diff options
| author | 3gg <3gg@shellblade.net> | 2025-12-27 12:03:39 -0800 |
|---|---|---|
| committer | 3gg <3gg@shellblade.net> | 2025-12-27 12:03:39 -0800 |
| commit | 5a079a2d114f96d4847d1ee305d5b7c16eeec50e (patch) | |
| tree | 8926ab44f168acf787d8e19608857b3af0f82758 /contrib/SDL-3.2.8/src/video/yuv2rgb/yuv_rgb_sse.c | |
Initial commit
Diffstat (limited to 'contrib/SDL-3.2.8/src/video/yuv2rgb/yuv_rgb_sse.c')
| -rw-r--r-- | contrib/SDL-3.2.8/src/video/yuv2rgb/yuv_rgb_sse.c | 460 |
1 files changed, 460 insertions, 0 deletions
diff --git a/contrib/SDL-3.2.8/src/video/yuv2rgb/yuv_rgb_sse.c b/contrib/SDL-3.2.8/src/video/yuv2rgb/yuv_rgb_sse.c new file mode 100644 index 0000000..37fe7e4 --- /dev/null +++ b/contrib/SDL-3.2.8/src/video/yuv2rgb/yuv_rgb_sse.c | |||
| @@ -0,0 +1,460 @@ | |||
| 1 | // Copyright 2016 Adrien Descamps | ||
| 2 | // Distributed under BSD 3-Clause License | ||
| 3 | #include "SDL_internal.h" | ||
| 4 | |||
| 5 | #ifdef SDL_HAVE_YUV | ||
| 6 | #include "yuv_rgb_internal.h" | ||
| 7 | |||
| 8 | #ifdef SDL_SSE2_INTRINSICS | ||
| 9 | |||
| 10 | /* SDL doesn't use these atm and compiling them adds seconds onto the build. --ryan. | ||
| 11 | #define SSE_FUNCTION_NAME yuv420_rgb565_sse | ||
| 12 | #define STD_FUNCTION_NAME yuv420_rgb565_std | ||
| 13 | #define YUV_FORMAT YUV_FORMAT_420 | ||
| 14 | #define RGB_FORMAT RGB_FORMAT_RGB565 | ||
| 15 | #define SSE_ALIGNED | ||
| 16 | #include "yuv_rgb_sse_func.h" | ||
| 17 | |||
| 18 | #define SSE_FUNCTION_NAME yuv420_rgb24_sse | ||
| 19 | #define STD_FUNCTION_NAME yuv420_rgb24_std | ||
| 20 | #define YUV_FORMAT YUV_FORMAT_420 | ||
| 21 | #define RGB_FORMAT RGB_FORMAT_RGB24 | ||
| 22 | #define SSE_ALIGNED | ||
| 23 | #include "yuv_rgb_sse_func.h" | ||
| 24 | |||
| 25 | #define SSE_FUNCTION_NAME yuv420_rgba_sse | ||
| 26 | #define STD_FUNCTION_NAME yuv420_rgba_std | ||
| 27 | #define YUV_FORMAT YUV_FORMAT_420 | ||
| 28 | #define RGB_FORMAT RGB_FORMAT_RGBA | ||
| 29 | #define SSE_ALIGNED | ||
| 30 | #include "yuv_rgb_sse_func.h" | ||
| 31 | |||
| 32 | #define SSE_FUNCTION_NAME yuv420_bgra_sse | ||
| 33 | #define STD_FUNCTION_NAME yuv420_bgra_std | ||
| 34 | #define YUV_FORMAT YUV_FORMAT_420 | ||
| 35 | #define RGB_FORMAT RGB_FORMAT_BGRA | ||
| 36 | #define SSE_ALIGNED | ||
| 37 | #include "yuv_rgb_sse_func.h" | ||
| 38 | |||
| 39 | #define SSE_FUNCTION_NAME yuv420_argb_sse | ||
| 40 | #define STD_FUNCTION_NAME yuv420_argb_std | ||
| 41 | #define YUV_FORMAT YUV_FORMAT_420 | ||
| 42 | #define RGB_FORMAT RGB_FORMAT_ARGB | ||
| 43 | #define SSE_ALIGNED | ||
| 44 | #include "yuv_rgb_sse_func.h" | ||
| 45 | |||
| 46 | #define SSE_FUNCTION_NAME yuv420_abgr_sse | ||
| 47 | #define STD_FUNCTION_NAME yuv420_abgr_std | ||
| 48 | #define YUV_FORMAT YUV_FORMAT_420 | ||
| 49 | #define RGB_FORMAT RGB_FORMAT_ABGR | ||
| 50 | #define SSE_ALIGNED | ||
| 51 | #include "yuv_rgb_sse_func.h" | ||
| 52 | |||
| 53 | #define SSE_FUNCTION_NAME yuv422_rgb565_sse | ||
| 54 | #define STD_FUNCTION_NAME yuv422_rgb565_std | ||
| 55 | #define YUV_FORMAT YUV_FORMAT_422 | ||
| 56 | #define RGB_FORMAT RGB_FORMAT_RGB565 | ||
| 57 | #define SSE_ALIGNED | ||
| 58 | #include "yuv_rgb_sse_func.h" | ||
| 59 | |||
| 60 | #define SSE_FUNCTION_NAME yuv422_rgb24_sse | ||
| 61 | #define STD_FUNCTION_NAME yuv422_rgb24_std | ||
| 62 | #define YUV_FORMAT YUV_FORMAT_422 | ||
| 63 | #define RGB_FORMAT RGB_FORMAT_RGB24 | ||
| 64 | #define SSE_ALIGNED | ||
| 65 | #include "yuv_rgb_sse_func.h" | ||
| 66 | |||
| 67 | #define SSE_FUNCTION_NAME yuv422_rgba_sse | ||
| 68 | #define STD_FUNCTION_NAME yuv422_rgba_std | ||
| 69 | #define YUV_FORMAT YUV_FORMAT_422 | ||
| 70 | #define RGB_FORMAT RGB_FORMAT_RGBA | ||
| 71 | #define SSE_ALIGNED | ||
| 72 | #include "yuv_rgb_sse_func.h" | ||
| 73 | |||
| 74 | #define SSE_FUNCTION_NAME yuv422_bgra_sse | ||
| 75 | #define STD_FUNCTION_NAME yuv422_bgra_std | ||
| 76 | #define YUV_FORMAT YUV_FORMAT_422 | ||
| 77 | #define RGB_FORMAT RGB_FORMAT_BGRA | ||
| 78 | #define SSE_ALIGNED | ||
| 79 | #include "yuv_rgb_sse_func.h" | ||
| 80 | |||
| 81 | #define SSE_FUNCTION_NAME yuv422_argb_sse | ||
| 82 | #define STD_FUNCTION_NAME yuv422_argb_std | ||
| 83 | #define YUV_FORMAT YUV_FORMAT_422 | ||
| 84 | #define RGB_FORMAT RGB_FORMAT_ARGB | ||
| 85 | #define SSE_ALIGNED | ||
| 86 | #include "yuv_rgb_sse_func.h" | ||
| 87 | |||
| 88 | #define SSE_FUNCTION_NAME yuv422_abgr_sse | ||
| 89 | #define STD_FUNCTION_NAME yuv422_abgr_std | ||
| 90 | #define YUV_FORMAT YUV_FORMAT_422 | ||
| 91 | #define RGB_FORMAT RGB_FORMAT_ABGR | ||
| 92 | #define SSE_ALIGNED | ||
| 93 | #include "yuv_rgb_sse_func.h" | ||
| 94 | |||
| 95 | #define SSE_FUNCTION_NAME yuvnv12_rgb565_sse | ||
| 96 | #define STD_FUNCTION_NAME yuvnv12_rgb565_std | ||
| 97 | #define YUV_FORMAT YUV_FORMAT_NV12 | ||
| 98 | #define RGB_FORMAT RGB_FORMAT_RGB565 | ||
| 99 | #define SSE_ALIGNED | ||
| 100 | #include "yuv_rgb_sse_func.h" | ||
| 101 | |||
| 102 | #define SSE_FUNCTION_NAME yuvnv12_rgb24_sse | ||
| 103 | #define STD_FUNCTION_NAME yuvnv12_rgb24_std | ||
| 104 | #define YUV_FORMAT YUV_FORMAT_NV12 | ||
| 105 | #define RGB_FORMAT RGB_FORMAT_RGB24 | ||
| 106 | #define SSE_ALIGNED | ||
| 107 | #include "yuv_rgb_sse_func.h" | ||
| 108 | |||
| 109 | #define SSE_FUNCTION_NAME yuvnv12_rgba_sse | ||
| 110 | #define STD_FUNCTION_NAME yuvnv12_rgba_std | ||
| 111 | #define YUV_FORMAT YUV_FORMAT_NV12 | ||
| 112 | #define RGB_FORMAT RGB_FORMAT_RGBA | ||
| 113 | #define SSE_ALIGNED | ||
| 114 | #include "yuv_rgb_sse_func.h" | ||
| 115 | |||
| 116 | #define SSE_FUNCTION_NAME yuvnv12_bgra_sse | ||
| 117 | #define STD_FUNCTION_NAME yuvnv12_bgra_std | ||
| 118 | #define YUV_FORMAT YUV_FORMAT_NV12 | ||
| 119 | #define RGB_FORMAT RGB_FORMAT_BGRA | ||
| 120 | #define SSE_ALIGNED | ||
| 121 | #include "yuv_rgb_sse_func.h" | ||
| 122 | |||
| 123 | #define SSE_FUNCTION_NAME yuvnv12_argb_sse | ||
| 124 | #define STD_FUNCTION_NAME yuvnv12_argb_std | ||
| 125 | #define YUV_FORMAT YUV_FORMAT_NV12 | ||
| 126 | #define RGB_FORMAT RGB_FORMAT_ARGB | ||
| 127 | #define SSE_ALIGNED | ||
| 128 | #include "yuv_rgb_sse_func.h" | ||
| 129 | |||
| 130 | #define SSE_FUNCTION_NAME yuvnv12_abgr_sse | ||
| 131 | #define STD_FUNCTION_NAME yuvnv12_abgr_std | ||
| 132 | #define YUV_FORMAT YUV_FORMAT_NV12 | ||
| 133 | #define RGB_FORMAT RGB_FORMAT_ABGR | ||
| 134 | #define SSE_ALIGNED | ||
| 135 | #include "yuv_rgb_sse_func.h" | ||
| 136 | */ | ||
| 137 | |||
| 138 | #define SSE_FUNCTION_NAME yuv420_rgb565_sseu | ||
| 139 | #define STD_FUNCTION_NAME yuv420_rgb565_std | ||
| 140 | #define YUV_FORMAT YUV_FORMAT_420 | ||
| 141 | #define RGB_FORMAT RGB_FORMAT_RGB565 | ||
| 142 | #include "yuv_rgb_sse_func.h" | ||
| 143 | |||
| 144 | #define SSE_FUNCTION_NAME yuv420_rgb24_sseu | ||
| 145 | #define STD_FUNCTION_NAME yuv420_rgb24_std | ||
| 146 | #define YUV_FORMAT YUV_FORMAT_420 | ||
| 147 | #define RGB_FORMAT RGB_FORMAT_RGB24 | ||
| 148 | #include "yuv_rgb_sse_func.h" | ||
| 149 | |||
| 150 | #define SSE_FUNCTION_NAME yuv420_rgba_sseu | ||
| 151 | #define STD_FUNCTION_NAME yuv420_rgba_std | ||
| 152 | #define YUV_FORMAT YUV_FORMAT_420 | ||
| 153 | #define RGB_FORMAT RGB_FORMAT_RGBA | ||
| 154 | #include "yuv_rgb_sse_func.h" | ||
| 155 | |||
| 156 | #define SSE_FUNCTION_NAME yuv420_bgra_sseu | ||
| 157 | #define STD_FUNCTION_NAME yuv420_bgra_std | ||
| 158 | #define YUV_FORMAT YUV_FORMAT_420 | ||
| 159 | #define RGB_FORMAT RGB_FORMAT_BGRA | ||
| 160 | #include "yuv_rgb_sse_func.h" | ||
| 161 | |||
| 162 | #define SSE_FUNCTION_NAME yuv420_argb_sseu | ||
| 163 | #define STD_FUNCTION_NAME yuv420_argb_std | ||
| 164 | #define YUV_FORMAT YUV_FORMAT_420 | ||
| 165 | #define RGB_FORMAT RGB_FORMAT_ARGB | ||
| 166 | #include "yuv_rgb_sse_func.h" | ||
| 167 | |||
| 168 | #define SSE_FUNCTION_NAME yuv420_abgr_sseu | ||
| 169 | #define STD_FUNCTION_NAME yuv420_abgr_std | ||
| 170 | #define YUV_FORMAT YUV_FORMAT_420 | ||
| 171 | #define RGB_FORMAT RGB_FORMAT_ABGR | ||
| 172 | #include "yuv_rgb_sse_func.h" | ||
| 173 | |||
| 174 | #define SSE_FUNCTION_NAME yuv422_rgb565_sseu | ||
| 175 | #define STD_FUNCTION_NAME yuv422_rgb565_std | ||
| 176 | #define YUV_FORMAT YUV_FORMAT_422 | ||
| 177 | #define RGB_FORMAT RGB_FORMAT_RGB565 | ||
| 178 | #include "yuv_rgb_sse_func.h" | ||
| 179 | |||
| 180 | #define SSE_FUNCTION_NAME yuv422_rgb24_sseu | ||
| 181 | #define STD_FUNCTION_NAME yuv422_rgb24_std | ||
| 182 | #define YUV_FORMAT YUV_FORMAT_422 | ||
| 183 | #define RGB_FORMAT RGB_FORMAT_RGB24 | ||
| 184 | #include "yuv_rgb_sse_func.h" | ||
| 185 | |||
| 186 | #define SSE_FUNCTION_NAME yuv422_rgba_sseu | ||
| 187 | #define STD_FUNCTION_NAME yuv422_rgba_std | ||
| 188 | #define YUV_FORMAT YUV_FORMAT_422 | ||
| 189 | #define RGB_FORMAT RGB_FORMAT_RGBA | ||
| 190 | #include "yuv_rgb_sse_func.h" | ||
| 191 | |||
| 192 | #define SSE_FUNCTION_NAME yuv422_bgra_sseu | ||
| 193 | #define STD_FUNCTION_NAME yuv422_bgra_std | ||
| 194 | #define YUV_FORMAT YUV_FORMAT_422 | ||
| 195 | #define RGB_FORMAT RGB_FORMAT_BGRA | ||
| 196 | #include "yuv_rgb_sse_func.h" | ||
| 197 | |||
| 198 | #define SSE_FUNCTION_NAME yuv422_argb_sseu | ||
| 199 | #define STD_FUNCTION_NAME yuv422_argb_std | ||
| 200 | #define YUV_FORMAT YUV_FORMAT_422 | ||
| 201 | #define RGB_FORMAT RGB_FORMAT_ARGB | ||
| 202 | #include "yuv_rgb_sse_func.h" | ||
| 203 | |||
| 204 | #define SSE_FUNCTION_NAME yuv422_abgr_sseu | ||
| 205 | #define STD_FUNCTION_NAME yuv422_abgr_std | ||
| 206 | #define YUV_FORMAT YUV_FORMAT_422 | ||
| 207 | #define RGB_FORMAT RGB_FORMAT_ABGR | ||
| 208 | #include "yuv_rgb_sse_func.h" | ||
| 209 | |||
| 210 | #define SSE_FUNCTION_NAME yuvnv12_rgb565_sseu | ||
| 211 | #define STD_FUNCTION_NAME yuvnv12_rgb565_std | ||
| 212 | #define YUV_FORMAT YUV_FORMAT_NV12 | ||
| 213 | #define RGB_FORMAT RGB_FORMAT_RGB565 | ||
| 214 | #include "yuv_rgb_sse_func.h" | ||
| 215 | |||
| 216 | #define SSE_FUNCTION_NAME yuvnv12_rgb24_sseu | ||
| 217 | #define STD_FUNCTION_NAME yuvnv12_rgb24_std | ||
| 218 | #define YUV_FORMAT YUV_FORMAT_NV12 | ||
| 219 | #define RGB_FORMAT RGB_FORMAT_RGB24 | ||
| 220 | #include "yuv_rgb_sse_func.h" | ||
| 221 | |||
| 222 | #define SSE_FUNCTION_NAME yuvnv12_rgba_sseu | ||
| 223 | #define STD_FUNCTION_NAME yuvnv12_rgba_std | ||
| 224 | #define YUV_FORMAT YUV_FORMAT_NV12 | ||
| 225 | #define RGB_FORMAT RGB_FORMAT_RGBA | ||
| 226 | #include "yuv_rgb_sse_func.h" | ||
| 227 | |||
| 228 | #define SSE_FUNCTION_NAME yuvnv12_bgra_sseu | ||
| 229 | #define STD_FUNCTION_NAME yuvnv12_bgra_std | ||
| 230 | #define YUV_FORMAT YUV_FORMAT_NV12 | ||
| 231 | #define RGB_FORMAT RGB_FORMAT_BGRA | ||
| 232 | #include "yuv_rgb_sse_func.h" | ||
| 233 | |||
| 234 | #define SSE_FUNCTION_NAME yuvnv12_argb_sseu | ||
| 235 | #define STD_FUNCTION_NAME yuvnv12_argb_std | ||
| 236 | #define YUV_FORMAT YUV_FORMAT_NV12 | ||
| 237 | #define RGB_FORMAT RGB_FORMAT_ARGB | ||
| 238 | #include "yuv_rgb_sse_func.h" | ||
| 239 | |||
| 240 | #define SSE_FUNCTION_NAME yuvnv12_abgr_sseu | ||
| 241 | #define STD_FUNCTION_NAME yuvnv12_abgr_std | ||
| 242 | #define YUV_FORMAT YUV_FORMAT_NV12 | ||
| 243 | #define RGB_FORMAT RGB_FORMAT_ABGR | ||
| 244 | #include "yuv_rgb_sse_func.h" | ||
| 245 | |||
| 246 | |||
| 247 | /* SDL doesn't use these atm and compiling them adds seconds onto the build. --ryan. | ||
| 248 | #define UNPACK_RGB24_32_STEP1(RGB1, RGB2, RGB3, RGB4, RGB5, RGB6, R1, R2, G1, G2, B1, B2) \ | ||
| 249 | R1 = _mm_unpacklo_epi8(RGB1, RGB4); \ | ||
| 250 | R2 = _mm_unpackhi_epi8(RGB1, RGB4); \ | ||
| 251 | G1 = _mm_unpacklo_epi8(RGB2, RGB5); \ | ||
| 252 | G2 = _mm_unpackhi_epi8(RGB2, RGB5); \ | ||
| 253 | B1 = _mm_unpacklo_epi8(RGB3, RGB6); \ | ||
| 254 | B2 = _mm_unpackhi_epi8(RGB3, RGB6); | ||
| 255 | |||
| 256 | #define UNPACK_RGB24_32_STEP2(RGB1, RGB2, RGB3, RGB4, RGB5, RGB6, R1, R2, G1, G2, B1, B2) \ | ||
| 257 | RGB1 = _mm_unpacklo_epi8(R1, G2); \ | ||
| 258 | RGB2 = _mm_unpackhi_epi8(R1, G2); \ | ||
| 259 | RGB3 = _mm_unpacklo_epi8(R2, B1); \ | ||
| 260 | RGB4 = _mm_unpackhi_epi8(R2, B1); \ | ||
| 261 | RGB5 = _mm_unpacklo_epi8(G1, B2); \ | ||
| 262 | RGB6 = _mm_unpackhi_epi8(G1, B2); \ | ||
| 263 | |||
| 264 | #define UNPACK_RGB24_32(RGB1, RGB2, RGB3, RGB4, RGB5, RGB6, R1, R2, G1, G2, B1, B2) \ | ||
| 265 | UNPACK_RGB24_32_STEP1(RGB1, RGB2, RGB3, RGB4, RGB5, RGB6, R1, R2, G1, G2, B1, B2) \ | ||
| 266 | UNPACK_RGB24_32_STEP2(RGB1, RGB2, RGB3, RGB4, RGB5, RGB6, R1, R2, G1, G2, B1, B2) \ | ||
| 267 | UNPACK_RGB24_32_STEP1(RGB1, RGB2, RGB3, RGB4, RGB5, RGB6, R1, R2, G1, G2, B1, B2) \ | ||
| 268 | UNPACK_RGB24_32_STEP2(RGB1, RGB2, RGB3, RGB4, RGB5, RGB6, R1, R2, G1, G2, B1, B2) \ | ||
| 269 | UNPACK_RGB24_32_STEP1(RGB1, RGB2, RGB3, RGB4, RGB5, RGB6, R1, R2, G1, G2, B1, B2) \ | ||
| 270 | |||
| 271 | #define RGB2YUV_16(R, G, B, Y, U, V) \ | ||
| 272 | Y = _mm_add_epi16(_mm_mullo_epi16(R, _mm_set1_epi16(param->matrix[0][0])), \ | ||
| 273 | _mm_mullo_epi16(G, _mm_set1_epi16(param->matrix[0][1]))); \ | ||
| 274 | Y = _mm_add_epi16(Y, _mm_mullo_epi16(B, _mm_set1_epi16(param->matrix[0][2]))); \ | ||
| 275 | Y = _mm_add_epi16(Y, _mm_set1_epi16((param->y_shift)<<PRECISION)); \ | ||
| 276 | Y = _mm_srai_epi16(Y, PRECISION); \ | ||
| 277 | U = _mm_add_epi16(_mm_mullo_epi16(R, _mm_set1_epi16(param->matrix[1][0])), \ | ||
| 278 | _mm_mullo_epi16(G, _mm_set1_epi16(param->matrix[1][1]))); \ | ||
| 279 | U = _mm_add_epi16(U, _mm_mullo_epi16(B, _mm_set1_epi16(param->matrix[1][2]))); \ | ||
| 280 | U = _mm_add_epi16(U, _mm_set1_epi16(128<<PRECISION)); \ | ||
| 281 | U = _mm_srai_epi16(U, PRECISION); \ | ||
| 282 | V = _mm_add_epi16(_mm_mullo_epi16(R, _mm_set1_epi16(param->matrix[2][0])), \ | ||
| 283 | _mm_mullo_epi16(G, _mm_set1_epi16(param->matrix[2][1]))); \ | ||
| 284 | V = _mm_add_epi16(V, _mm_mullo_epi16(B, _mm_set1_epi16(param->matrix[2][2]))); \ | ||
| 285 | V = _mm_add_epi16(V, _mm_set1_epi16(128<<PRECISION)); \ | ||
| 286 | V = _mm_srai_epi16(V, PRECISION); | ||
| 287 | */ | ||
| 288 | |||
| 289 | #if 0 // SDL doesn't use these atm and compiling them adds seconds onto the build. --ryan. | ||
| 290 | #define RGB2YUV_32 \ | ||
| 291 | __m128i r1, r2, b1, b2, g1, g2; \ | ||
| 292 | __m128i r_16, g_16, b_16; \ | ||
| 293 | __m128i y1_16, y2_16, u1_16, u2_16, v1_16, v2_16, y, u1, u2, v1, v2, u1_tmp, u2_tmp, v1_tmp, v2_tmp; \ | ||
| 294 | __m128i rgb1 = LOAD_SI128((const __m128i*)(rgb_ptr1)), \ | ||
| 295 | rgb2 = LOAD_SI128((const __m128i*)(rgb_ptr1+16)), \ | ||
| 296 | rgb3 = LOAD_SI128((const __m128i*)(rgb_ptr1+32)), \ | ||
| 297 | rgb4 = LOAD_SI128((const __m128i*)(rgb_ptr2)), \ | ||
| 298 | rgb5 = LOAD_SI128((const __m128i*)(rgb_ptr2+16)), \ | ||
| 299 | rgb6 = LOAD_SI128((const __m128i*)(rgb_ptr2+32)); \ | ||
| 300 | /* unpack rgb24 data to r, g and b data in separate channels*/ \ | ||
| 301 | UNPACK_RGB24_32(rgb1, rgb2, rgb3, rgb4, rgb5, rgb6, r1, r2, g1, g2, b1, b2) \ | ||
| 302 | /* process pixels of first line */ \ | ||
| 303 | r_16 = _mm_unpacklo_epi8(r1, _mm_setzero_si128()); \ | ||
| 304 | g_16 = _mm_unpacklo_epi8(g1, _mm_setzero_si128()); \ | ||
| 305 | b_16 = _mm_unpacklo_epi8(b1, _mm_setzero_si128()); \ | ||
| 306 | RGB2YUV_16(r_16, g_16, b_16, y1_16, u1_16, v1_16) \ | ||
| 307 | r_16 = _mm_unpackhi_epi8(r1, _mm_setzero_si128()); \ | ||
| 308 | g_16 = _mm_unpackhi_epi8(g1, _mm_setzero_si128()); \ | ||
| 309 | b_16 = _mm_unpackhi_epi8(b1, _mm_setzero_si128()); \ | ||
| 310 | RGB2YUV_16(r_16, g_16, b_16, y2_16, u2_16, v2_16) \ | ||
| 311 | y = _mm_packus_epi16(y1_16, y2_16); \ | ||
| 312 | u1 = _mm_packus_epi16(u1_16, u2_16); \ | ||
| 313 | v1 = _mm_packus_epi16(v1_16, v2_16); \ | ||
| 314 | /* save Y values */ \ | ||
| 315 | SAVE_SI128((__m128i*)(y_ptr1), y); \ | ||
| 316 | /* process pixels of second line */ \ | ||
| 317 | r_16 = _mm_unpacklo_epi8(r2, _mm_setzero_si128()); \ | ||
| 318 | g_16 = _mm_unpacklo_epi8(g2, _mm_setzero_si128()); \ | ||
| 319 | b_16 = _mm_unpacklo_epi8(b2, _mm_setzero_si128()); \ | ||
| 320 | RGB2YUV_16(r_16, g_16, b_16, y1_16, u1_16, v1_16) \ | ||
| 321 | r_16 = _mm_unpackhi_epi8(r2, _mm_setzero_si128()); \ | ||
| 322 | g_16 = _mm_unpackhi_epi8(g2, _mm_setzero_si128()); \ | ||
| 323 | b_16 = _mm_unpackhi_epi8(b2, _mm_setzero_si128()); \ | ||
| 324 | RGB2YUV_16(r_16, g_16, b_16, y2_16, u2_16, v2_16) \ | ||
| 325 | y = _mm_packus_epi16(y1_16, y2_16); \ | ||
| 326 | u2 = _mm_packus_epi16(u1_16, u2_16); \ | ||
| 327 | v2 = _mm_packus_epi16(v1_16, v2_16); \ | ||
| 328 | /* save Y values */ \ | ||
| 329 | SAVE_SI128((__m128i*)(y_ptr2), y); \ | ||
| 330 | /* vertical subsampling of u/v values */ \ | ||
| 331 | u1_tmp = _mm_avg_epu8(u1, u2); \ | ||
| 332 | v1_tmp = _mm_avg_epu8(v1, v2); \ | ||
| 333 | /* do the same again with next data */ \ | ||
| 334 | rgb1 = LOAD_SI128((const __m128i*)(rgb_ptr1+48)); \ | ||
| 335 | rgb2 = LOAD_SI128((const __m128i*)(rgb_ptr1+64)); \ | ||
| 336 | rgb3 = LOAD_SI128((const __m128i*)(rgb_ptr1+80)); \ | ||
| 337 | rgb4 = LOAD_SI128((const __m128i*)(rgb_ptr2+48)); \ | ||
| 338 | rgb5 = LOAD_SI128((const __m128i*)(rgb_ptr2+64)); \ | ||
| 339 | rgb6 = LOAD_SI128((const __m128i*)(rgb_ptr2+80)); \ | ||
| 340 | /* unpack rgb24 data to r, g and b data in separate channels*/ \ | ||
| 341 | UNPACK_RGB24_32(rgb1, rgb2, rgb3, rgb4, rgb5, rgb6, r1, r2, g1, g2, b1, b2) \ | ||
| 342 | /* process pixels of first line */ \ | ||
| 343 | r_16 = _mm_unpacklo_epi8(r1, _mm_setzero_si128()); \ | ||
| 344 | g_16 = _mm_unpacklo_epi8(g1, _mm_setzero_si128()); \ | ||
| 345 | b_16 = _mm_unpacklo_epi8(b1, _mm_setzero_si128()); \ | ||
| 346 | RGB2YUV_16(r_16, g_16, b_16, y1_16, u1_16, v1_16) \ | ||
| 347 | r_16 = _mm_unpackhi_epi8(r1, _mm_setzero_si128()); \ | ||
| 348 | g_16 = _mm_unpackhi_epi8(g1, _mm_setzero_si128()); \ | ||
| 349 | b_16 = _mm_unpackhi_epi8(b1, _mm_setzero_si128()); \ | ||
| 350 | RGB2YUV_16(r_16, g_16, b_16, y2_16, u2_16, v2_16) \ | ||
| 351 | y = _mm_packus_epi16(y1_16, y2_16); \ | ||
| 352 | u1 = _mm_packus_epi16(u1_16, u2_16); \ | ||
| 353 | v1 = _mm_packus_epi16(v1_16, v2_16); \ | ||
| 354 | /* save Y values */ \ | ||
| 355 | SAVE_SI128((__m128i*)(y_ptr1+16), y); \ | ||
| 356 | /* process pixels of second line */ \ | ||
| 357 | r_16 = _mm_unpacklo_epi8(r2, _mm_setzero_si128()); \ | ||
| 358 | g_16 = _mm_unpacklo_epi8(g2, _mm_setzero_si128()); \ | ||
| 359 | b_16 = _mm_unpacklo_epi8(b2, _mm_setzero_si128()); \ | ||
| 360 | RGB2YUV_16(r_16, g_16, b_16, y1_16, u1_16, v1_16) \ | ||
| 361 | r_16 = _mm_unpackhi_epi8(r2, _mm_setzero_si128()); \ | ||
| 362 | g_16 = _mm_unpackhi_epi8(g2, _mm_setzero_si128()); \ | ||
| 363 | b_16 = _mm_unpackhi_epi8(b2, _mm_setzero_si128()); \ | ||
| 364 | RGB2YUV_16(r_16, g_16, b_16, y2_16, u2_16, v2_16) \ | ||
| 365 | y = _mm_packus_epi16(y1_16, y2_16); \ | ||
| 366 | u2 = _mm_packus_epi16(u1_16, u2_16); \ | ||
| 367 | v2 = _mm_packus_epi16(v1_16, v2_16); \ | ||
| 368 | /* save Y values */ \ | ||
| 369 | SAVE_SI128((__m128i*)(y_ptr2+16), y); \ | ||
| 370 | /* vertical subsampling of u/v values */ \ | ||
| 371 | u2_tmp = _mm_avg_epu8(u1, u2); \ | ||
| 372 | v2_tmp = _mm_avg_epu8(v1, v2); \ | ||
| 373 | /* horizontal subsampling of u/v values */ \ | ||
| 374 | u1 = _mm_packus_epi16(_mm_srl_epi16(u1_tmp, _mm_cvtsi32_si128(8)), _mm_srl_epi16(u2_tmp, _mm_cvtsi32_si128(8))); \ | ||
| 375 | v1 = _mm_packus_epi16(_mm_srl_epi16(v1_tmp, _mm_cvtsi32_si128(8)), _mm_srl_epi16(v2_tmp, _mm_cvtsi32_si128(8))); \ | ||
| 376 | u2 = _mm_packus_epi16(_mm_and_si128(u1_tmp, _mm_set1_epi16(0xFF)), _mm_and_si128(u2_tmp, _mm_set1_epi16(0xFF))); \ | ||
| 377 | v2 = _mm_packus_epi16(_mm_and_si128(v1_tmp, _mm_set1_epi16(0xFF)), _mm_and_si128(v2_tmp, _mm_set1_epi16(0xFF))); \ | ||
| 378 | u1 = _mm_avg_epu8(u1, u2); \ | ||
| 379 | v1 = _mm_avg_epu8(v1, v2); \ | ||
| 380 | SAVE_SI128((__m128i*)(u_ptr), u1); \ | ||
| 381 | SAVE_SI128((__m128i*)(v_ptr), v1); | ||
| 382 | #endif | ||
| 383 | |||
| 384 | /* SDL doesn't use these atm and compiling them adds seconds onto the build. --ryan. | ||
| 385 | void SDL_TARGETING("sse2") rgb24_yuv420_sse(uint32_t width, uint32_t height, | ||
| 386 | const uint8_t *RGB, uint32_t RGB_stride, | ||
| 387 | uint8_t *Y, uint8_t *U, uint8_t *V, uint32_t Y_stride, uint32_t UV_stride, | ||
| 388 | YCbCrType yuv_type) | ||
| 389 | { | ||
| 390 | #define LOAD_SI128 _mm_load_si128 | ||
| 391 | #define SAVE_SI128 _mm_stream_si128 | ||
| 392 | const RGB2YUVParam *const param = &(RGB2YUV[yuv_type]); | ||
| 393 | |||
| 394 | uint32_t xpos, ypos; | ||
| 395 | for(ypos=0; ypos<(height-1); ypos+=2) | ||
| 396 | { | ||
| 397 | const uint8_t *rgb_ptr1=RGB+ypos*RGB_stride, | ||
| 398 | *rgb_ptr2=RGB+(ypos+1)*RGB_stride; | ||
| 399 | |||
| 400 | uint8_t *y_ptr1=Y+ypos*Y_stride, | ||
| 401 | *y_ptr2=Y+(ypos+1)*Y_stride, | ||
| 402 | *u_ptr=U+(ypos/2)*UV_stride, | ||
| 403 | *v_ptr=V+(ypos/2)*UV_stride; | ||
| 404 | |||
| 405 | for(xpos=0; xpos<(width-31); xpos+=32) | ||
| 406 | { | ||
| 407 | RGB2YUV_32 | ||
| 408 | |||
| 409 | rgb_ptr1+=96; | ||
| 410 | rgb_ptr2+=96; | ||
| 411 | y_ptr1+=32; | ||
| 412 | y_ptr2+=32; | ||
| 413 | u_ptr+=16; | ||
| 414 | v_ptr+=16; | ||
| 415 | } | ||
| 416 | } | ||
| 417 | #undef LOAD_SI128 | ||
| 418 | #undef SAVE_SI128 | ||
| 419 | } | ||
| 420 | |||
| 421 | void SDL_TARGETING("sse2") rgb24_yuv420_sseu(uint32_t width, uint32_t height, | ||
| 422 | const uint8_t *RGB, uint32_t RGB_stride, | ||
| 423 | uint8_t *Y, uint8_t *U, uint8_t *V, uint32_t Y_stride, uint32_t UV_stride, | ||
| 424 | YCbCrType yuv_type) | ||
| 425 | { | ||
| 426 | #define LOAD_SI128 _mm_loadu_si128 | ||
| 427 | #define SAVE_SI128 _mm_storeu_si128 | ||
| 428 | const RGB2YUVParam *const param = &(RGB2YUV[yuv_type]); | ||
| 429 | |||
| 430 | uint32_t xpos, ypos; | ||
| 431 | for(ypos=0; ypos<(height-1); ypos+=2) | ||
| 432 | { | ||
| 433 | const uint8_t *rgb_ptr1=RGB+ypos*RGB_stride, | ||
| 434 | *rgb_ptr2=RGB+(ypos+1)*RGB_stride; | ||
| 435 | |||
| 436 | uint8_t *y_ptr1=Y+ypos*Y_stride, | ||
| 437 | *y_ptr2=Y+(ypos+1)*Y_stride, | ||
| 438 | *u_ptr=U+(ypos/2)*UV_stride, | ||
| 439 | *v_ptr=V+(ypos/2)*UV_stride; | ||
| 440 | |||
| 441 | for(xpos=0; xpos<(width-31); xpos+=32) | ||
| 442 | { | ||
| 443 | RGB2YUV_32 | ||
| 444 | |||
| 445 | rgb_ptr1+=96; | ||
| 446 | rgb_ptr2+=96; | ||
| 447 | y_ptr1+=32; | ||
| 448 | y_ptr2+=32; | ||
| 449 | u_ptr+=16; | ||
| 450 | v_ptr+=16; | ||
| 451 | } | ||
| 452 | } | ||
| 453 | #undef LOAD_SI128 | ||
| 454 | #undef SAVE_SI128 | ||
| 455 | } | ||
| 456 | */ | ||
| 457 | |||
| 458 | #endif // SDL_SSE2_INTRINSICS | ||
| 459 | |||
| 460 | #endif // SDL_HAVE_YUV | ||
