 64 * 2); /* for msan */ \ 669 int r = width & MASK; \ 670 int n = width & ~MASK; \ 671 if (n > 0) { \ 672 ANY_SIMD(src_ptr, dst_ptr, n); \ 673 } \ 674 memcpy(temp, src_ptr + (n >> UVSHIFT) * SBPP, SS(r, UVSHIFT) * SBPP); \ 675 memcpy(temp + 64, dst_ptr + n * BPP, r * BPP); \ 676 ANY_SIMD(temp, temp + 64, MASK + 1); \ 677 memcpy(dst_ptr + n * BPP, temp + 64, r * BPP); \ 678 } 679 680 #ifdef HAS_ARGBCOPYALPHAROW_AVX2 681 ANY11B(ARGBCopyAlphaRow_Any_AVX2, ARGBCopyAlphaRow_AVX2, 0, 4, 4, 15) 682 #endif 683 #ifdef HAS_ARGBCOPYALPHAROW_SSE2 684 ANY11B(ARGBCopyAlphaRow_Any_SSE2, ARGBCopyAlphaRow_SSE2, 0, 4, 4, 7) 685 #endif 686 #ifdef HAS_ARGBCOPYYTOALPHAROW_AVX2 687 ANY11B(ARGBCopyYToAlphaRow_Any_AVX2, ARGBCopyYToAlphaRow_AVX2, 0, 1, 4, 15) 688 #endif 689 #ifdef HAS_ARGBCOPYYTOALPHAROW_SSE2 690 ANY11B(ARGBCopyYToAlphaRow_Any_SSE2, ARGBCopyYToAlphaRow_SSE2, 0, 1, 4, 7) 691 #endif 692 #undef ANY11B 693 694 // Any 1 to 1 with parameter. 695 #define ANY11P(NAMEANY, ANY_SIMD, T, SBPP, BPP, MASK) \ 696 void NAMEANY(const uint8_t* src_ptr, uint8_t* dst_ptr, T param, int width) { \ 697 SIMD_ALIGNED(uint8_t temp[64 * 2]); \ 698 memset(temp, 0, 64); /* for msan */ \ 699 int r = width & MASK; \ 700 int n = width & ~MASK; \ 701 if (n > 0) { \ 702 ANY_SIMD(src_ptr, dst_ptr, param, n); \ 703 } \ 704 memcpy(temp, src_ptr + n * SBPP, r * SBPP); \ 705 ANY_SIMD(temp, temp + 64, param, MASK + 1); \ 706 memcpy(dst_ptr + n * BPP, temp + 64, r * BPP); \ 707 } 708 709 #if defined(HAS_ARGBTORGB565DITHERROW_SSE2) 710 ANY11P(ARGBToRGB565DitherRow_Any_SSE2, 711 ARGBToRGB565DitherRow_SSE2, 712 const uint32_t, 713 4, 714 2, 715 3) 716 #endif 717 #if defined(HAS_ARGBTORGB565DITHERROW_AVX2) 718 ANY11P(ARGBToRGB565DitherRow_Any_AVX2, 719 ARGBToRGB565DitherRow_AVX2, 720 const uint32_t, 721 4, 722 2, 723 7) 724 #endif 725 #if defined(HAS_ARGBTORGB565DITHERROW_NEON) 726 ANY11P(ARGBToRGB565DitherRow_Any_NEON, 727 ARGBToRGB565DitherRow_NEON, 728 const uint32_t, 729 4, 730 2, 731 7) 732 #endif 733 #if defined(HAS_ARGBTORGB565DITHERROW_MSA) 734 ANY11P(ARGBToRGB565DitherRow_Any_MSA, 735 ARGBToRGB565DitherRow_MSA, 736 const uint32_t, 737 4, 738 2, 739 7) 740 #endif 741 #ifdef HAS_ARGBSHUFFLEROW_SSSE3 742 ANY11P(ARGBShuffleRow_Any_SSSE3, ARGBShuffleRow_SSSE3, const uint8_t*, 4, 4, 7) 743 #endif 744 #ifdef HAS_ARGBSHUFFLEROW_AVX2 745 ANY11P(ARGBShuffleRow_Any_AVX2, ARGBShuffleRow_AVX2, const uint8_t*, 4, 4, 15) 746 #endif 747 #ifdef HAS_ARGBSHUFFLEROW_NEON 748 ANY11P(ARGBShuffleRow_Any_NEON, ARGBShuffleRow_NEON, const uint8_t*, 4, 4, 3) 749 #endif 750 #ifdef HAS_ARGBSHUFFLEROW_MSA 751 ANY11P(ARGBShuffleRow_Any_MSA, ARGBShuffleRow_MSA, const uint8_t*, 4, 4, 7) 752 #endif 753 #undef ANY11P 754 755 // Any 1 to 1 with parameter and shorts. BPP measures in shorts. 756 #define ANY11C(NAMEANY, ANY_SIMD, SBPP, BPP, STYPE, DTYPE, MASK) \ 757 void NAMEANY(const STYPE* src_ptr, DTYPE* dst_ptr, int scale, int width) { \ 758 SIMD_ALIGNED(STYPE temp[32]); \ 759 SIMD_ALIGNED(DTYPE out[32]); \ 760 memset(temp, 0, 32 * SBPP); /* for msan */ \ 761 int r = width & MASK; \ 762 int n = width & ~MASK; \ 763 if (n > 0) { \ 764 ANY_SIMD(src_ptr, dst_ptr, scale, n); \ 765 } \ 766 memcpy(temp, src_ptr + n, r * SBPP); \ 767 ANY_SIMD(temp, out, scale, MASK + 1); \ 768 memcpy(dst_ptr + n, out, r * BPP); \ 769 } 770 771 #ifdef HAS_CONVERT16TO8ROW_SSSE3 772 ANY11C(Convert16To8Row_Any_SSSE3, 773 Convert16To8Row_SSSE3, 774 2, 775 1, 776 uint16_t, 777 uint8_t, 778 15) 779 #endif 780 #ifdef HAS_CONVERT16TO8ROW_AVX2 781 ANY11C(Convert16To8Row_Any_AVX2, 782 Convert16To8Row_AVX2, 783 2, 784 1, 785 uint16_t, 786 uint8_t, 787 31) 788 #endif 789 #ifdef HAS_CONVERT8TO16ROW_SSE2 790 ANY11C(Convert8To16Row_Any_SSE2, 791 Convert8To16Row_SSE2, 792 1, 793 2, 794 uint8_t, 795 uint16_t, 796 15) 797 #endif 798 #ifdef HAS_CONVERT8TO16ROW_AVX2 799 ANY11C(Convert8To16Row_Any_AVX2, 800 Convert8To16Row_AVX2, 801 1, 802 2, 803 uint8_t, 804 uint16_t, 805 31) 806 #endif 807 #undef ANY11C 808 809 // Any 1 to 1 with parameter and shorts to byte. BPP measures in shorts. 810 #define ANY11P16(NAMEANY, ANY_SIMD, ST, T, SBPP, BPP, MASK) \ 811 void NAMEANY(const ST* src_ptr, T* dst_ptr, float param, int width) { \ 812 SIMD_ALIGNED(ST temp[32]); \ 813 SIMD_ALIGNED(T out[32]); \ 814 memset(temp, 0, SBPP * 32); /* for msan */ \ 815 int r = width & MASK; \ 816 int n = width & ~MASK; \ 817 if (n > 0) { \ 818 ANY_SIMD(src_ptr, dst_ptr, param, n); \ 819 } \ 820 memcpy(temp, src_ptr + n, r * SBPP); \ 821 ANY_SIMD(temp, out, param, MASK + 1); \ 822 memcpy(dst_ptr + n, out, r * BPP); \ 823 } 824 825 #ifdef HAS_HALFFLOATROW_SSE2 826 ANY11P16(HalfFloatRow_Any_SSE2, HalfFloatRow_SSE2, uint16_t, uint16_t, 2, 2, 7) 827 #endif 828 #ifdef HAS_HALFFLOATROW_AVX2 829 ANY11P16(HalfFloatRow_Any_AVX2, HalfFloatRow_AVX2, uint16_t, uint16_t, 2, 2, 15) 830 #endif 831 #ifdef HAS_HALFFLOATROW_F16C 832 ANY11P16(HalfFloatRow_Any_F16C, HalfFloatRow_F16C, uint16_t, uint16_t, 2, 2, 15) 833 ANY11P16(HalfFloat1Row_Any_F16C, 834 HalfFloat1Row_F16C, 835 uint16_t, 836 uint16_t, 837 2, 838 2, 839 15) 840 #endif 841 #ifdef HAS_HALFFLOATROW_NEON 842 ANY11P16(HalfFloatRow_Any_NEON, HalfFloatRow_NEON, uint16_t, uint16_t, 2, 2, 7) 843 ANY11P16(HalfFloat1Row_Any_NEON, 844 HalfFloat1Row_NEON, 845 uint16_t, 846 uint16_t, 847 2, 848 2, 849 7) 850 #endif 851 #ifdef HAS_HALFFLOATROW_MSA 852 ANY11P16(HalfFloatRow_Any_MSA, HalfFloatRow_MSA, uint16_t, uint16_t, 2, 2, 31) 853 #endif 854 #ifdef HAS_BYTETOFLOATROW_NEON 855 ANY11P16(ByteToFloatRow_Any_NEON, ByteToFloatRow_NEON, uint8_t, float, 1, 3, 7) 856 #endif 857 #undef ANY11P16 858 859 // Any 1 to 1 with yuvconstants 860 #define ANY11C(NAMEANY, ANY_SIMD, UVSHIFT, SBPP, BPP, MASK) \ 861 void NAMEANY(const uint8_t* src_ptr, uint8_t* dst_ptr, \ 862 const struct YuvConstants* yuvconstants, int width) { \ 863 SIMD_ALIGNED(uint8_t temp[128 * 2]); \ 864 memset(temp, 0, 128); /* for YUY2 and msan */ \ 865 int r = width & MASK; \ 866 int n = width & ~MASK; \ 867 if (n > 0) { \ 868 ANY_SIMD(src_ptr, dst_ptr, yuvconstants, n); \ 869 } \ 870 memcpy(temp, src_ptr + (n >> UVSHIFT) * SBPP, SS(r, UVSHIFT) * SBPP); \ 871 ANY_SIMD(temp, temp + 128, yuvconstants, MASK + 1); \ 872 memcpy(dst_ptr + n * BPP, temp + 128, r * BPP); \ 873 } 874 #if defined(HAS_YUY2TOARGBROW_SSSE3) 875 ANY11C(YUY2ToARGBRow_Any_SSSE3, YUY2ToARGBRow_SSSE3, 1, 4, 4, 15) 876 ANY11C(UYVYToARGBRow_Any_SSSE3, UYVYToARGBRow_SSSE3, 1, 4, 4, 15) 877 #endif 878 #if defined(HAS_YUY2TOARGBROW_AVX2) 879 ANY11C(YUY2ToARGBRow_Any_AVX2, YUY2ToARGBRow_AVX2, 1, 4, 4, 31) 880 ANY11C(UYVYToARGBRow_Any_AVX2, UYVYToARGBRow_AVX2, 1, 4, 4, 31) 881 #endif 882 #if defined(HAS_YUY2TOARGBROW_NEON) 883 ANY11C(YUY2ToARGBRow_Any_NEON, YUY2ToARGBRow_NEON, 1, 4, 4, 7) 884 ANY11C(UYVYToARGBRow_Any_NEON, UYVYToARGBRow_NEON, 1, 4, 4, 7) 885 #endif 886 #if defined(HAS_YUY2TOARGBROW_MSA) 887 ANY11C(YUY2ToARGBRow_Any_MSA, YUY2ToARGBRow_MSA, 1, 4, 4, 7) 888 ANY11C(UYVYToARGBRow_Any_MSA, UYVYToARGBRow_MSA, 1, 4, 4, 7) 889 #endif 890 #undef ANY11C 891 892 // Any 1 to 1 interpolate. Takes 2 rows of source via stride. 893 #define ANY11T(NAMEANY, ANY_SIMD, SBPP, BPP, MASK) \ 894 void NAMEANY(uint8_t* dst_ptr, const uint8_t* src_ptr, \ 895 ptrdiff_t src_stride_ptr, int width, int source_y_fraction) { \ 896 SIMD_ALIGNED(uint8_t temp[64 * 3]); \ 897 memset(temp, 0, 64 * 2); /* for msan */ \ 898 int r = width & MASK; \ 899 int n = width & ~MASK; \ 900 if (n > 0) { \ 901 ANY_SIMD(dst_ptr, src_ptr, src_stride_ptr, n, source_y_fraction); \ 902 } \ 903 memcpy(temp, src_ptr + n * SBPP, r * SBPP); \ 904 memcpy(temp + 64, src_ptr + src_stride_ptr + n * SBPP, r * SBPP); \ 905 ANY_SIMD(temp + 128, temp, 64, MASK + 1, source_y_fraction); \ 906 memcpy(dst_ptr + n * BPP, temp + 128, r * BPP); \ 907 } 908 909 #ifdef HAS_INTERPOLATEROW_AVX2 910 ANY11T(InterpolateRow_Any_AVX2, InterpolateRow_AVX2, 1, 1, 31) 911 #endif 912 #ifdef HAS_INTERPOLATEROW_SSSE3 913 ANY11T(InterpolateRow_Any_SSSE3, InterpolateRow_SSSE3, 1, 1, 15) 914 #endif 915 #ifdef HAS_INTERPOLATEROW_NEON 916 ANY11T(InterpolateRow_Any_NEON, InterpolateRow_NEON, 1, 1, 15) 917 #endif 918 #ifdef HAS_INTERPOLATEROW_MSA 919 ANY11T(InterpolateRow_Any_MSA, InterpolateRow_MSA, 1, 1, 31) 920 #endif 921 #undef ANY11T 922 923 // Any 1 to 1 mirror. 924 #define ANY11M(NAMEANY, ANY_SIMD, BPP, MASK) \ 925 void NAMEANY(const uint8_t* src_ptr, uint8_t* dst_ptr, int width) { \ 926 SIMD_ALIGNED(uint8_t temp[64 * 2]); \ 927 memset(temp, 0, 64); /* for msan */ \ 928 int r = width & MASK; \ 929 int n = width & ~MASK; \ 930 if (n > 0) { \ 931 ANY_SIMD(src_ptr + r * BPP, dst_ptr, n); \ 932 } \ 933 memcpy(temp, src_ptr, r* BPP); \ 934 ANY_SIMD(temp, temp + 64, MASK + 1); \ 935 memcpy(dst_ptr + n * BPP, temp + 64 + (MASK + 1 - r) * BPP, r * BPP); \ 936 } 937 938 #ifdef HAS_MIRRORROW_AVX2 939 ANY11M(MirrorRow_Any_AVX2, MirrorRow_AVX2, 1, 31) 940 #endif 941 #ifdef HAS_MIRRORROW_SSSE3 942 ANY11M(MirrorRow_Any_SSSE3, MirrorRow_SSSE3, 1, 15) 943 #endif 944 #ifdef HAS_MIRRORROW_NEON 945 ANY11M(MirrorRow_Any_NEON, MirrorRow_NEON, 1, 15) 946 #endif 947 #ifdef HAS_MIRRORROW_MSA 948 ANY11M(MirrorRow_Any_MSA, MirrorRow_MSA, 1, 63) 949 #endif 950 #ifdef HAS_ARGBMIRRORROW_AVX2 951 ANY11M(ARGBMirrorRow_Any_AVX2, ARGBMirrorRow_AVX2, 4, 7) 952 #endif 953 #ifdef HAS_ARGBMIRRORROW_SSE2 954 ANY11M(ARGBMirrorRow_Any_SSE2, ARGBMirrorRow_SSE2, 4, 3) 955 #endif 956 #ifdef HAS_ARGBMIRRORROW_NEON 957 ANY11M(ARGBMirrorRow_Any_NEON, ARGBMirrorRow_NEON, 4, 3) 958 #endif 959 #ifdef HAS_ARGBMIRRORROW_MSA 960 ANY11M(ARGBMirrorRow_Any_MSA, ARGBMirrorRow_MSA, 4, 15) 961 #endif 962 #undef ANY11M 963 964 // Any 1 plane. (memset) 965 #define ANY1(NAMEANY, ANY_SIMD, T, BPP, MASK) \ 966 void NAMEANY(uint8_t* dst_ptr, T v32, int width) { \ 967 SIMD_ALIGNED(uint8_t temp[64]); \ 968 int r = width & MASK; \ 969 int n = width & ~MASK; \ 970 if (n > 0) { \ 971 ANY_SIMD(dst_ptr, v32, n); \ 972 } \ 973 ANY_SIMD(temp, v32, MASK + 1); \ 974 memcpy(dst_ptr + n * BPP, temp, r * BPP); \ 975 } 976 977 #ifdef HAS_SETROW_X86 978 ANY1(SetRow_Any_X86, SetRow_X86, uint8_t, 1, 3) 979 #endif 980 #ifdef HAS_SETROW_NEON 981 ANY1(SetRow_Any_NEON, SetRow_NEON, uint8_t, 1, 15) 982 #endif 983 #ifdef HAS_ARGBSETROW_NEON 984 ANY1(ARGBSetRow_Any_NEON, ARGBSetRow_NEON, uint32_t, 4, 3) 985 #endif 986 #ifdef HAS_ARGBSETROW_MSA 987 ANY1(ARGBSetRow_Any_MSA, ARGBSetRow_MSA, uint32_t, 4, 3) 988 #endif 989 #undef ANY1 990 991 // Any 1 to 2. Outputs UV planes. 992 #define ANY12(NAMEANY, ANY_SIMD, UVSHIFT, BPP, DUVSHIFT, MASK) \ 993 void NAMEANY(const uint8_t* src_ptr, uint8_t* dst_u, uint8_t* dst_v, \ 994 int width) { \ 995 SIMD_ALIGNED(uint8_t temp[128 * 3]); \ 996 memset(temp, 0, 128); /* for msan */ \ 997 int r = width & MASK; \ 998 int n = width & ~MASK; \ 999 if (n > 0) { \ 1000 ANY_SIMD(src_ptr, dst_u, dst_v, n); \ 1001 } \ 1002 memcpy(temp, src_ptr + (n >> UVSHIFT) * BPP, SS(r, UVSHIFT) * BPP); \ 1003 ANY_SIMD(temp, temp + 128, temp + 256, MASK + 1); \ 1004 memcpy(dst_u + (n >> DUVSHIFT), temp + 128, SS(r, DUVSHIFT)); \ 1005 memcpy(dst_v + (n >> DUVSHIFT), temp + 256, SS(r, DUVSHIFT)); \ 1006 } 1007 1008 #ifdef HAS_SPLITUVROW_SSE2 1009 ANY12(SplitUVRow_Any_SSE2, SplitUVRow_SSE2, 0, 2, 0, 15) 1010 #endif 1011 #ifdef HAS_SPLITUVROW_AVX2 1012 ANY12(SplitUVRow_Any_AVX2, SplitUVRow_AVX2, 0, 2, 0, 31) 1013 #endif 1014 #ifdef HAS_SPLITUVROW_NEON 1015 ANY12(SplitUVRow_Any_NEON, SplitUVRow_NEON, 0, 2, 0, 15) 1016 #endif 1017 #ifdef HAS_SPLITUVROW_MSA 1018 ANY12(SplitUVRow_Any_MSA, SplitUVRow_MSA, 0, 2, 0, 31) 1019 #endif 1020 #ifdef HAS_ARGBTOUV444ROW_SSSE3 1021 ANY12(ARGBToUV444Row_Any_SSSE3, ARGBToUV444Row_SSSE3, 0, 4, 0, 15) 1022 #endif 1023 #ifdef HAS_YUY2TOUV422ROW_AVX2 1024 ANY12(YUY2ToUV422Row_Any_AVX2, YUY2ToUV422Row_AVX2, 1, 4, 1, 31) 1025 ANY12(UYVYToUV422Row_Any_AVX2, UYVYToUV422Row_AVX2, 1, 4, 1, 31) 1026 #endif 1027 #ifdef HAS_YUY2TOUV422ROW_SSE2 1028 ANY12(YUY2ToUV422Row_Any_SSE2, YUY2ToUV422Row_SSE2, 1, 4, 1, 15) 1029 ANY12(UYVYToUV422Row_Any_SSE2, UYVYToUV422Row_SSE2, 1, 4, 1, 15) 1030 #endif 1031 #ifdef HAS_YUY2TOUV422ROW_NEON 1032 ANY12(ARGBToUV444Row_Any_NEON, ARGBToUV444Row_NEON, 0, 4, 0, 7) 1033 ANY12(YUY2ToUV422Row_Any_NEON, YUY2ToUV422Row_NEON, 1, 4, 1, 15) 1034 ANY12(UYVYToUV422Row_Any_NEON, UYVYToUV422Row_NEON, 1, 4, 1, 15) 1035 #endif 1036 #ifdef HAS_YUY2TOUV422ROW_MSA 1037 ANY12(ARGBToUV444Row_Any_MSA, ARGBToUV444Row_MSA, 0, 4, 0, 15) 1038 ANY12(YUY2ToUV422Row_Any_MSA, YUY2ToUV422Row_MSA, 1, 4, 1, 31) 1039 ANY12(UYVYToUV422Row_Any_MSA, UYVYToUV422Row_MSA, 1, 4, 1, 31) 1040 #endif 1041 #undef ANY12 1042 1043 // Any 1 to 3. Outputs RGB planes. 1044 #define ANY13(NAMEANY, ANY_SIMD, BPP, MASK) \ 1045 void NAMEANY(const uint8_t* src_ptr, uint8_t* dst_r, uint8_t* dst_g, \ 1046 uint8_t* dst_b, int width) { \ 1047 SIMD_ALIGNED(uint8_t temp[16 * 6]); \ 1048 memset(temp, 0, 16 * 3); /* for msan */ \ 1049 int r = width & MASK; \ 1050 int n = width & ~MASK; \ 1051 if (n > 0) { \ 1052 ANY_SIMD(src_ptr, dst_r, dst_g, dst_b, n); \ 1053 } \ 1054 memcpy(temp, src_ptr + n * BPP, r * BPP); \ 1055 ANY_SIMD(temp, temp + 16 * 3, temp + 16 * 4, temp + 16 * 5, MASK + 1); \ 1056 memcpy(dst_r + n, temp + 16 * 3, r); \ 1057 memcpy(dst_g + n, temp + 16 * 4, r); \ 1058 memcpy(dst_b + n, temp + 16 * 5, r); \ 1059 } 1060 1061 #ifdef HAS_SPLITRGBROW_SSSE3 1062 ANY13(SplitRGBRow_Any_SSSE3, SplitRGBRow_SSSE3, 3, 15) 1063 #endif 1064 #ifdef HAS_SPLITRGBROW_NEON 1065 ANY13(SplitRGBRow_Any_NEON, SplitRGBRow_NEON, 3, 15) 1066 #endif 1067 1068 // Any 1 to 2 with source stride (2 rows of source). Outputs UV planes. 1069 // 128 byte row allows for 32 avx ARGB pixels. 1070 #define ANY12S(NAMEANY, ANY_SIMD, UVSHIFT, BPP, MASK) \ 1071 void NAMEANY(const uint8_t* src_ptr, int src_stride_ptr, uint8_t* dst_u, \ 1072 uint8_t* dst_v, int width) { \ 1073 SIMD_ALIGNED(uint8_t temp[128 * 4]); \ 1074 memset(temp, 0, 128 * 2); /* for msan */ \ 1075 int r = width & MASK; \ 1076 int n = width & ~MASK; \ 1077 if (n > 0) { \ 1078 ANY_SIMD(src_ptr, src_stride_ptr, dst_u, dst_v, n); \ 1079 } \ 1080 memcpy(temp, src_ptr + (n >> UVSHIFT) * BPP, SS(r, UVSHIFT) * BPP); \ 1081 memcpy(temp + 128, src_ptr + src_stride_ptr + (n >> UVSHIFT) * BPP, \ 1082 SS(r, UVSHIFT) * BPP); \ 1083 if ((width & 1) && UVSHIFT == 0) { /* repeat last pixel for subsample */ \ 1084 memcpy(temp + SS(r, UVSHIFT) * BPP, temp + SS(r, UVSHIFT) * BPP - BPP, \ 1085 BPP); \ 1086 memcpy(temp + 128 + SS(r, UVSHIFT) * BPP, \ 1087 temp + 128 + SS(r, UVSHIFT) * BPP - BPP, BPP); \ 1088 } \ 1089 ANY_SIMD(temp, 128, temp + 256, temp + 384, MASK + 1); \ 1090 memcpy(dst_u + (n >> 1), temp + 256, SS(r, 1)); \ 1091 memcpy(dst_v + (n >> 1), temp + 384, SS(r, 1)); \ 1092 } 1093 1094 #ifdef HAS_ARGBTOUVROW_AVX2 1095 ANY12S(ARGBToUVRow_Any_AVX2, ARGBToUVRow_AVX2, 0, 4, 31) 1096 #endif 1097 #ifdef HAS_ARGBTOUVJROW_AVX2 1098 ANY12S(ARGBToUVJRow_Any_AVX2, ARGBToUVJRow_AVX2, 0, 4, 31) 1099 #endif 1100 #ifdef HAS_ARGBTOUVROW_SSSE3 1101 ANY12S(ARGBToUVRow_Any_SSSE3, ARGBToUVRow_SSSE3, 0, 4, 15) 1102 ANY12S(ARGBToUVJRow_Any_SSSE3, ARGBToUVJRow_SSSE3, 0, 4, 15) 1103 ANY12S(BGRAToUVRow_Any_SSSE3, BGRAToUVRow_SSSE3, 0, 4, 15) 1104 ANY12S(ABGRToUVRow_Any_SSSE3, ABGRToUVRow_SSSE3, 0, 4, 15) 1105 ANY12S(RGBAToUVRow_Any_SSSE3, RGBAToUVRow_SSSE3, 0, 4, 15) 1106 #endif 1107 #ifdef HAS_YUY2TOUVROW_AVX2 1108 ANY12S(YUY2ToUVRow_Any_AVX2, YUY2ToUVRow_AVX2, 1, 4, 31) 1109 ANY12S(UYVYToUVRow_Any_AVX2, UYVYToUVRow_AVX2, 1, 4, 31) 1110 #endif 1111 #ifdef HAS_YUY2TOUVROW_SSE2 1112 ANY12S(YUY2ToUVRow_Any_SSE2, YUY2ToUVRow_SSE2, 1, 4, 15) 1113 ANY12S(UYVYToUVRow_Any_SSE2, UYVYToUVRow_SSE2, 1, 4, 15) 1114 #endif 1115 #ifdef HAS_ARGBTOUVROW_NEON 1116 ANY12S(ARGBToUVRow_Any_NEON, ARGBToUVRow_NEON, 0, 4, 15) 1117 #endif 1118 #ifdef HAS_ARGBTOUVROW_MSA 1119 ANY12S(ARGBToUVRow_Any_MSA, ARGBToUVRow_MSA, 0, 4, 31) 1120 #endif 1121 #ifdef HAS_ARGBTOUVJROW_NEON 1122 ANY12S(ARGBToUVJRow_Any_NEON, ARGBToUVJRow_NEON, 0, 4, 15) 1123 #endif 1124 #ifdef HAS_ARGBTOUVJROW_MSA 1125 ANY12S(ARGBToUVJRow_Any_MSA, ARGBToUVJRow_MSA, 0, 4, 31) 1126 #endif 1127 #ifdef HAS_BGRATOUVROW_NEON 1128 ANY12S(BGRAToUVRow_Any_NEON, BGRAToUVRow_NEON, 0, 4, 15) 1129 #endif 1130 #ifdef HAS_BGRATOUVROW_MSA 1131 ANY12S(BGRAToUVRow_Any_MSA, BGRAToUVRow_MSA, 0, 4, 31) 1132 #endif 1133 #ifdef HAS_ABGRTOUVROW_NEON 1134 ANY12S(ABGRToUVRow_Any_NEON, ABGRToUVRow_NEON, 0, 4, 15) 1135 #endif 1136 #ifdef HAS_ABGRTOUVROW_MSA 1137 ANY12S(ABGRToUVRow_Any_MSA, ABGRToUVRow_MSA, 0, 4, 31) 1138 #endif 1139 #ifdef HAS_RGBATOUVROW_NEON 1140 ANY12S(RGBAToUVRow_Any_NEON, RGBAToUVRow_NEON, 0, 4, 15) 1141 #endif 1142 #ifdef HAS_RGBATOUVROW_MSA 1143 ANY12S(RGBAToUVRow_Any_MSA, RGBAToUVRow_MSA, 0, 4, 31) 1144 #endif 1145 #ifdef HAS_RGB24TOUVROW_NEON 1146 ANY12S(RGB24ToUVRow_Any_NEON, RGB24ToUVRow_NEON, 0, 3, 15) 1147 #endif 1148 #ifdef HAS_RGB24TOUVROW_MSA 1149 ANY12S(RGB24ToUVRow_Any_MSA, RGB24ToUVRow_MSA, 0, 3, 15) 1150 #endif 1151 #ifdef HAS_RAWTOUVROW_NEON 1152 ANY12S(RAWToUVRow_Any_NEON, RAWToUVRow_NEON, 0, 3, 15) 1153 #endif 1154 #ifdef HAS_RAWTOUVROW_MSA 1155 ANY12S(RAWToUVRow_Any_MSA, RAWToUVRow_MSA, 0, 3, 15) 1156 #endif 1157 #ifdef HAS_RGB565TOUVROW_NEON 1158 ANY12S(RGB565ToUVRow_Any_NEON, RGB565ToUVRow_NEON, 0, 2, 15) 1159 #endif 1160 #ifdef HAS_RGB565TOUVROW_MSA 1161 ANY12S(RGB565ToUVRow_Any_MSA, RGB565ToUVRow_MSA, 0, 2, 15) 1162 #endif 1163 #ifdef HAS_ARGB1555TOUVROW_NEON 1164 ANY12S(ARGB1555ToUVRow_Any_NEON, ARGB1555ToUVRow_NEON, 0, 2, 15) 1165 #endif 1166 #ifdef HAS_ARGB1555TOUVROW_MSA 1167 ANY12S(ARGB1555ToUVRow_Any_MSA, ARGB1555ToUVRow_MSA, 0, 2, 15) 1168 #endif 1169 #ifdef HAS_ARGB4444TOUVROW_NEON 1170 ANY12S(ARGB4444ToUVRow_Any_NEON, ARGB4444ToUVRow_NEON, 0, 2, 15) 1171 #endif 1172 #ifdef HAS_YUY2TOUVROW_NEON 1173 ANY12S(YUY2ToUVRow_Any_NEON, YUY2ToUVRow_NEON, 1, 4, 15) 1174 #endif 1175 #ifdef HAS_UYVYTOUVROW_NEON 1176 ANY12S(UYVYToUVRow_Any_NEON, UYVYToUVRow_NEON, 1, 4, 15) 1177 #endif 1178 #ifdef HAS_YUY2TOUVROW_MSA 1179 ANY12S(YUY2ToUVRow_Any_MSA, YUY2ToUVRow_MSA, 1, 4, 31) 1180 #endif 1181 #ifdef HAS_UYVYTOUVROW_MSA 1182 ANY12S(UYVYToUVRow_Any_MSA, UYVYToUVRow_MSA, 1, 4, 31) 1183 #endif 1184 #undef ANY12S 1185 1186 #ifdef __cplusplus 1187 } // extern "C" 1188 } // namespace libyuv 1189 #endif 1190