| OLD | NEW |
| 1 /* | 1 /* |
| 2 * Copyright 2012 The LibYuv Project Authors. All rights reserved. | 2 * Copyright 2012 The LibYuv Project Authors. All rights reserved. |
| 3 * | 3 * |
| 4 * Use of this source code is governed by a BSD-style license | 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 | 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 | 6 * tree. An additional intellectual property rights grant can be found |
| 7 * in the file PATENTS. All contributing project authors may | 7 * in the file PATENTS. All contributing project authors may |
| 8 * be found in the AUTHORS file in the root of the source tree. | 8 * be found in the AUTHORS file in the root of the source tree. |
| 9 */ | 9 */ |
| 10 | 10 |
| 11 #include "libyuv/row.h" | 11 #include "libyuv/row.h" |
| 12 | 12 |
| 13 #include <string.h> // For memset. | 13 #include <string.h> // For memset. |
| 14 | 14 |
| 15 #include "libyuv/basic_types.h" | 15 #include "libyuv/basic_types.h" |
| 16 | 16 |
| 17 #ifdef __cplusplus | 17 #ifdef __cplusplus |
| 18 namespace libyuv { | 18 namespace libyuv { |
| 19 extern "C" { | 19 extern "C" { |
| 20 #endif | 20 #endif |
| 21 | 21 |
| 22 // Subsampled source needs to be increase by 1 of not even. | 22 // Subsampled source needs to be increase by 1 of not even. |
| 23 #define SS(width, shift) (((width) + (1 << (shift)) - 1) >> (shift)) | 23 #define SS(width, shift) (((width) + (1 << (shift)) - 1) >> (shift)) |
| 24 | 24 |
| 25 // Any 4 planes to 1 with yuvconstants | 25 // Any 4 planes to 1 with yuvconstants |
| 26 #define ANY41C(NAMEANY, ANY_SIMD, UVSHIFT, DUVSHIFT, BPP, MASK) \ | 26 #define ANY41C(NAMEANY, ANY_SIMD, UVSHIFT, DUVSHIFT, BPP, MASK) \ |
| 27 void NAMEANY(const uint8* y_buf, const uint8* u_buf, const uint8* v_buf, \ | 27 void NAMEANY(const uint8* y_buf, const uint8* u_buf, const uint8* v_buf, \ |
| 28 const uint8* a_buf, uint8* dst_ptr, \ | 28 const uint8* a_buf, uint8* dst_ptr, \ |
| 29 struct YuvConstants* yuvconstants, int width) { \ | 29 const struct YuvConstants* yuvconstants, int width) { \ |
| 30 SIMD_ALIGNED(uint8 temp[64 * 5]); \ | 30 SIMD_ALIGNED(uint8 temp[64 * 5]); \ |
| 31 memset(temp, 0, 64 * 4); /* for msan */ \ | 31 memset(temp, 0, 64 * 4); /* for msan */ \ |
| 32 int r = width & MASK; \ | 32 int r = width & MASK; \ |
| 33 int n = width & ~MASK; \ | 33 int n = width & ~MASK; \ |
| 34 if (n > 0) { \ | 34 if (n > 0) { \ |
| 35 ANY_SIMD(y_buf, u_buf, v_buf, a_buf, dst_ptr, yuvconstants, n); \ | 35 ANY_SIMD(y_buf, u_buf, v_buf, a_buf, dst_ptr, yuvconstants, n); \ |
| 36 } \ | 36 } \ |
| 37 memcpy(temp, y_buf + n, r); \ | 37 memcpy(temp, y_buf + n, r); \ |
| 38 memcpy(temp + 64, u_buf + (n >> UVSHIFT), SS(r, UVSHIFT)); \ | 38 memcpy(temp + 64, u_buf + (n >> UVSHIFT), SS(r, UVSHIFT)); \ |
| 39 memcpy(temp + 128, v_buf + (n >> UVSHIFT), SS(r, UVSHIFT)); \ | 39 memcpy(temp + 128, v_buf + (n >> UVSHIFT), SS(r, UVSHIFT)); \ |
| (...skipping 40 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... |
| 80 ANY31(I422ToYUY2Row_Any_NEON, I422ToYUY2Row_NEON, 1, 1, 4, 15) | 80 ANY31(I422ToYUY2Row_Any_NEON, I422ToYUY2Row_NEON, 1, 1, 4, 15) |
| 81 #endif | 81 #endif |
| 82 #ifdef HAS_I422TOUYVYROW_NEON | 82 #ifdef HAS_I422TOUYVYROW_NEON |
| 83 ANY31(I422ToUYVYRow_Any_NEON, I422ToUYVYRow_NEON, 1, 1, 4, 15) | 83 ANY31(I422ToUYVYRow_Any_NEON, I422ToUYVYRow_NEON, 1, 1, 4, 15) |
| 84 #endif | 84 #endif |
| 85 #undef ANY31 | 85 #undef ANY31 |
| 86 | 86 |
| 87 // Any 3 planes to 1 with yuvconstants | 87 // Any 3 planes to 1 with yuvconstants |
| 88 #define ANY31C(NAMEANY, ANY_SIMD, UVSHIFT, DUVSHIFT, BPP, MASK) \ | 88 #define ANY31C(NAMEANY, ANY_SIMD, UVSHIFT, DUVSHIFT, BPP, MASK) \ |
| 89 void NAMEANY(const uint8* y_buf, const uint8* u_buf, const uint8* v_buf, \ | 89 void NAMEANY(const uint8* y_buf, const uint8* u_buf, const uint8* v_buf, \ |
| 90 uint8* dst_ptr, struct YuvConstants* yuvconstants, \ | 90 uint8* dst_ptr, const struct YuvConstants* yuvconstants, \ |
| 91 int width) { \ | 91 int width) { \ |
| 92 SIMD_ALIGNED(uint8 temp[64 * 4]); \ | 92 SIMD_ALIGNED(uint8 temp[64 * 4]); \ |
| 93 memset(temp, 0, 64 * 3); /* for YUY2 and msan */ \ | 93 memset(temp, 0, 64 * 3); /* for YUY2 and msan */ \ |
| 94 int r = width & MASK; \ | 94 int r = width & MASK; \ |
| 95 int n = width & ~MASK; \ | 95 int n = width & ~MASK; \ |
| 96 if (n > 0) { \ | 96 if (n > 0) { \ |
| 97 ANY_SIMD(y_buf, u_buf, v_buf, dst_ptr, yuvconstants, n); \ | 97 ANY_SIMD(y_buf, u_buf, v_buf, dst_ptr, yuvconstants, n); \ |
| 98 } \ | 98 } \ |
| 99 memcpy(temp, y_buf + n, r); \ | 99 memcpy(temp, y_buf + n, r); \ |
| 100 memcpy(temp + 64, u_buf + (n >> UVSHIFT), SS(r, UVSHIFT)); \ | 100 memcpy(temp + 64, u_buf + (n >> UVSHIFT), SS(r, UVSHIFT)); \ |
| (...skipping 146 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... |
| 247 ANY21(SobelXYRow_Any_SSE2, SobelXYRow_SSE2, 0, 1, 1, 4, 15) | 247 ANY21(SobelXYRow_Any_SSE2, SobelXYRow_SSE2, 0, 1, 1, 4, 15) |
| 248 #endif | 248 #endif |
| 249 #ifdef HAS_SOBELXYROW_NEON | 249 #ifdef HAS_SOBELXYROW_NEON |
| 250 ANY21(SobelXYRow_Any_NEON, SobelXYRow_NEON, 0, 1, 1, 4, 7) | 250 ANY21(SobelXYRow_Any_NEON, SobelXYRow_NEON, 0, 1, 1, 4, 7) |
| 251 #endif | 251 #endif |
| 252 #undef ANY21 | 252 #undef ANY21 |
| 253 | 253 |
| 254 // Any 2 planes to 1 with yuvconstants | 254 // Any 2 planes to 1 with yuvconstants |
| 255 #define ANY21C(NAMEANY, ANY_SIMD, UVSHIFT, SBPP, SBPP2, BPP, MASK) \ | 255 #define ANY21C(NAMEANY, ANY_SIMD, UVSHIFT, SBPP, SBPP2, BPP, MASK) \ |
| 256 void NAMEANY(const uint8* y_buf, const uint8* uv_buf, \ | 256 void NAMEANY(const uint8* y_buf, const uint8* uv_buf, \ |
| 257 uint8* dst_ptr, struct YuvConstants* yuvconstants, \ | 257 uint8* dst_ptr, const struct YuvConstants* yuvconstants, \ |
| 258 int width) { \ | 258 int width) { \ |
| 259 SIMD_ALIGNED(uint8 temp[64 * 3]); \ | 259 SIMD_ALIGNED(uint8 temp[64 * 3]); \ |
| 260 memset(temp, 0, 64 * 2); /* for msan */ \ | 260 memset(temp, 0, 64 * 2); /* for msan */ \ |
| 261 int r = width & MASK; \ | 261 int r = width & MASK; \ |
| 262 int n = width & ~MASK; \ | 262 int n = width & ~MASK; \ |
| 263 if (n > 0) { \ | 263 if (n > 0) { \ |
| 264 ANY_SIMD(y_buf, uv_buf, dst_ptr, yuvconstants, n); \ | 264 ANY_SIMD(y_buf, uv_buf, dst_ptr, yuvconstants, n); \ |
| 265 } \ | 265 } \ |
| 266 memcpy(temp, y_buf + n * SBPP, r * SBPP); \ | 266 memcpy(temp, y_buf + n * SBPP, r * SBPP); \ |
| 267 memcpy(temp + 64, uv_buf + (n >> UVSHIFT) * SBPP2, \ | 267 memcpy(temp + 64, uv_buf + (n >> UVSHIFT) * SBPP2, \ |
| (...skipping 194 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... |
| 462 ANY11(ARGBUnattenuateRow_Any_AVX2, ARGBUnattenuateRow_AVX2, 0, 4, 4, 7) | 462 ANY11(ARGBUnattenuateRow_Any_AVX2, ARGBUnattenuateRow_AVX2, 0, 4, 4, 7) |
| 463 #endif | 463 #endif |
| 464 #ifdef HAS_ARGBATTENUATEROW_NEON | 464 #ifdef HAS_ARGBATTENUATEROW_NEON |
| 465 ANY11(ARGBAttenuateRow_Any_NEON, ARGBAttenuateRow_NEON, 0, 4, 4, 7) | 465 ANY11(ARGBAttenuateRow_Any_NEON, ARGBAttenuateRow_NEON, 0, 4, 4, 7) |
| 466 #endif | 466 #endif |
| 467 #undef ANY11 | 467 #undef ANY11 |
| 468 | 468 |
| 469 // Any 1 to 1 with yuvconstants | 469 // Any 1 to 1 with yuvconstants |
| 470 #define ANY11C(NAMEANY, ANY_SIMD, UVSHIFT, SBPP, BPP, MASK) \ | 470 #define ANY11C(NAMEANY, ANY_SIMD, UVSHIFT, SBPP, BPP, MASK) \ |
| 471 void NAMEANY(const uint8* src_ptr, uint8* dst_ptr, \ | 471 void NAMEANY(const uint8* src_ptr, uint8* dst_ptr, \ |
| 472 struct YuvConstants* yuvconstants, int width) { \ | 472 const struct YuvConstants* yuvconstants, int width) { \ |
| 473 SIMD_ALIGNED(uint8 temp[128 * 2]); \ | 473 SIMD_ALIGNED(uint8 temp[128 * 2]); \ |
| 474 memset(temp, 0, 128); /* for YUY2 and msan */ \ | 474 memset(temp, 0, 128); /* for YUY2 and msan */ \ |
| 475 int r = width & MASK; \ | 475 int r = width & MASK; \ |
| 476 int n = width & ~MASK; \ | 476 int n = width & ~MASK; \ |
| 477 if (n > 0) { \ | 477 if (n > 0) { \ |
| 478 ANY_SIMD(src_ptr, dst_ptr, yuvconstants, n); \ | 478 ANY_SIMD(src_ptr, dst_ptr, yuvconstants, n); \ |
| 479 } \ | 479 } \ |
| 480 memcpy(temp, src_ptr + (n >> UVSHIFT) * SBPP, SS(r, UVSHIFT) * SBPP); \ | 480 memcpy(temp, src_ptr + (n >> UVSHIFT) * SBPP, SS(r, UVSHIFT) * SBPP); \ |
| 481 ANY_SIMD(temp, temp + 128, yuvconstants, MASK + 1); \ | 481 ANY_SIMD(temp, temp + 128, yuvconstants, MASK + 1); \ |
| 482 memcpy(dst_ptr + n * BPP, temp + 128, r * BPP); \ | 482 memcpy(dst_ptr + n * BPP, temp + 128, r * BPP); \ |
| (...skipping 312 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... |
| 795 #endif | 795 #endif |
| 796 #ifdef HAS_UYVYTOUVROW_NEON | 796 #ifdef HAS_UYVYTOUVROW_NEON |
| 797 ANY12S(UYVYToUVRow_Any_NEON, UYVYToUVRow_NEON, 1, 4, 15) | 797 ANY12S(UYVYToUVRow_Any_NEON, UYVYToUVRow_NEON, 1, 4, 15) |
| 798 #endif | 798 #endif |
| 799 #undef ANY12S | 799 #undef ANY12S |
| 800 | 800 |
| 801 #ifdef __cplusplus | 801 #ifdef __cplusplus |
| 802 } // extern "C" | 802 } // extern "C" |
| 803 } // namespace libyuv | 803 } // namespace libyuv |
| 804 #endif | 804 #endif |
| OLD | NEW |