1 /* 2 * Copyright (c) 2011 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 #ifndef LIBYUV_SOURCE_ROW_H_ 12 #define LIBYUV_SOURCE_ROW_H_ 13 14 #include "libyuv/basic_types.h" 15 16 // The following are available on all x86 platforms 17 #if (defined(WIN32) || defined(__x86_64__) || defined(__i386__)) \ 18 && !defined(COVERAGE_ENABLED) && !defined(TARGET_IPHONE_SIMULATOR) 19 #define HAS_ARGBTOYROW_SSSE3 20 #define HAS_BG24TOARGBROW_SSSE3 21 #define HAS_RAWTOARGBROW_SSSE3 22 #define HAS_RGB24TOYROW_SSSE3 23 #define HAS_RAWTOYROW_SSSE3 24 #define HAS_RGB24TOUVROW_SSSE3 25 #define HAS_RAWTOUVROW_SSSE3 26 #endif 27 28 // The following are available only on Windows 29 #if defined(WIN32) \ 30 && !defined(COVERAGE_ENABLED) && !defined(TARGET_IPHONE_SIMULATOR) 31 #define HAS_BGRATOYROW_SSSE3 32 #define HAS_ABGRTOYROW_SSSE3 33 #define HAS_ARGBTOUVROW_SSSE3 34 #define HAS_BGRATOUVROW_SSSE3 35 #define HAS_ABGRTOUVROW_SSSE3 36 #endif 37 38 extern "C" { 39 #ifdef HAS_ARGBTOYROW_SSSE3 40 void ARGBToYRow_SSSE3(const uint8* src_argb, uint8* dst_y, int pix); 41 void BGRAToYRow_SSSE3(const uint8* src_argb, uint8* dst_y, int pix); 42 void ABGRToYRow_SSSE3(const uint8* src_argb, uint8* dst_y, int pix); 43 void ARGBToUVRow_SSSE3(const uint8* src_argb0, int src_stride_argb, 44 uint8* dst_u, uint8* dst_v, int width); 45 void BGRAToUVRow_SSSE3(const uint8* src_argb0, int src_stride_argb, 46 uint8* dst_u, uint8* dst_v, int width); 47 void ABGRToUVRow_SSSE3(const uint8* src_argb0, int src_stride_argb, 48 uint8* dst_u, uint8* dst_v, int width); 49 #endif 50 #if defined(HAS_BG24TOARGBROW_SSSE3) && defined(HAS_ARGBTOYROW_SSSE3) 51 #define HASRGB24TOYROW_SSSE3 52 #endif 53 #ifdef HASRGB24TOYROW_SSSE3 54 void RGB24ToYRow_SSSE3(const uint8* src_argb, uint8* dst_y, int pix); 55 void RAWToYRow_SSSE3(const uint8* src_argb, uint8* dst_y, int pix); 56 void RGB24ToUVRow_SSSE3(const uint8* src_argb0, int src_stride_argb, 57 uint8* dst_u, uint8* dst_v, int width); 58 void RAWToUVRow_SSSE3(const uint8* src_argb0, int src_stride_argb, 59 uint8* dst_u, uint8* dst_v, int width); 60 #endif 61 void ARGBToYRow_C(const uint8* src_argb, uint8* dst_y, int pix); 62 void BGRAToYRow_C(const uint8* src_argb, uint8* dst_y, int pix); 63 void ABGRToYRow_C(const uint8* src_argb, uint8* dst_y, int pix); 64 void RGB24ToYRow_C(const uint8* src_argb, uint8* dst_y, int pix); 65 void RAWToYRow_C(const uint8* src_argb, uint8* dst_y, int pix); 66 void ARGBToUVRow_C(const uint8* src_argb0, int src_stride_argb, 67 uint8* dst_u, uint8* dst_v, int width); 68 void BGRAToUVRow_C(const uint8* src_argb0, int src_stride_argb, 69 uint8* dst_u, uint8* dst_v, int width); 70 void ABGRToUVRow_C(const uint8* src_argb0, int src_stride_argb, 71 uint8* dst_u, uint8* dst_v, int width); 72 void RGB24ToUVRow_C(const uint8* src_argb0, int src_stride_argb, 73 uint8* dst_u, uint8* dst_v, int width); 74 void RAWToUVRow_C(const uint8* src_argb0, int src_stride_argb, 75 uint8* dst_u, uint8* dst_v, int width); 76 77 #ifdef HAS_BG24TOARGBROW_SSSE3 78 void BG24ToARGBRow_SSSE3(const uint8* src_bg24, uint8* dst_argb, int pix); 79 void RAWToARGBRow_SSSE3(const uint8* src_bg24, uint8* dst_argb, int pix); 80 #endif 81 void BG24ToARGBRow_C(const uint8* src_bg24, uint8* dst_argb, int pix); 82 void RAWToARGBRow_C(const uint8* src_bg24, uint8* dst_argb, int pix); 83 84 #if defined(_MSC_VER) 85 #define SIMD_ALIGNED(var) __declspec(align(16)) var 86 #define TALIGN16(t, var) static __declspec(align(16)) t _ ## var 87 #else 88 #define SIMD_ALIGNED(var) var __attribute__((aligned(16))) 89 #define TALIGN16(t, var) t var __attribute__((aligned(16))) 90 #endif 91 92 #ifdef OSX 93 extern SIMD_ALIGNED(const int16 kCoefficientsRgbY[768][4]); 94 extern SIMD_ALIGNED(const int16 kCoefficientsBgraY[768][4]); 95 extern SIMD_ALIGNED(const int16 kCoefficientsAbgrY[768][4]); 96 #else 97 extern SIMD_ALIGNED(const int16 _kCoefficientsRgbY[768][4]); 98 extern SIMD_ALIGNED(const int16 _kCoefficientsBgraY[768][4]); 99 extern SIMD_ALIGNED(const int16 _kCoefficientsAbgrY[768][4]); 100 #endif 101 void FastConvertYUVToRGB32Row(const uint8* y_buf, 102 const uint8* u_buf, 103 const uint8* v_buf, 104 uint8* rgb_buf, 105 int width); 106 107 void FastConvertYUVToBGRARow(const uint8* y_buf, 108 const uint8* u_buf, 109 const uint8* v_buf, 110 uint8* rgb_buf, 111 int width); 112 113 void FastConvertYUVToABGRRow(const uint8* y_buf, 114 const uint8* u_buf, 115 const uint8* v_buf, 116 uint8* rgb_buf, 117 int width); 118 119 void FastConvertYUV444ToRGB32Row(const uint8* y_buf, 120 const uint8* u_buf, 121 const uint8* v_buf, 122 uint8* rgb_buf, 123 int width); 124 125 void FastConvertYToRGB32Row(const uint8* y_buf, 126 uint8* rgb_buf, 127 int width); 128 129 // Method to force C version. 130 //#define USE_MMX 0 131 //#define USE_SSE2 0 132 133 #if !defined(USE_MMX) 134 // Windows, Mac and Linux use MMX 135 #if defined(__i386__) || defined(_MSC_VER) 136 #define USE_MMX 1 137 #else 138 #define USE_MMX 0 139 #endif 140 #endif 141 142 #if !defined(USE_SSE2) 143 #if defined(__SSE2__) || defined(ARCH_CPU_X86_64) || _M_IX86_FP==2 144 #define USE_SSE2 1 145 #else 146 #define USE_SSE2 0 147 #endif 148 #endif 149 150 // x64 uses MMX2 (SSE) so emms is not required. 151 // Warning C4799: function has no EMMS instruction. 152 // EMMS() is slow and should be called by the calling function once per image. 153 #if USE_MMX && !defined(ARCH_CPU_X86_64) 154 #if defined(_MSC_VER) 155 #define EMMS() __asm emms 156 #pragma warning(disable: 4799) 157 #else 158 #define EMMS() asm("emms") 159 #endif 160 #else 161 #define EMMS() 162 #endif 163 164 165 } // extern "C" 166 167 #endif // LIBYUV_SOURCE_ROW_H_ 168