| 1 | /* |
| 2 | * Copyright (C) 2001-2011 Michael Niedermayer <michaelni@gmx.at> |
| 3 | * |
| 4 | * This file is part of FFmpeg. |
| 5 | * |
| 6 | * FFmpeg is free software; you can redistribute it and/or |
| 7 | * modify it under the terms of the GNU Lesser General Public |
| 8 | * License as published by the Free Software Foundation; either |
| 9 | * version 2.1 of the License, or (at your option) any later version. |
| 10 | * |
| 11 | * FFmpeg is distributed in the hope that it will be useful, |
| 12 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 13 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
| 14 | * Lesser General Public License for more details. |
| 15 | * |
| 16 | * You should have received a copy of the GNU Lesser General Public |
| 17 | * License along with FFmpeg; if not, write to the Free Software |
| 18 | * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
| 19 | */ |
| 20 | |
| 21 | #include <stdint.h> |
| 22 | |
| 23 | #include "libavutil/x86/asm.h" |
| 24 | #include "libswscale/swscale_internal.h" |
| 25 | |
| 26 | #undef REAL_MOVNTQ |
| 27 | #undef MOVNTQ |
| 28 | #undef MOVNTQ2 |
| 29 | #undef PREFETCH |
| 30 | |
| 31 | |
| 32 | #if COMPILE_TEMPLATE_MMXEXT |
| 33 | #define REAL_MOVNTQ(a,b) "movntq " #a ", " #b " \n\t" |
| 34 | #define MOVNTQ2 "movntq " |
| 35 | #else |
| 36 | #define REAL_MOVNTQ(a,b) "movq " #a ", " #b " \n\t" |
| 37 | #define MOVNTQ2 "movq " |
| 38 | #endif |
| 39 | #define MOVNTQ(a,b) REAL_MOVNTQ(a,b) |
| 40 | |
| 41 | #if !COMPILE_TEMPLATE_MMXEXT |
| 42 | static av_always_inline void |
| 43 | dither_8to16(const uint8_t *srcDither, int rot) |
| 44 | { |
| 45 | if (rot) { |
| 46 | __asm__ volatile("pxor %%mm0, %%mm0\n\t" |
| 47 | "movq (%0), %%mm3\n\t" |
| 48 | "movq %%mm3, %%mm4\n\t" |
| 49 | "psrlq $24, %%mm3\n\t" |
| 50 | "psllq $40, %%mm4\n\t" |
| 51 | "por %%mm4, %%mm3\n\t" |
| 52 | "movq %%mm3, %%mm4\n\t" |
| 53 | "punpcklbw %%mm0, %%mm3\n\t" |
| 54 | "punpckhbw %%mm0, %%mm4\n\t" |
| 55 | :: "r"(srcDither) |
| 56 | ); |
| 57 | } else { |
| 58 | __asm__ volatile("pxor %%mm0, %%mm0\n\t" |
| 59 | "movq (%0), %%mm3\n\t" |
| 60 | "movq %%mm3, %%mm4\n\t" |
| 61 | "punpcklbw %%mm0, %%mm3\n\t" |
| 62 | "punpckhbw %%mm0, %%mm4\n\t" |
| 63 | :: "r"(srcDither) |
| 64 | ); |
| 65 | } |
| 66 | } |
| 67 | #endif |
| 68 | |
| 69 | static void RENAME(yuv2yuvX)(const int16_t *filter, int filterSize, |
| 70 | const int16_t **src, uint8_t *dest, int dstW, |
| 71 | const uint8_t *dither, int offset) |
| 72 | { |
| 73 | dither_8to16(dither, offset); |
| 74 | filterSize--; |
| 75 | __asm__ volatile( |
| 76 | "movd %0, %%mm1\n\t" |
| 77 | "punpcklwd %%mm1, %%mm1\n\t" |
| 78 | "punpckldq %%mm1, %%mm1\n\t" |
| 79 | "psllw $3, %%mm1\n\t" |
| 80 | "paddw %%mm1, %%mm3\n\t" |
| 81 | "paddw %%mm1, %%mm4\n\t" |
| 82 | "psraw $4, %%mm3\n\t" |
| 83 | "psraw $4, %%mm4\n\t" |
| 84 | ::"m"(filterSize) |
| 85 | ); |
| 86 | |
| 87 | __asm__ volatile(\ |
| 88 | "movq %%mm3, %%mm6\n\t" |
| 89 | "movq %%mm4, %%mm7\n\t" |
| 90 | "movl %3, %%ecx\n\t" |
| 91 | "mov %0, %%"REG_d" \n\t"\ |
| 92 | "mov (%%"REG_d"), %%"REG_S" \n\t"\ |
| 93 | ".p2align 4 \n\t" /* FIXME Unroll? */\ |
| 94 | "1: \n\t"\ |
| 95 | "movq 8(%%"REG_d"), %%mm0 \n\t" /* filterCoeff */\ |
| 96 | "movq (%%"REG_S", %%"REG_c", 2), %%mm2 \n\t" /* srcData */\ |
| 97 | "movq 8(%%"REG_S", %%"REG_c", 2), %%mm5 \n\t" /* srcData */\ |
| 98 | "add $16, %%"REG_d" \n\t"\ |
| 99 | "mov (%%"REG_d"), %%"REG_S" \n\t"\ |
| 100 | "test %%"REG_S", %%"REG_S" \n\t"\ |
| 101 | "pmulhw %%mm0, %%mm2 \n\t"\ |
| 102 | "pmulhw %%mm0, %%mm5 \n\t"\ |
| 103 | "paddw %%mm2, %%mm3 \n\t"\ |
| 104 | "paddw %%mm5, %%mm4 \n\t"\ |
| 105 | " jnz 1b \n\t"\ |
| 106 | "psraw $3, %%mm3 \n\t"\ |
| 107 | "psraw $3, %%mm4 \n\t"\ |
| 108 | "packuswb %%mm4, %%mm3 \n\t" |
| 109 | MOVNTQ2 " %%mm3, (%1, %%"REG_c")\n\t" |
| 110 | "add $8, %%"REG_c" \n\t"\ |
| 111 | "cmp %2, %%"REG_c" \n\t"\ |
| 112 | "movq %%mm6, %%mm3\n\t" |
| 113 | "movq %%mm7, %%mm4\n\t" |
| 114 | "mov %0, %%"REG_d" \n\t"\ |
| 115 | "mov (%%"REG_d"), %%"REG_S" \n\t"\ |
| 116 | "jb 1b \n\t"\ |
| 117 | :: "g" (filter), |
| 118 | "r" (dest-offset), "g" ((x86_reg)(dstW+offset)), "m" (offset) |
| 119 | : "%"REG_d, "%"REG_S, "%"REG_c |
| 120 | ); |
| 121 | } |
| 122 | |
| 123 | #define YSCALEYUV2PACKEDX_UV \ |
| 124 | __asm__ volatile(\ |
| 125 | "xor %%"REG_a", %%"REG_a" \n\t"\ |
| 126 | ".p2align 4 \n\t"\ |
| 127 | "nop \n\t"\ |
| 128 | "1: \n\t"\ |
| 129 | "lea "CHR_MMX_FILTER_OFFSET"(%0), %%"REG_d" \n\t"\ |
| 130 | "mov (%%"REG_d"), %%"REG_S" \n\t"\ |
| 131 | "movq "VROUNDER_OFFSET"(%0), %%mm3 \n\t"\ |
| 132 | "movq %%mm3, %%mm4 \n\t"\ |
| 133 | ".p2align 4 \n\t"\ |
| 134 | "2: \n\t"\ |
| 135 | "movq 8(%%"REG_d"), %%mm0 \n\t" /* filterCoeff */\ |
| 136 | "movq (%%"REG_S", %%"REG_a"), %%mm2 \n\t" /* UsrcData */\ |
| 137 | "add %6, %%"REG_S" \n\t" \ |
| 138 | "movq (%%"REG_S", %%"REG_a"), %%mm5 \n\t" /* VsrcData */\ |
| 139 | "add $16, %%"REG_d" \n\t"\ |
| 140 | "mov (%%"REG_d"), %%"REG_S" \n\t"\ |
| 141 | "pmulhw %%mm0, %%mm2 \n\t"\ |
| 142 | "pmulhw %%mm0, %%mm5 \n\t"\ |
| 143 | "paddw %%mm2, %%mm3 \n\t"\ |
| 144 | "paddw %%mm5, %%mm4 \n\t"\ |
| 145 | "test %%"REG_S", %%"REG_S" \n\t"\ |
| 146 | " jnz 2b \n\t"\ |
| 147 | |
| 148 | #define YSCALEYUV2PACKEDX_YA(offset,coeff,src1,src2,dst1,dst2) \ |
| 149 | "lea "offset"(%0), %%"REG_d" \n\t"\ |
| 150 | "mov (%%"REG_d"), %%"REG_S" \n\t"\ |
| 151 | "movq "VROUNDER_OFFSET"(%0), "#dst1" \n\t"\ |
| 152 | "movq "#dst1", "#dst2" \n\t"\ |
| 153 | ".p2align 4 \n\t"\ |
| 154 | "2: \n\t"\ |
| 155 | "movq 8(%%"REG_d"), "#coeff" \n\t" /* filterCoeff */\ |
| 156 | "movq (%%"REG_S", %%"REG_a", 2), "#src1" \n\t" /* Y1srcData */\ |
| 157 | "movq 8(%%"REG_S", %%"REG_a", 2), "#src2" \n\t" /* Y2srcData */\ |
| 158 | "add $16, %%"REG_d" \n\t"\ |
| 159 | "mov (%%"REG_d"), %%"REG_S" \n\t"\ |
| 160 | "pmulhw "#coeff", "#src1" \n\t"\ |
| 161 | "pmulhw "#coeff", "#src2" \n\t"\ |
| 162 | "paddw "#src1", "#dst1" \n\t"\ |
| 163 | "paddw "#src2", "#dst2" \n\t"\ |
| 164 | "test %%"REG_S", %%"REG_S" \n\t"\ |
| 165 | " jnz 2b \n\t"\ |
| 166 | |
| 167 | #define YSCALEYUV2PACKEDX \ |
| 168 | YSCALEYUV2PACKEDX_UV \ |
| 169 | YSCALEYUV2PACKEDX_YA(LUM_MMX_FILTER_OFFSET,%%mm0,%%mm2,%%mm5,%%mm1,%%mm7) \ |
| 170 | |
| 171 | #define YSCALEYUV2PACKEDX_END \ |
| 172 | :: "r" (&c->redDither), \ |
| 173 | "m" (dummy), "m" (dummy), "m" (dummy),\ |
| 174 | "r" (dest), "m" (dstW_reg), "m"(uv_off) \ |
| 175 | NAMED_CONSTRAINTS_ADD(bF8,bFC) \ |
| 176 | : "%"REG_a, "%"REG_d, "%"REG_S \ |
| 177 | ); |
| 178 | |
| 179 | #define YSCALEYUV2PACKEDX_ACCURATE_UV \ |
| 180 | __asm__ volatile(\ |
| 181 | "xor %%"REG_a", %%"REG_a" \n\t"\ |
| 182 | ".p2align 4 \n\t"\ |
| 183 | "nop \n\t"\ |
| 184 | "1: \n\t"\ |
| 185 | "lea "CHR_MMX_FILTER_OFFSET"(%0), %%"REG_d" \n\t"\ |
| 186 | "mov (%%"REG_d"), %%"REG_S" \n\t"\ |
| 187 | "pxor %%mm4, %%mm4 \n\t"\ |
| 188 | "pxor %%mm5, %%mm5 \n\t"\ |
| 189 | "pxor %%mm6, %%mm6 \n\t"\ |
| 190 | "pxor %%mm7, %%mm7 \n\t"\ |
| 191 | ".p2align 4 \n\t"\ |
| 192 | "2: \n\t"\ |
| 193 | "movq (%%"REG_S", %%"REG_a"), %%mm0 \n\t" /* UsrcData */\ |
| 194 | "add %6, %%"REG_S" \n\t" \ |
| 195 | "movq (%%"REG_S", %%"REG_a"), %%mm2 \n\t" /* VsrcData */\ |
| 196 | "mov "STR(APCK_PTR2)"(%%"REG_d"), %%"REG_S" \n\t"\ |
| 197 | "movq (%%"REG_S", %%"REG_a"), %%mm1 \n\t" /* UsrcData */\ |
| 198 | "movq %%mm0, %%mm3 \n\t"\ |
| 199 | "punpcklwd %%mm1, %%mm0 \n\t"\ |
| 200 | "punpckhwd %%mm1, %%mm3 \n\t"\ |
| 201 | "movq "STR(APCK_COEF)"(%%"REG_d"),%%mm1 \n\t" /* filterCoeff */\ |
| 202 | "pmaddwd %%mm1, %%mm0 \n\t"\ |
| 203 | "pmaddwd %%mm1, %%mm3 \n\t"\ |
| 204 | "paddd %%mm0, %%mm4 \n\t"\ |
| 205 | "paddd %%mm3, %%mm5 \n\t"\ |
| 206 | "add %6, %%"REG_S" \n\t" \ |
| 207 | "movq (%%"REG_S", %%"REG_a"), %%mm3 \n\t" /* VsrcData */\ |
| 208 | "mov "STR(APCK_SIZE)"(%%"REG_d"), %%"REG_S" \n\t"\ |
| 209 | "add $"STR(APCK_SIZE)", %%"REG_d" \n\t"\ |
| 210 | "test %%"REG_S", %%"REG_S" \n\t"\ |
| 211 | "movq %%mm2, %%mm0 \n\t"\ |
| 212 | "punpcklwd %%mm3, %%mm2 \n\t"\ |
| 213 | "punpckhwd %%mm3, %%mm0 \n\t"\ |
| 214 | "pmaddwd %%mm1, %%mm2 \n\t"\ |
| 215 | "pmaddwd %%mm1, %%mm0 \n\t"\ |
| 216 | "paddd %%mm2, %%mm6 \n\t"\ |
| 217 | "paddd %%mm0, %%mm7 \n\t"\ |
| 218 | " jnz 2b \n\t"\ |
| 219 | "psrad $16, %%mm4 \n\t"\ |
| 220 | "psrad $16, %%mm5 \n\t"\ |
| 221 | "psrad $16, %%mm6 \n\t"\ |
| 222 | "psrad $16, %%mm7 \n\t"\ |
| 223 | "movq "VROUNDER_OFFSET"(%0), %%mm0 \n\t"\ |
| 224 | "packssdw %%mm5, %%mm4 \n\t"\ |
| 225 | "packssdw %%mm7, %%mm6 \n\t"\ |
| 226 | "paddw %%mm0, %%mm4 \n\t"\ |
| 227 | "paddw %%mm0, %%mm6 \n\t"\ |
| 228 | "movq %%mm4, "U_TEMP"(%0) \n\t"\ |
| 229 | "movq %%mm6, "V_TEMP"(%0) \n\t"\ |
| 230 | |
| 231 | #define YSCALEYUV2PACKEDX_ACCURATE_YA(offset) \ |
| 232 | "lea "offset"(%0), %%"REG_d" \n\t"\ |
| 233 | "mov (%%"REG_d"), %%"REG_S" \n\t"\ |
| 234 | "pxor %%mm1, %%mm1 \n\t"\ |
| 235 | "pxor %%mm5, %%mm5 \n\t"\ |
| 236 | "pxor %%mm7, %%mm7 \n\t"\ |
| 237 | "pxor %%mm6, %%mm6 \n\t"\ |
| 238 | ".p2align 4 \n\t"\ |
| 239 | "2: \n\t"\ |
| 240 | "movq (%%"REG_S", %%"REG_a", 2), %%mm0 \n\t" /* Y1srcData */\ |
| 241 | "movq 8(%%"REG_S", %%"REG_a", 2), %%mm2 \n\t" /* Y2srcData */\ |
| 242 | "mov "STR(APCK_PTR2)"(%%"REG_d"), %%"REG_S" \n\t"\ |
| 243 | "movq (%%"REG_S", %%"REG_a", 2), %%mm4 \n\t" /* Y1srcData */\ |
| 244 | "movq %%mm0, %%mm3 \n\t"\ |
| 245 | "punpcklwd %%mm4, %%mm0 \n\t"\ |
| 246 | "punpckhwd %%mm4, %%mm3 \n\t"\ |
| 247 | "movq "STR(APCK_COEF)"(%%"REG_d"), %%mm4 \n\t" /* filterCoeff */\ |
| 248 | "pmaddwd %%mm4, %%mm0 \n\t"\ |
| 249 | "pmaddwd %%mm4, %%mm3 \n\t"\ |
| 250 | "paddd %%mm0, %%mm1 \n\t"\ |
| 251 | "paddd %%mm3, %%mm5 \n\t"\ |
| 252 | "movq 8(%%"REG_S", %%"REG_a", 2), %%mm3 \n\t" /* Y2srcData */\ |
| 253 | "mov "STR(APCK_SIZE)"(%%"REG_d"), %%"REG_S" \n\t"\ |
| 254 | "add $"STR(APCK_SIZE)", %%"REG_d" \n\t"\ |
| 255 | "test %%"REG_S", %%"REG_S" \n\t"\ |
| 256 | "movq %%mm2, %%mm0 \n\t"\ |
| 257 | "punpcklwd %%mm3, %%mm2 \n\t"\ |
| 258 | "punpckhwd %%mm3, %%mm0 \n\t"\ |
| 259 | "pmaddwd %%mm4, %%mm2 \n\t"\ |
| 260 | "pmaddwd %%mm4, %%mm0 \n\t"\ |
| 261 | "paddd %%mm2, %%mm7 \n\t"\ |
| 262 | "paddd %%mm0, %%mm6 \n\t"\ |
| 263 | " jnz 2b \n\t"\ |
| 264 | "psrad $16, %%mm1 \n\t"\ |
| 265 | "psrad $16, %%mm5 \n\t"\ |
| 266 | "psrad $16, %%mm7 \n\t"\ |
| 267 | "psrad $16, %%mm6 \n\t"\ |
| 268 | "movq "VROUNDER_OFFSET"(%0), %%mm0 \n\t"\ |
| 269 | "packssdw %%mm5, %%mm1 \n\t"\ |
| 270 | "packssdw %%mm6, %%mm7 \n\t"\ |
| 271 | "paddw %%mm0, %%mm1 \n\t"\ |
| 272 | "paddw %%mm0, %%mm7 \n\t"\ |
| 273 | "movq "U_TEMP"(%0), %%mm3 \n\t"\ |
| 274 | "movq "V_TEMP"(%0), %%mm4 \n\t"\ |
| 275 | |
| 276 | #define YSCALEYUV2PACKEDX_ACCURATE \ |
| 277 | YSCALEYUV2PACKEDX_ACCURATE_UV \ |
| 278 | YSCALEYUV2PACKEDX_ACCURATE_YA(LUM_MMX_FILTER_OFFSET) |
| 279 | |
| 280 | #define YSCALEYUV2RGBX \ |
| 281 | "psubw "U_OFFSET"(%0), %%mm3 \n\t" /* (U-128)8*/\ |
| 282 | "psubw "V_OFFSET"(%0), %%mm4 \n\t" /* (V-128)8*/\ |
| 283 | "movq %%mm3, %%mm2 \n\t" /* (U-128)8*/\ |
| 284 | "movq %%mm4, %%mm5 \n\t" /* (V-128)8*/\ |
| 285 | "pmulhw "UG_COEFF"(%0), %%mm3 \n\t"\ |
| 286 | "pmulhw "VG_COEFF"(%0), %%mm4 \n\t"\ |
| 287 | /* mm2=(U-128)8, mm3=ug, mm4=vg mm5=(V-128)8 */\ |
| 288 | "pmulhw "UB_COEFF"(%0), %%mm2 \n\t"\ |
| 289 | "pmulhw "VR_COEFF"(%0), %%mm5 \n\t"\ |
| 290 | "psubw "Y_OFFSET"(%0), %%mm1 \n\t" /* 8(Y-16)*/\ |
| 291 | "psubw "Y_OFFSET"(%0), %%mm7 \n\t" /* 8(Y-16)*/\ |
| 292 | "pmulhw "Y_COEFF"(%0), %%mm1 \n\t"\ |
| 293 | "pmulhw "Y_COEFF"(%0), %%mm7 \n\t"\ |
| 294 | /* mm1= Y1, mm2=ub, mm3=ug, mm4=vg mm5=vr, mm7=Y2 */\ |
| 295 | "paddw %%mm3, %%mm4 \n\t"\ |
| 296 | "movq %%mm2, %%mm0 \n\t"\ |
| 297 | "movq %%mm5, %%mm6 \n\t"\ |
| 298 | "movq %%mm4, %%mm3 \n\t"\ |
| 299 | "punpcklwd %%mm2, %%mm2 \n\t"\ |
| 300 | "punpcklwd %%mm5, %%mm5 \n\t"\ |
| 301 | "punpcklwd %%mm4, %%mm4 \n\t"\ |
| 302 | "paddw %%mm1, %%mm2 \n\t"\ |
| 303 | "paddw %%mm1, %%mm5 \n\t"\ |
| 304 | "paddw %%mm1, %%mm4 \n\t"\ |
| 305 | "punpckhwd %%mm0, %%mm0 \n\t"\ |
| 306 | "punpckhwd %%mm6, %%mm6 \n\t"\ |
| 307 | "punpckhwd %%mm3, %%mm3 \n\t"\ |
| 308 | "paddw %%mm7, %%mm0 \n\t"\ |
| 309 | "paddw %%mm7, %%mm6 \n\t"\ |
| 310 | "paddw %%mm7, %%mm3 \n\t"\ |
| 311 | /* mm0=B1, mm2=B2, mm3=G2, mm4=G1, mm5=R1, mm6=R2 */\ |
| 312 | "packuswb %%mm0, %%mm2 \n\t"\ |
| 313 | "packuswb %%mm6, %%mm5 \n\t"\ |
| 314 | "packuswb %%mm3, %%mm4 \n\t"\ |
| 315 | |
| 316 | #define REAL_WRITEBGR32(dst, dstw, index, b, g, r, a, q0, q2, q3, t) \ |
| 317 | "movq "#b", "#q2" \n\t" /* B */\ |
| 318 | "movq "#r", "#t" \n\t" /* R */\ |
| 319 | "punpcklbw "#g", "#b" \n\t" /* GBGBGBGB 0 */\ |
| 320 | "punpcklbw "#a", "#r" \n\t" /* ARARARAR 0 */\ |
| 321 | "punpckhbw "#g", "#q2" \n\t" /* GBGBGBGB 2 */\ |
| 322 | "punpckhbw "#a", "#t" \n\t" /* ARARARAR 2 */\ |
| 323 | "movq "#b", "#q0" \n\t" /* GBGBGBGB 0 */\ |
| 324 | "movq "#q2", "#q3" \n\t" /* GBGBGBGB 2 */\ |
| 325 | "punpcklwd "#r", "#q0" \n\t" /* ARGBARGB 0 */\ |
| 326 | "punpckhwd "#r", "#b" \n\t" /* ARGBARGB 1 */\ |
| 327 | "punpcklwd "#t", "#q2" \n\t" /* ARGBARGB 2 */\ |
| 328 | "punpckhwd "#t", "#q3" \n\t" /* ARGBARGB 3 */\ |
| 329 | \ |
| 330 | MOVNTQ( q0, (dst, index, 4))\ |
| 331 | MOVNTQ( b, 8(dst, index, 4))\ |
| 332 | MOVNTQ( q2, 16(dst, index, 4))\ |
| 333 | MOVNTQ( q3, 24(dst, index, 4))\ |
| 334 | \ |
| 335 | "add $8, "#index" \n\t"\ |
| 336 | "cmp "dstw", "#index" \n\t"\ |
| 337 | " jb 1b \n\t" |
| 338 | #define WRITEBGR32(dst, dstw, index, b, g, r, a, q0, q2, q3, t) REAL_WRITEBGR32(dst, dstw, index, b, g, r, a, q0, q2, q3, t) |
| 339 | |
| 340 | static void RENAME(yuv2rgb32_X_ar)(SwsContext *c, const int16_t *lumFilter, |
| 341 | const int16_t **lumSrc, int lumFilterSize, |
| 342 | const int16_t *chrFilter, const int16_t **chrUSrc, |
| 343 | const int16_t **chrVSrc, |
| 344 | int chrFilterSize, const int16_t **alpSrc, |
| 345 | uint8_t *dest, int dstW, int dstY) |
| 346 | { |
| 347 | x86_reg dummy=0; |
| 348 | x86_reg dstW_reg = dstW; |
| 349 | x86_reg uv_off = c->uv_offx2; |
| 350 | |
| 351 | if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) { |
| 352 | YSCALEYUV2PACKEDX_ACCURATE |
| 353 | YSCALEYUV2RGBX |
| 354 | "movq %%mm2, "U_TEMP"(%0) \n\t" |
| 355 | "movq %%mm4, "V_TEMP"(%0) \n\t" |
| 356 | "movq %%mm5, "Y_TEMP"(%0) \n\t" |
| 357 | YSCALEYUV2PACKEDX_ACCURATE_YA(ALP_MMX_FILTER_OFFSET) |
| 358 | "movq "Y_TEMP"(%0), %%mm5 \n\t" |
| 359 | "psraw $3, %%mm1 \n\t" |
| 360 | "psraw $3, %%mm7 \n\t" |
| 361 | "packuswb %%mm7, %%mm1 \n\t" |
| 362 | WRITEBGR32(%4, "%5", %%REGa, %%mm3, %%mm4, %%mm5, %%mm1, %%mm0, %%mm7, %%mm2, %%mm6) |
| 363 | YSCALEYUV2PACKEDX_END |
| 364 | } else { |
| 365 | YSCALEYUV2PACKEDX_ACCURATE |
| 366 | YSCALEYUV2RGBX |
| 367 | "pcmpeqd %%mm7, %%mm7 \n\t" |
| 368 | WRITEBGR32(%4, "%5", %%REGa, %%mm2, %%mm4, %%mm5, %%mm7, %%mm0, %%mm1, %%mm3, %%mm6) |
| 369 | YSCALEYUV2PACKEDX_END |
| 370 | } |
| 371 | } |
| 372 | |
| 373 | static void RENAME(yuv2rgb32_X)(SwsContext *c, const int16_t *lumFilter, |
| 374 | const int16_t **lumSrc, int lumFilterSize, |
| 375 | const int16_t *chrFilter, const int16_t **chrUSrc, |
| 376 | const int16_t **chrVSrc, |
| 377 | int chrFilterSize, const int16_t **alpSrc, |
| 378 | uint8_t *dest, int dstW, int dstY) |
| 379 | { |
| 380 | x86_reg dummy=0; |
| 381 | x86_reg dstW_reg = dstW; |
| 382 | x86_reg uv_off = c->uv_offx2; |
| 383 | |
| 384 | if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) { |
| 385 | YSCALEYUV2PACKEDX |
| 386 | YSCALEYUV2RGBX |
| 387 | YSCALEYUV2PACKEDX_YA(ALP_MMX_FILTER_OFFSET, %%mm0, %%mm3, %%mm6, %%mm1, %%mm7) |
| 388 | "psraw $3, %%mm1 \n\t" |
| 389 | "psraw $3, %%mm7 \n\t" |
| 390 | "packuswb %%mm7, %%mm1 \n\t" |
| 391 | WRITEBGR32(%4, "%5", %%REGa, %%mm2, %%mm4, %%mm5, %%mm1, %%mm0, %%mm7, %%mm3, %%mm6) |
| 392 | YSCALEYUV2PACKEDX_END |
| 393 | } else { |
| 394 | YSCALEYUV2PACKEDX |
| 395 | YSCALEYUV2RGBX |
| 396 | "pcmpeqd %%mm7, %%mm7 \n\t" |
| 397 | WRITEBGR32(%4, "%5", %%REGa, %%mm2, %%mm4, %%mm5, %%mm7, %%mm0, %%mm1, %%mm3, %%mm6) |
| 398 | YSCALEYUV2PACKEDX_END |
| 399 | } |
| 400 | } |
| 401 | |
| 402 | #define REAL_WRITERGB16(dst, dstw, index) \ |
| 403 | "pand "MANGLE(bF8)", %%mm2 \n\t" /* B */\ |
| 404 | "pand "MANGLE(bFC)", %%mm4 \n\t" /* G */\ |
| 405 | "pand "MANGLE(bF8)", %%mm5 \n\t" /* R */\ |
| 406 | "psrlq $3, %%mm2 \n\t"\ |
| 407 | \ |
| 408 | "movq %%mm2, %%mm1 \n\t"\ |
| 409 | "movq %%mm4, %%mm3 \n\t"\ |
| 410 | \ |
| 411 | "punpcklbw %%mm7, %%mm3 \n\t"\ |
| 412 | "punpcklbw %%mm5, %%mm2 \n\t"\ |
| 413 | "punpckhbw %%mm7, %%mm4 \n\t"\ |
| 414 | "punpckhbw %%mm5, %%mm1 \n\t"\ |
| 415 | \ |
| 416 | "psllq $3, %%mm3 \n\t"\ |
| 417 | "psllq $3, %%mm4 \n\t"\ |
| 418 | \ |
| 419 | "por %%mm3, %%mm2 \n\t"\ |
| 420 | "por %%mm4, %%mm1 \n\t"\ |
| 421 | \ |
| 422 | MOVNTQ(%%mm2, (dst, index, 2))\ |
| 423 | MOVNTQ(%%mm1, 8(dst, index, 2))\ |
| 424 | \ |
| 425 | "add $8, "#index" \n\t"\ |
| 426 | "cmp "dstw", "#index" \n\t"\ |
| 427 | " jb 1b \n\t" |
| 428 | #define WRITERGB16(dst, dstw, index) REAL_WRITERGB16(dst, dstw, index) |
| 429 | |
| 430 | static void RENAME(yuv2rgb565_X_ar)(SwsContext *c, const int16_t *lumFilter, |
| 431 | const int16_t **lumSrc, int lumFilterSize, |
| 432 | const int16_t *chrFilter, const int16_t **chrUSrc, |
| 433 | const int16_t **chrVSrc, |
| 434 | int chrFilterSize, const int16_t **alpSrc, |
| 435 | uint8_t *dest, int dstW, int dstY) |
| 436 | { |
| 437 | x86_reg dummy=0; |
| 438 | x86_reg dstW_reg = dstW; |
| 439 | x86_reg uv_off = c->uv_offx2; |
| 440 | |
| 441 | YSCALEYUV2PACKEDX_ACCURATE |
| 442 | YSCALEYUV2RGBX |
| 443 | "pxor %%mm7, %%mm7 \n\t" |
| 444 | /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */ |
| 445 | #ifdef DITHER1XBPP |
| 446 | "paddusb "BLUE_DITHER"(%0), %%mm2\n\t" |
| 447 | "paddusb "GREEN_DITHER"(%0), %%mm4\n\t" |
| 448 | "paddusb "RED_DITHER"(%0), %%mm5\n\t" |
| 449 | #endif |
| 450 | WRITERGB16(%4, "%5", %%REGa) |
| 451 | YSCALEYUV2PACKEDX_END |
| 452 | } |
| 453 | |
| 454 | static void RENAME(yuv2rgb565_X)(SwsContext *c, const int16_t *lumFilter, |
| 455 | const int16_t **lumSrc, int lumFilterSize, |
| 456 | const int16_t *chrFilter, const int16_t **chrUSrc, |
| 457 | const int16_t **chrVSrc, |
| 458 | int chrFilterSize, const int16_t **alpSrc, |
| 459 | uint8_t *dest, int dstW, int dstY) |
| 460 | { |
| 461 | x86_reg dummy=0; |
| 462 | x86_reg dstW_reg = dstW; |
| 463 | x86_reg uv_off = c->uv_offx2; |
| 464 | |
| 465 | YSCALEYUV2PACKEDX |
| 466 | YSCALEYUV2RGBX |
| 467 | "pxor %%mm7, %%mm7 \n\t" |
| 468 | /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */ |
| 469 | #ifdef DITHER1XBPP |
| 470 | "paddusb "BLUE_DITHER"(%0), %%mm2 \n\t" |
| 471 | "paddusb "GREEN_DITHER"(%0), %%mm4 \n\t" |
| 472 | "paddusb "RED_DITHER"(%0), %%mm5 \n\t" |
| 473 | #endif |
| 474 | WRITERGB16(%4, "%5", %%REGa) |
| 475 | YSCALEYUV2PACKEDX_END |
| 476 | } |
| 477 | |
| 478 | #define REAL_WRITERGB15(dst, dstw, index) \ |
| 479 | "pand "MANGLE(bF8)", %%mm2 \n\t" /* B */\ |
| 480 | "pand "MANGLE(bF8)", %%mm4 \n\t" /* G */\ |
| 481 | "pand "MANGLE(bF8)", %%mm5 \n\t" /* R */\ |
| 482 | "psrlq $3, %%mm2 \n\t"\ |
| 483 | "psrlq $1, %%mm5 \n\t"\ |
| 484 | \ |
| 485 | "movq %%mm2, %%mm1 \n\t"\ |
| 486 | "movq %%mm4, %%mm3 \n\t"\ |
| 487 | \ |
| 488 | "punpcklbw %%mm7, %%mm3 \n\t"\ |
| 489 | "punpcklbw %%mm5, %%mm2 \n\t"\ |
| 490 | "punpckhbw %%mm7, %%mm4 \n\t"\ |
| 491 | "punpckhbw %%mm5, %%mm1 \n\t"\ |
| 492 | \ |
| 493 | "psllq $2, %%mm3 \n\t"\ |
| 494 | "psllq $2, %%mm4 \n\t"\ |
| 495 | \ |
| 496 | "por %%mm3, %%mm2 \n\t"\ |
| 497 | "por %%mm4, %%mm1 \n\t"\ |
| 498 | \ |
| 499 | MOVNTQ(%%mm2, (dst, index, 2))\ |
| 500 | MOVNTQ(%%mm1, 8(dst, index, 2))\ |
| 501 | \ |
| 502 | "add $8, "#index" \n\t"\ |
| 503 | "cmp "dstw", "#index" \n\t"\ |
| 504 | " jb 1b \n\t" |
| 505 | #define WRITERGB15(dst, dstw, index) REAL_WRITERGB15(dst, dstw, index) |
| 506 | |
| 507 | static void RENAME(yuv2rgb555_X_ar)(SwsContext *c, const int16_t *lumFilter, |
| 508 | const int16_t **lumSrc, int lumFilterSize, |
| 509 | const int16_t *chrFilter, const int16_t **chrUSrc, |
| 510 | const int16_t **chrVSrc, |
| 511 | int chrFilterSize, const int16_t **alpSrc, |
| 512 | uint8_t *dest, int dstW, int dstY) |
| 513 | { |
| 514 | x86_reg dummy=0; |
| 515 | x86_reg dstW_reg = dstW; |
| 516 | x86_reg uv_off = c->uv_offx2; |
| 517 | |
| 518 | YSCALEYUV2PACKEDX_ACCURATE |
| 519 | YSCALEYUV2RGBX |
| 520 | "pxor %%mm7, %%mm7 \n\t" |
| 521 | /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */ |
| 522 | #ifdef DITHER1XBPP |
| 523 | "paddusb "BLUE_DITHER"(%0), %%mm2\n\t" |
| 524 | "paddusb "GREEN_DITHER"(%0), %%mm4\n\t" |
| 525 | "paddusb "RED_DITHER"(%0), %%mm5\n\t" |
| 526 | #endif |
| 527 | WRITERGB15(%4, "%5", %%REGa) |
| 528 | YSCALEYUV2PACKEDX_END |
| 529 | } |
| 530 | |
| 531 | static void RENAME(yuv2rgb555_X)(SwsContext *c, const int16_t *lumFilter, |
| 532 | const int16_t **lumSrc, int lumFilterSize, |
| 533 | const int16_t *chrFilter, const int16_t **chrUSrc, |
| 534 | const int16_t **chrVSrc, |
| 535 | int chrFilterSize, const int16_t **alpSrc, |
| 536 | uint8_t *dest, int dstW, int dstY) |
| 537 | { |
| 538 | x86_reg dummy=0; |
| 539 | x86_reg dstW_reg = dstW; |
| 540 | x86_reg uv_off = c->uv_offx2; |
| 541 | |
| 542 | YSCALEYUV2PACKEDX |
| 543 | YSCALEYUV2RGBX |
| 544 | "pxor %%mm7, %%mm7 \n\t" |
| 545 | /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */ |
| 546 | #ifdef DITHER1XBPP |
| 547 | "paddusb "BLUE_DITHER"(%0), %%mm2 \n\t" |
| 548 | "paddusb "GREEN_DITHER"(%0), %%mm4 \n\t" |
| 549 | "paddusb "RED_DITHER"(%0), %%mm5 \n\t" |
| 550 | #endif |
| 551 | WRITERGB15(%4, "%5", %%REGa) |
| 552 | YSCALEYUV2PACKEDX_END |
| 553 | } |
| 554 | |
| 555 | #define WRITEBGR24MMX(dst, dstw, index) \ |
| 556 | /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */\ |
| 557 | "movq %%mm2, %%mm1 \n\t" /* B */\ |
| 558 | "movq %%mm5, %%mm6 \n\t" /* R */\ |
| 559 | "punpcklbw %%mm4, %%mm2 \n\t" /* GBGBGBGB 0 */\ |
| 560 | "punpcklbw %%mm7, %%mm5 \n\t" /* 0R0R0R0R 0 */\ |
| 561 | "punpckhbw %%mm4, %%mm1 \n\t" /* GBGBGBGB 2 */\ |
| 562 | "punpckhbw %%mm7, %%mm6 \n\t" /* 0R0R0R0R 2 */\ |
| 563 | "movq %%mm2, %%mm0 \n\t" /* GBGBGBGB 0 */\ |
| 564 | "movq %%mm1, %%mm3 \n\t" /* GBGBGBGB 2 */\ |
| 565 | "punpcklwd %%mm5, %%mm0 \n\t" /* 0RGB0RGB 0 */\ |
| 566 | "punpckhwd %%mm5, %%mm2 \n\t" /* 0RGB0RGB 1 */\ |
| 567 | "punpcklwd %%mm6, %%mm1 \n\t" /* 0RGB0RGB 2 */\ |
| 568 | "punpckhwd %%mm6, %%mm3 \n\t" /* 0RGB0RGB 3 */\ |
| 569 | \ |
| 570 | "movq %%mm0, %%mm4 \n\t" /* 0RGB0RGB 0 */\ |
| 571 | "movq %%mm2, %%mm6 \n\t" /* 0RGB0RGB 1 */\ |
| 572 | "movq %%mm1, %%mm5 \n\t" /* 0RGB0RGB 2 */\ |
| 573 | "movq %%mm3, %%mm7 \n\t" /* 0RGB0RGB 3 */\ |
| 574 | \ |
| 575 | "psllq $40, %%mm0 \n\t" /* RGB00000 0 */\ |
| 576 | "psllq $40, %%mm2 \n\t" /* RGB00000 1 */\ |
| 577 | "psllq $40, %%mm1 \n\t" /* RGB00000 2 */\ |
| 578 | "psllq $40, %%mm3 \n\t" /* RGB00000 3 */\ |
| 579 | \ |
| 580 | "punpckhdq %%mm4, %%mm0 \n\t" /* 0RGBRGB0 0 */\ |
| 581 | "punpckhdq %%mm6, %%mm2 \n\t" /* 0RGBRGB0 1 */\ |
| 582 | "punpckhdq %%mm5, %%mm1 \n\t" /* 0RGBRGB0 2 */\ |
| 583 | "punpckhdq %%mm7, %%mm3 \n\t" /* 0RGBRGB0 3 */\ |
| 584 | \ |
| 585 | "psrlq $8, %%mm0 \n\t" /* 00RGBRGB 0 */\ |
| 586 | "movq %%mm2, %%mm6 \n\t" /* 0RGBRGB0 1 */\ |
| 587 | "psllq $40, %%mm2 \n\t" /* GB000000 1 */\ |
| 588 | "por %%mm2, %%mm0 \n\t" /* GBRGBRGB 0 */\ |
| 589 | MOVNTQ(%%mm0, (dst))\ |
| 590 | \ |
| 591 | "psrlq $24, %%mm6 \n\t" /* 0000RGBR 1 */\ |
| 592 | "movq %%mm1, %%mm5 \n\t" /* 0RGBRGB0 2 */\ |
| 593 | "psllq $24, %%mm1 \n\t" /* BRGB0000 2 */\ |
| 594 | "por %%mm1, %%mm6 \n\t" /* BRGBRGBR 1 */\ |
| 595 | MOVNTQ(%%mm6, 8(dst))\ |
| 596 | \ |
| 597 | "psrlq $40, %%mm5 \n\t" /* 000000RG 2 */\ |
| 598 | "psllq $8, %%mm3 \n\t" /* RGBRGB00 3 */\ |
| 599 | "por %%mm3, %%mm5 \n\t" /* RGBRGBRG 2 */\ |
| 600 | MOVNTQ(%%mm5, 16(dst))\ |
| 601 | \ |
| 602 | "add $24, "#dst" \n\t"\ |
| 603 | \ |
| 604 | "add $8, "#index" \n\t"\ |
| 605 | "cmp "dstw", "#index" \n\t"\ |
| 606 | " jb 1b \n\t" |
| 607 | |
| 608 | #define WRITEBGR24MMXEXT(dst, dstw, index) \ |
| 609 | /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */\ |
| 610 | "movq "MANGLE(ff_M24A)", %%mm0 \n\t"\ |
| 611 | "movq "MANGLE(ff_M24C)", %%mm7 \n\t"\ |
| 612 | "pshufw $0x50, %%mm2, %%mm1 \n\t" /* B3 B2 B3 B2 B1 B0 B1 B0 */\ |
| 613 | "pshufw $0x50, %%mm4, %%mm3 \n\t" /* G3 G2 G3 G2 G1 G0 G1 G0 */\ |
| 614 | "pshufw $0x00, %%mm5, %%mm6 \n\t" /* R1 R0 R1 R0 R1 R0 R1 R0 */\ |
| 615 | \ |
| 616 | "pand %%mm0, %%mm1 \n\t" /* B2 B1 B0 */\ |
| 617 | "pand %%mm0, %%mm3 \n\t" /* G2 G1 G0 */\ |
| 618 | "pand %%mm7, %%mm6 \n\t" /* R1 R0 */\ |
| 619 | \ |
| 620 | "psllq $8, %%mm3 \n\t" /* G2 G1 G0 */\ |
| 621 | "por %%mm1, %%mm6 \n\t"\ |
| 622 | "por %%mm3, %%mm6 \n\t"\ |
| 623 | MOVNTQ(%%mm6, (dst))\ |
| 624 | \ |
| 625 | "psrlq $8, %%mm4 \n\t" /* 00 G7 G6 G5 G4 G3 G2 G1 */\ |
| 626 | "pshufw $0xA5, %%mm2, %%mm1 \n\t" /* B5 B4 B5 B4 B3 B2 B3 B2 */\ |
| 627 | "pshufw $0x55, %%mm4, %%mm3 \n\t" /* G4 G3 G4 G3 G4 G3 G4 G3 */\ |
| 628 | "pshufw $0xA5, %%mm5, %%mm6 \n\t" /* R5 R4 R5 R4 R3 R2 R3 R2 */\ |
| 629 | \ |
| 630 | "pand "MANGLE(ff_M24B)", %%mm1 \n\t" /* B5 B4 B3 */\ |
| 631 | "pand %%mm7, %%mm3 \n\t" /* G4 G3 */\ |
| 632 | "pand %%mm0, %%mm6 \n\t" /* R4 R3 R2 */\ |
| 633 | \ |
| 634 | "por %%mm1, %%mm3 \n\t" /* B5 G4 B4 G3 B3 */\ |
| 635 | "por %%mm3, %%mm6 \n\t"\ |
| 636 | MOVNTQ(%%mm6, 8(dst))\ |
| 637 | \ |
| 638 | "pshufw $0xFF, %%mm2, %%mm1 \n\t" /* B7 B6 B7 B6 B7 B6 B6 B7 */\ |
| 639 | "pshufw $0xFA, %%mm4, %%mm3 \n\t" /* 00 G7 00 G7 G6 G5 G6 G5 */\ |
| 640 | "pshufw $0xFA, %%mm5, %%mm6 \n\t" /* R7 R6 R7 R6 R5 R4 R5 R4 */\ |
| 641 | \ |
| 642 | "pand %%mm7, %%mm1 \n\t" /* B7 B6 */\ |
| 643 | "pand %%mm0, %%mm3 \n\t" /* G7 G6 G5 */\ |
| 644 | "pand "MANGLE(ff_M24B)", %%mm6 \n\t" /* R7 R6 R5 */\ |
| 645 | \ |
| 646 | "por %%mm1, %%mm3 \n\t"\ |
| 647 | "por %%mm3, %%mm6 \n\t"\ |
| 648 | MOVNTQ(%%mm6, 16(dst))\ |
| 649 | \ |
| 650 | "add $24, "#dst" \n\t"\ |
| 651 | \ |
| 652 | "add $8, "#index" \n\t"\ |
| 653 | "cmp "dstw", "#index" \n\t"\ |
| 654 | " jb 1b \n\t" |
| 655 | |
| 656 | #if COMPILE_TEMPLATE_MMXEXT |
| 657 | #undef WRITEBGR24 |
| 658 | #define WRITEBGR24(dst, dstw, index) WRITEBGR24MMXEXT(dst, dstw, index) |
| 659 | #else |
| 660 | #undef WRITEBGR24 |
| 661 | #define WRITEBGR24(dst, dstw, index) WRITEBGR24MMX(dst, dstw, index) |
| 662 | #endif |
| 663 | |
| 664 | #if HAVE_6REGS |
| 665 | static void RENAME(yuv2bgr24_X_ar)(SwsContext *c, const int16_t *lumFilter, |
| 666 | const int16_t **lumSrc, int lumFilterSize, |
| 667 | const int16_t *chrFilter, const int16_t **chrUSrc, |
| 668 | const int16_t **chrVSrc, |
| 669 | int chrFilterSize, const int16_t **alpSrc, |
| 670 | uint8_t *dest, int dstW, int dstY) |
| 671 | { |
| 672 | x86_reg dummy=0; |
| 673 | x86_reg dstW_reg = dstW; |
| 674 | x86_reg uv_off = c->uv_offx2; |
| 675 | |
| 676 | YSCALEYUV2PACKEDX_ACCURATE |
| 677 | YSCALEYUV2RGBX |
| 678 | "pxor %%mm7, %%mm7 \n\t" |
| 679 | "lea (%%"REG_a", %%"REG_a", 2), %%"REG_c"\n\t" //FIXME optimize |
| 680 | "add %4, %%"REG_c" \n\t" |
| 681 | WRITEBGR24(%%REGc, "%5", %%REGa) |
| 682 | :: "r" (&c->redDither), |
| 683 | "m" (dummy), "m" (dummy), "m" (dummy), |
| 684 | "r" (dest), "m" (dstW_reg), "m"(uv_off) |
| 685 | NAMED_CONSTRAINTS_ADD(ff_M24A,ff_M24C,ff_M24B) |
| 686 | : "%"REG_a, "%"REG_c, "%"REG_d, "%"REG_S |
| 687 | ); |
| 688 | } |
| 689 | |
| 690 | static void RENAME(yuv2bgr24_X)(SwsContext *c, const int16_t *lumFilter, |
| 691 | const int16_t **lumSrc, int lumFilterSize, |
| 692 | const int16_t *chrFilter, const int16_t **chrUSrc, |
| 693 | const int16_t **chrVSrc, |
| 694 | int chrFilterSize, const int16_t **alpSrc, |
| 695 | uint8_t *dest, int dstW, int dstY) |
| 696 | { |
| 697 | x86_reg dummy=0; |
| 698 | x86_reg dstW_reg = dstW; |
| 699 | x86_reg uv_off = c->uv_offx2; |
| 700 | |
| 701 | YSCALEYUV2PACKEDX |
| 702 | YSCALEYUV2RGBX |
| 703 | "pxor %%mm7, %%mm7 \n\t" |
| 704 | "lea (%%"REG_a", %%"REG_a", 2), %%"REG_c" \n\t" //FIXME optimize |
| 705 | "add %4, %%"REG_c" \n\t" |
| 706 | WRITEBGR24(%%REGc, "%5", %%REGa) |
| 707 | :: "r" (&c->redDither), |
| 708 | "m" (dummy), "m" (dummy), "m" (dummy), |
| 709 | "r" (dest), "m" (dstW_reg), "m"(uv_off) |
| 710 | NAMED_CONSTRAINTS_ADD(ff_M24A,ff_M24C,ff_M24B) |
| 711 | : "%"REG_a, "%"REG_c, "%"REG_d, "%"REG_S |
| 712 | ); |
| 713 | } |
| 714 | #endif /* HAVE_6REGS */ |
| 715 | |
| 716 | #define REAL_WRITEYUY2(dst, dstw, index) \ |
| 717 | "packuswb %%mm3, %%mm3 \n\t"\ |
| 718 | "packuswb %%mm4, %%mm4 \n\t"\ |
| 719 | "packuswb %%mm7, %%mm1 \n\t"\ |
| 720 | "punpcklbw %%mm4, %%mm3 \n\t"\ |
| 721 | "movq %%mm1, %%mm7 \n\t"\ |
| 722 | "punpcklbw %%mm3, %%mm1 \n\t"\ |
| 723 | "punpckhbw %%mm3, %%mm7 \n\t"\ |
| 724 | \ |
| 725 | MOVNTQ(%%mm1, (dst, index, 2))\ |
| 726 | MOVNTQ(%%mm7, 8(dst, index, 2))\ |
| 727 | \ |
| 728 | "add $8, "#index" \n\t"\ |
| 729 | "cmp "dstw", "#index" \n\t"\ |
| 730 | " jb 1b \n\t" |
| 731 | #define WRITEYUY2(dst, dstw, index) REAL_WRITEYUY2(dst, dstw, index) |
| 732 | |
| 733 | static void RENAME(yuv2yuyv422_X_ar)(SwsContext *c, const int16_t *lumFilter, |
| 734 | const int16_t **lumSrc, int lumFilterSize, |
| 735 | const int16_t *chrFilter, const int16_t **chrUSrc, |
| 736 | const int16_t **chrVSrc, |
| 737 | int chrFilterSize, const int16_t **alpSrc, |
| 738 | uint8_t *dest, int dstW, int dstY) |
| 739 | { |
| 740 | x86_reg dummy=0; |
| 741 | x86_reg dstW_reg = dstW; |
| 742 | x86_reg uv_off = c->uv_offx2; |
| 743 | |
| 744 | YSCALEYUV2PACKEDX_ACCURATE |
| 745 | /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */ |
| 746 | "psraw $3, %%mm3 \n\t" |
| 747 | "psraw $3, %%mm4 \n\t" |
| 748 | "psraw $3, %%mm1 \n\t" |
| 749 | "psraw $3, %%mm7 \n\t" |
| 750 | WRITEYUY2(%4, "%5", %%REGa) |
| 751 | YSCALEYUV2PACKEDX_END |
| 752 | } |
| 753 | |
| 754 | static void RENAME(yuv2yuyv422_X)(SwsContext *c, const int16_t *lumFilter, |
| 755 | const int16_t **lumSrc, int lumFilterSize, |
| 756 | const int16_t *chrFilter, const int16_t **chrUSrc, |
| 757 | const int16_t **chrVSrc, |
| 758 | int chrFilterSize, const int16_t **alpSrc, |
| 759 | uint8_t *dest, int dstW, int dstY) |
| 760 | { |
| 761 | x86_reg dummy=0; |
| 762 | x86_reg dstW_reg = dstW; |
| 763 | x86_reg uv_off = c->uv_offx2; |
| 764 | |
| 765 | YSCALEYUV2PACKEDX |
| 766 | /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */ |
| 767 | "psraw $3, %%mm3 \n\t" |
| 768 | "psraw $3, %%mm4 \n\t" |
| 769 | "psraw $3, %%mm1 \n\t" |
| 770 | "psraw $3, %%mm7 \n\t" |
| 771 | WRITEYUY2(%4, "%5", %%REGa) |
| 772 | YSCALEYUV2PACKEDX_END |
| 773 | } |
| 774 | |
| 775 | #define REAL_YSCALEYUV2RGB_UV(index, c) \ |
| 776 | "xor "#index", "#index" \n\t"\ |
| 777 | ".p2align 4 \n\t"\ |
| 778 | "1: \n\t"\ |
| 779 | "movq (%2, "#index"), %%mm2 \n\t" /* uvbuf0[eax]*/\ |
| 780 | "movq (%3, "#index"), %%mm3 \n\t" /* uvbuf1[eax]*/\ |
| 781 | "add "UV_OFF_BYTE"("#c"), "#index" \n\t" \ |
| 782 | "movq (%2, "#index"), %%mm5 \n\t" /* uvbuf0[eax+2048]*/\ |
| 783 | "movq (%3, "#index"), %%mm4 \n\t" /* uvbuf1[eax+2048]*/\ |
| 784 | "sub "UV_OFF_BYTE"("#c"), "#index" \n\t" \ |
| 785 | "psubw %%mm3, %%mm2 \n\t" /* uvbuf0[eax] - uvbuf1[eax]*/\ |
| 786 | "psubw %%mm4, %%mm5 \n\t" /* uvbuf0[eax+2048] - uvbuf1[eax+2048]*/\ |
| 787 | "movq "CHR_MMX_FILTER_OFFSET"+8("#c"), %%mm0 \n\t"\ |
| 788 | "pmulhw %%mm0, %%mm2 \n\t" /* (uvbuf0[eax] - uvbuf1[eax])uvalpha1>>16*/\ |
| 789 | "pmulhw %%mm0, %%mm5 \n\t" /* (uvbuf0[eax+2048] - uvbuf1[eax+2048])uvalpha1>>16*/\ |
| 790 | "psraw $4, %%mm3 \n\t" /* uvbuf0[eax] - uvbuf1[eax] >>4*/\ |
| 791 | "psraw $4, %%mm4 \n\t" /* uvbuf0[eax+2048] - uvbuf1[eax+2048] >>4*/\ |
| 792 | "paddw %%mm2, %%mm3 \n\t" /* uvbuf0[eax]uvalpha1 - uvbuf1[eax](1-uvalpha1)*/\ |
| 793 | "paddw %%mm5, %%mm4 \n\t" /* uvbuf0[eax+2048]uvalpha1 - uvbuf1[eax+2048](1-uvalpha1)*/\ |
| 794 | "psubw "U_OFFSET"("#c"), %%mm3 \n\t" /* (U-128)8*/\ |
| 795 | "psubw "V_OFFSET"("#c"), %%mm4 \n\t" /* (V-128)8*/\ |
| 796 | "movq %%mm3, %%mm2 \n\t" /* (U-128)8*/\ |
| 797 | "movq %%mm4, %%mm5 \n\t" /* (V-128)8*/\ |
| 798 | "pmulhw "UG_COEFF"("#c"), %%mm3 \n\t"\ |
| 799 | "pmulhw "VG_COEFF"("#c"), %%mm4 \n\t"\ |
| 800 | /* mm2=(U-128)8, mm3=ug, mm4=vg mm5=(V-128)8 */\ |
| 801 | |
| 802 | #define REAL_YSCALEYUV2RGB_YA(index, c, b1, b2) \ |
| 803 | "movq ("#b1", "#index", 2), %%mm0 \n\t" /*buf0[eax]*/\ |
| 804 | "movq ("#b2", "#index", 2), %%mm1 \n\t" /*buf1[eax]*/\ |
| 805 | "movq 8("#b1", "#index", 2), %%mm6 \n\t" /*buf0[eax]*/\ |
| 806 | "movq 8("#b2", "#index", 2), %%mm7 \n\t" /*buf1[eax]*/\ |
| 807 | "psubw %%mm1, %%mm0 \n\t" /* buf0[eax] - buf1[eax]*/\ |
| 808 | "psubw %%mm7, %%mm6 \n\t" /* buf0[eax] - buf1[eax]*/\ |
| 809 | "pmulhw "LUM_MMX_FILTER_OFFSET"+8("#c"), %%mm0 \n\t" /* (buf0[eax] - buf1[eax])yalpha1>>16*/\ |
| 810 | "pmulhw "LUM_MMX_FILTER_OFFSET"+8("#c"), %%mm6 \n\t" /* (buf0[eax] - buf1[eax])yalpha1>>16*/\ |
| 811 | "psraw $4, %%mm1 \n\t" /* buf0[eax] - buf1[eax] >>4*/\ |
| 812 | "psraw $4, %%mm7 \n\t" /* buf0[eax] - buf1[eax] >>4*/\ |
| 813 | "paddw %%mm0, %%mm1 \n\t" /* buf0[eax]yalpha1 + buf1[eax](1-yalpha1) >>16*/\ |
| 814 | "paddw %%mm6, %%mm7 \n\t" /* buf0[eax]yalpha1 + buf1[eax](1-yalpha1) >>16*/\ |
| 815 | |
| 816 | #define REAL_YSCALEYUV2RGB_COEFF(c) \ |
| 817 | "pmulhw "UB_COEFF"("#c"), %%mm2 \n\t"\ |
| 818 | "pmulhw "VR_COEFF"("#c"), %%mm5 \n\t"\ |
| 819 | "psubw "Y_OFFSET"("#c"), %%mm1 \n\t" /* 8(Y-16)*/\ |
| 820 | "psubw "Y_OFFSET"("#c"), %%mm7 \n\t" /* 8(Y-16)*/\ |
| 821 | "pmulhw "Y_COEFF"("#c"), %%mm1 \n\t"\ |
| 822 | "pmulhw "Y_COEFF"("#c"), %%mm7 \n\t"\ |
| 823 | /* mm1= Y1, mm2=ub, mm3=ug, mm4=vg mm5=vr, mm7=Y2 */\ |
| 824 | "paddw %%mm3, %%mm4 \n\t"\ |
| 825 | "movq %%mm2, %%mm0 \n\t"\ |
| 826 | "movq %%mm5, %%mm6 \n\t"\ |
| 827 | "movq %%mm4, %%mm3 \n\t"\ |
| 828 | "punpcklwd %%mm2, %%mm2 \n\t"\ |
| 829 | "punpcklwd %%mm5, %%mm5 \n\t"\ |
| 830 | "punpcklwd %%mm4, %%mm4 \n\t"\ |
| 831 | "paddw %%mm1, %%mm2 \n\t"\ |
| 832 | "paddw %%mm1, %%mm5 \n\t"\ |
| 833 | "paddw %%mm1, %%mm4 \n\t"\ |
| 834 | "punpckhwd %%mm0, %%mm0 \n\t"\ |
| 835 | "punpckhwd %%mm6, %%mm6 \n\t"\ |
| 836 | "punpckhwd %%mm3, %%mm3 \n\t"\ |
| 837 | "paddw %%mm7, %%mm0 \n\t"\ |
| 838 | "paddw %%mm7, %%mm6 \n\t"\ |
| 839 | "paddw %%mm7, %%mm3 \n\t"\ |
| 840 | /* mm0=B1, mm2=B2, mm3=G2, mm4=G1, mm5=R1, mm6=R2 */\ |
| 841 | "packuswb %%mm0, %%mm2 \n\t"\ |
| 842 | "packuswb %%mm6, %%mm5 \n\t"\ |
| 843 | "packuswb %%mm3, %%mm4 \n\t"\ |
| 844 | |
| 845 | #define YSCALEYUV2RGB_YA(index, c, b1, b2) REAL_YSCALEYUV2RGB_YA(index, c, b1, b2) |
| 846 | |
| 847 | #define YSCALEYUV2RGB(index, c) \ |
| 848 | REAL_YSCALEYUV2RGB_UV(index, c) \ |
| 849 | REAL_YSCALEYUV2RGB_YA(index, c, %0, %1) \ |
| 850 | REAL_YSCALEYUV2RGB_COEFF(c) |
| 851 | |
| 852 | /** |
| 853 | * vertical bilinear scale YV12 to RGB |
| 854 | */ |
| 855 | static void RENAME(yuv2rgb32_2)(SwsContext *c, const int16_t *buf[2], |
| 856 | const int16_t *ubuf[2], const int16_t *vbuf[2], |
| 857 | const int16_t *abuf[2], uint8_t *dest, |
| 858 | int dstW, int yalpha, int uvalpha, int y) |
| 859 | { |
| 860 | const int16_t *buf0 = buf[0], *buf1 = buf[1], |
| 861 | *ubuf0 = ubuf[0], *ubuf1 = ubuf[1]; |
| 862 | |
| 863 | if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) { |
| 864 | const int16_t *abuf0 = abuf[0], *abuf1 = abuf[1]; |
| 865 | #if ARCH_X86_64 |
| 866 | __asm__ volatile( |
| 867 | YSCALEYUV2RGB(%%r8, %5) |
| 868 | YSCALEYUV2RGB_YA(%%r8, %5, %6, %7) |
| 869 | "psraw $3, %%mm1 \n\t" /* abuf0[eax] - abuf1[eax] >>7*/ |
| 870 | "psraw $3, %%mm7 \n\t" /* abuf0[eax] - abuf1[eax] >>7*/ |
| 871 | "packuswb %%mm7, %%mm1 \n\t" |
| 872 | WRITEBGR32(%4, DSTW_OFFSET"(%5)", %%r8, %%mm2, %%mm4, %%mm5, %%mm1, %%mm0, %%mm7, %%mm3, %%mm6) |
| 873 | :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "r" (dest), |
| 874 | "a" (&c->redDither), |
| 875 | "r" (abuf0), "r" (abuf1) |
| 876 | : "%r8" |
| 877 | ); |
| 878 | #else |
| 879 | c->u_temp=(intptr_t)abuf0; |
| 880 | c->v_temp=(intptr_t)abuf1; |
| 881 | __asm__ volatile( |
| 882 | "mov %%"REG_b", "ESP_OFFSET"(%5) \n\t" |
| 883 | "mov %4, %%"REG_b" \n\t" |
| 884 | "push %%"REG_BP" \n\t" |
| 885 | YSCALEYUV2RGB(%%REGBP, %5) |
| 886 | "push %0 \n\t" |
| 887 | "push %1 \n\t" |
| 888 | "mov "U_TEMP"(%5), %0 \n\t" |
| 889 | "mov "V_TEMP"(%5), %1 \n\t" |
| 890 | YSCALEYUV2RGB_YA(%%REGBP, %5, %0, %1) |
| 891 | "psraw $3, %%mm1 \n\t" /* abuf0[eax] - abuf1[eax] >>7*/ |
| 892 | "psraw $3, %%mm7 \n\t" /* abuf0[eax] - abuf1[eax] >>7*/ |
| 893 | "packuswb %%mm7, %%mm1 \n\t" |
| 894 | "pop %1 \n\t" |
| 895 | "pop %0 \n\t" |
| 896 | WRITEBGR32(%%REGb, DSTW_OFFSET"(%5)", %%REGBP, %%mm2, %%mm4, %%mm5, %%mm1, %%mm0, %%mm7, %%mm3, %%mm6) |
| 897 | "pop %%"REG_BP" \n\t" |
| 898 | "mov "ESP_OFFSET"(%5), %%"REG_b" \n\t" |
| 899 | :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest), |
| 900 | "a" (&c->redDither) |
| 901 | ); |
| 902 | #endif |
| 903 | } else { |
| 904 | __asm__ volatile( |
| 905 | "mov %%"REG_b", "ESP_OFFSET"(%5) \n\t" |
| 906 | "mov %4, %%"REG_b" \n\t" |
| 907 | "push %%"REG_BP" \n\t" |
| 908 | YSCALEYUV2RGB(%%REGBP, %5) |
| 909 | "pcmpeqd %%mm7, %%mm7 \n\t" |
| 910 | WRITEBGR32(%%REGb, DSTW_OFFSET"(%5)", %%REGBP, %%mm2, %%mm4, %%mm5, %%mm7, %%mm0, %%mm1, %%mm3, %%mm6) |
| 911 | "pop %%"REG_BP" \n\t" |
| 912 | "mov "ESP_OFFSET"(%5), %%"REG_b" \n\t" |
| 913 | :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest), |
| 914 | "a" (&c->redDither) |
| 915 | ); |
| 916 | } |
| 917 | } |
| 918 | |
| 919 | static void RENAME(yuv2bgr24_2)(SwsContext *c, const int16_t *buf[2], |
| 920 | const int16_t *ubuf[2], const int16_t *vbuf[2], |
| 921 | const int16_t *abuf[2], uint8_t *dest, |
| 922 | int dstW, int yalpha, int uvalpha, int y) |
| 923 | { |
| 924 | const int16_t *buf0 = buf[0], *buf1 = buf[1], |
| 925 | *ubuf0 = ubuf[0], *ubuf1 = ubuf[1]; |
| 926 | |
| 927 | __asm__ volatile( |
| 928 | "mov %%"REG_b", "ESP_OFFSET"(%5) \n\t" |
| 929 | "mov %4, %%"REG_b" \n\t" |
| 930 | "push %%"REG_BP" \n\t" |
| 931 | YSCALEYUV2RGB(%%REGBP, %5) |
| 932 | "pxor %%mm7, %%mm7 \n\t" |
| 933 | WRITEBGR24(%%REGb, DSTW_OFFSET"(%5)", %%REGBP) |
| 934 | "pop %%"REG_BP" \n\t" |
| 935 | "mov "ESP_OFFSET"(%5), %%"REG_b" \n\t" |
| 936 | :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest), |
| 937 | "a" (&c->redDither) |
| 938 | NAMED_CONSTRAINTS_ADD(ff_M24A,ff_M24C,ff_M24B) |
| 939 | ); |
| 940 | } |
| 941 | |
| 942 | static void RENAME(yuv2rgb555_2)(SwsContext *c, const int16_t *buf[2], |
| 943 | const int16_t *ubuf[2], const int16_t *vbuf[2], |
| 944 | const int16_t *abuf[2], uint8_t *dest, |
| 945 | int dstW, int yalpha, int uvalpha, int y) |
| 946 | { |
| 947 | const int16_t *buf0 = buf[0], *buf1 = buf[1], |
| 948 | *ubuf0 = ubuf[0], *ubuf1 = ubuf[1]; |
| 949 | |
| 950 | __asm__ volatile( |
| 951 | "mov %%"REG_b", "ESP_OFFSET"(%5) \n\t" |
| 952 | "mov %4, %%"REG_b" \n\t" |
| 953 | "push %%"REG_BP" \n\t" |
| 954 | YSCALEYUV2RGB(%%REGBP, %5) |
| 955 | "pxor %%mm7, %%mm7 \n\t" |
| 956 | /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */ |
| 957 | #ifdef DITHER1XBPP |
| 958 | "paddusb "BLUE_DITHER"(%5), %%mm2 \n\t" |
| 959 | "paddusb "GREEN_DITHER"(%5), %%mm4 \n\t" |
| 960 | "paddusb "RED_DITHER"(%5), %%mm5 \n\t" |
| 961 | #endif |
| 962 | WRITERGB15(%%REGb, DSTW_OFFSET"(%5)", %%REGBP) |
| 963 | "pop %%"REG_BP" \n\t" |
| 964 | "mov "ESP_OFFSET"(%5), %%"REG_b" \n\t" |
| 965 | :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest), |
| 966 | "a" (&c->redDither) |
| 967 | NAMED_CONSTRAINTS_ADD(bF8) |
| 968 | ); |
| 969 | } |
| 970 | |
| 971 | static void RENAME(yuv2rgb565_2)(SwsContext *c, const int16_t *buf[2], |
| 972 | const int16_t *ubuf[2], const int16_t *vbuf[2], |
| 973 | const int16_t *abuf[2], uint8_t *dest, |
| 974 | int dstW, int yalpha, int uvalpha, int y) |
| 975 | { |
| 976 | const int16_t *buf0 = buf[0], *buf1 = buf[1], |
| 977 | *ubuf0 = ubuf[0], *ubuf1 = ubuf[1]; |
| 978 | |
| 979 | __asm__ volatile( |
| 980 | "mov %%"REG_b", "ESP_OFFSET"(%5) \n\t" |
| 981 | "mov %4, %%"REG_b" \n\t" |
| 982 | "push %%"REG_BP" \n\t" |
| 983 | YSCALEYUV2RGB(%%REGBP, %5) |
| 984 | "pxor %%mm7, %%mm7 \n\t" |
| 985 | /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */ |
| 986 | #ifdef DITHER1XBPP |
| 987 | "paddusb "BLUE_DITHER"(%5), %%mm2 \n\t" |
| 988 | "paddusb "GREEN_DITHER"(%5), %%mm4 \n\t" |
| 989 | "paddusb "RED_DITHER"(%5), %%mm5 \n\t" |
| 990 | #endif |
| 991 | WRITERGB16(%%REGb, DSTW_OFFSET"(%5)", %%REGBP) |
| 992 | "pop %%"REG_BP" \n\t" |
| 993 | "mov "ESP_OFFSET"(%5), %%"REG_b" \n\t" |
| 994 | :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest), |
| 995 | "a" (&c->redDither) |
| 996 | NAMED_CONSTRAINTS_ADD(bF8,bFC) |
| 997 | ); |
| 998 | } |
| 999 | |
| 1000 | #define REAL_YSCALEYUV2PACKED(index, c) \ |
| 1001 | "movq "CHR_MMX_FILTER_OFFSET"+8("#c"), %%mm0 \n\t"\ |
| 1002 | "movq "LUM_MMX_FILTER_OFFSET"+8("#c"), %%mm1 \n\t"\ |
| 1003 | "psraw $3, %%mm0 \n\t"\ |
| 1004 | "psraw $3, %%mm1 \n\t"\ |
| 1005 | "movq %%mm0, "CHR_MMX_FILTER_OFFSET"+8("#c") \n\t"\ |
| 1006 | "movq %%mm1, "LUM_MMX_FILTER_OFFSET"+8("#c") \n\t"\ |
| 1007 | "xor "#index", "#index" \n\t"\ |
| 1008 | ".p2align 4 \n\t"\ |
| 1009 | "1: \n\t"\ |
| 1010 | "movq (%2, "#index"), %%mm2 \n\t" /* uvbuf0[eax]*/\ |
| 1011 | "movq (%3, "#index"), %%mm3 \n\t" /* uvbuf1[eax]*/\ |
| 1012 | "add "UV_OFF_BYTE"("#c"), "#index" \n\t" \ |
| 1013 | "movq (%2, "#index"), %%mm5 \n\t" /* uvbuf0[eax+2048]*/\ |
| 1014 | "movq (%3, "#index"), %%mm4 \n\t" /* uvbuf1[eax+2048]*/\ |
| 1015 | "sub "UV_OFF_BYTE"("#c"), "#index" \n\t" \ |
| 1016 | "psubw %%mm3, %%mm2 \n\t" /* uvbuf0[eax] - uvbuf1[eax]*/\ |
| 1017 | "psubw %%mm4, %%mm5 \n\t" /* uvbuf0[eax+2048] - uvbuf1[eax+2048]*/\ |
| 1018 | "movq "CHR_MMX_FILTER_OFFSET"+8("#c"), %%mm0 \n\t"\ |
| 1019 | "pmulhw %%mm0, %%mm2 \n\t" /* (uvbuf0[eax] - uvbuf1[eax])uvalpha1>>16*/\ |
| 1020 | "pmulhw %%mm0, %%mm5 \n\t" /* (uvbuf0[eax+2048] - uvbuf1[eax+2048])uvalpha1>>16*/\ |
| 1021 | "psraw $7, %%mm3 \n\t" /* uvbuf0[eax] - uvbuf1[eax] >>4*/\ |
| 1022 | "psraw $7, %%mm4 \n\t" /* uvbuf0[eax+2048] - uvbuf1[eax+2048] >>4*/\ |
| 1023 | "paddw %%mm2, %%mm3 \n\t" /* uvbuf0[eax]uvalpha1 - uvbuf1[eax](1-uvalpha1)*/\ |
| 1024 | "paddw %%mm5, %%mm4 \n\t" /* uvbuf0[eax+2048]uvalpha1 - uvbuf1[eax+2048](1-uvalpha1)*/\ |
| 1025 | "movq (%0, "#index", 2), %%mm0 \n\t" /*buf0[eax]*/\ |
| 1026 | "movq (%1, "#index", 2), %%mm1 \n\t" /*buf1[eax]*/\ |
| 1027 | "movq 8(%0, "#index", 2), %%mm6 \n\t" /*buf0[eax]*/\ |
| 1028 | "movq 8(%1, "#index", 2), %%mm7 \n\t" /*buf1[eax]*/\ |
| 1029 | "psubw %%mm1, %%mm0 \n\t" /* buf0[eax] - buf1[eax]*/\ |
| 1030 | "psubw %%mm7, %%mm6 \n\t" /* buf0[eax] - buf1[eax]*/\ |
| 1031 | "pmulhw "LUM_MMX_FILTER_OFFSET"+8("#c"), %%mm0 \n\t" /* (buf0[eax] - buf1[eax])yalpha1>>16*/\ |
| 1032 | "pmulhw "LUM_MMX_FILTER_OFFSET"+8("#c"), %%mm6 \n\t" /* (buf0[eax] - buf1[eax])yalpha1>>16*/\ |
| 1033 | "psraw $7, %%mm1 \n\t" /* buf0[eax] - buf1[eax] >>4*/\ |
| 1034 | "psraw $7, %%mm7 \n\t" /* buf0[eax] - buf1[eax] >>4*/\ |
| 1035 | "paddw %%mm0, %%mm1 \n\t" /* buf0[eax]yalpha1 + buf1[eax](1-yalpha1) >>16*/\ |
| 1036 | "paddw %%mm6, %%mm7 \n\t" /* buf0[eax]yalpha1 + buf1[eax](1-yalpha1) >>16*/\ |
| 1037 | |
| 1038 | #define YSCALEYUV2PACKED(index, c) REAL_YSCALEYUV2PACKED(index, c) |
| 1039 | |
| 1040 | static void RENAME(yuv2yuyv422_2)(SwsContext *c, const int16_t *buf[2], |
| 1041 | const int16_t *ubuf[2], const int16_t *vbuf[2], |
| 1042 | const int16_t *abuf[2], uint8_t *dest, |
| 1043 | int dstW, int yalpha, int uvalpha, int y) |
| 1044 | { |
| 1045 | const int16_t *buf0 = buf[0], *buf1 = buf[1], |
| 1046 | *ubuf0 = ubuf[0], *ubuf1 = ubuf[1]; |
| 1047 | |
| 1048 | __asm__ volatile( |
| 1049 | "mov %%"REG_b", "ESP_OFFSET"(%5) \n\t" |
| 1050 | "mov %4, %%"REG_b" \n\t" |
| 1051 | "push %%"REG_BP" \n\t" |
| 1052 | YSCALEYUV2PACKED(%%REGBP, %5) |
| 1053 | WRITEYUY2(%%REGb, DSTW_OFFSET"(%5)", %%REGBP) |
| 1054 | "pop %%"REG_BP" \n\t" |
| 1055 | "mov "ESP_OFFSET"(%5), %%"REG_b" \n\t" |
| 1056 | :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest), |
| 1057 | "a" (&c->redDither) |
| 1058 | ); |
| 1059 | } |
| 1060 | |
| 1061 | #define REAL_YSCALEYUV2RGB1(index, c) \ |
| 1062 | "xor "#index", "#index" \n\t"\ |
| 1063 | ".p2align 4 \n\t"\ |
| 1064 | "1: \n\t"\ |
| 1065 | "movq (%2, "#index"), %%mm3 \n\t" /* uvbuf0[eax]*/\ |
| 1066 | "add "UV_OFF_BYTE"("#c"), "#index" \n\t" \ |
| 1067 | "movq (%2, "#index"), %%mm4 \n\t" /* uvbuf0[eax+2048]*/\ |
| 1068 | "sub "UV_OFF_BYTE"("#c"), "#index" \n\t" \ |
| 1069 | "psraw $4, %%mm3 \n\t" /* uvbuf0[eax] - uvbuf1[eax] >>4*/\ |
| 1070 | "psraw $4, %%mm4 \n\t" /* uvbuf0[eax+2048] - uvbuf1[eax+2048] >>4*/\ |
| 1071 | "psubw "U_OFFSET"("#c"), %%mm3 \n\t" /* (U-128)8*/\ |
| 1072 | "psubw "V_OFFSET"("#c"), %%mm4 \n\t" /* (V-128)8*/\ |
| 1073 | "movq %%mm3, %%mm2 \n\t" /* (U-128)8*/\ |
| 1074 | "movq %%mm4, %%mm5 \n\t" /* (V-128)8*/\ |
| 1075 | "pmulhw "UG_COEFF"("#c"), %%mm3 \n\t"\ |
| 1076 | "pmulhw "VG_COEFF"("#c"), %%mm4 \n\t"\ |
| 1077 | /* mm2=(U-128)8, mm3=ug, mm4=vg mm5=(V-128)8 */\ |
| 1078 | "movq (%0, "#index", 2), %%mm1 \n\t" /*buf0[eax]*/\ |
| 1079 | "movq 8(%0, "#index", 2), %%mm7 \n\t" /*buf0[eax]*/\ |
| 1080 | "psraw $4, %%mm1 \n\t" /* buf0[eax] - buf1[eax] >>4*/\ |
| 1081 | "psraw $4, %%mm7 \n\t" /* buf0[eax] - buf1[eax] >>4*/\ |
| 1082 | "pmulhw "UB_COEFF"("#c"), %%mm2 \n\t"\ |
| 1083 | "pmulhw "VR_COEFF"("#c"), %%mm5 \n\t"\ |
| 1084 | "psubw "Y_OFFSET"("#c"), %%mm1 \n\t" /* 8(Y-16)*/\ |
| 1085 | "psubw "Y_OFFSET"("#c"), %%mm7 \n\t" /* 8(Y-16)*/\ |
| 1086 | "pmulhw "Y_COEFF"("#c"), %%mm1 \n\t"\ |
| 1087 | "pmulhw "Y_COEFF"("#c"), %%mm7 \n\t"\ |
| 1088 | /* mm1= Y1, mm2=ub, mm3=ug, mm4=vg mm5=vr, mm7=Y2 */\ |
| 1089 | "paddw %%mm3, %%mm4 \n\t"\ |
| 1090 | "movq %%mm2, %%mm0 \n\t"\ |
| 1091 | "movq %%mm5, %%mm6 \n\t"\ |
| 1092 | "movq %%mm4, %%mm3 \n\t"\ |
| 1093 | "punpcklwd %%mm2, %%mm2 \n\t"\ |
| 1094 | "punpcklwd %%mm5, %%mm5 \n\t"\ |
| 1095 | "punpcklwd %%mm4, %%mm4 \n\t"\ |
| 1096 | "paddw %%mm1, %%mm2 \n\t"\ |
| 1097 | "paddw %%mm1, %%mm5 \n\t"\ |
| 1098 | "paddw %%mm1, %%mm4 \n\t"\ |
| 1099 | "punpckhwd %%mm0, %%mm0 \n\t"\ |
| 1100 | "punpckhwd %%mm6, %%mm6 \n\t"\ |
| 1101 | "punpckhwd %%mm3, %%mm3 \n\t"\ |
| 1102 | "paddw %%mm7, %%mm0 \n\t"\ |
| 1103 | "paddw %%mm7, %%mm6 \n\t"\ |
| 1104 | "paddw %%mm7, %%mm3 \n\t"\ |
| 1105 | /* mm0=B1, mm2=B2, mm3=G2, mm4=G1, mm5=R1, mm6=R2 */\ |
| 1106 | "packuswb %%mm0, %%mm2 \n\t"\ |
| 1107 | "packuswb %%mm6, %%mm5 \n\t"\ |
| 1108 | "packuswb %%mm3, %%mm4 \n\t"\ |
| 1109 | |
| 1110 | #define YSCALEYUV2RGB1(index, c) REAL_YSCALEYUV2RGB1(index, c) |
| 1111 | |
| 1112 | // do vertical chrominance interpolation |
| 1113 | #define REAL_YSCALEYUV2RGB1b(index, c) \ |
| 1114 | "xor "#index", "#index" \n\t"\ |
| 1115 | ".p2align 4 \n\t"\ |
| 1116 | "1: \n\t"\ |
| 1117 | "movq (%2, "#index"), %%mm2 \n\t" /* uvbuf0[eax]*/\ |
| 1118 | "movq (%3, "#index"), %%mm3 \n\t" /* uvbuf1[eax]*/\ |
| 1119 | "add "UV_OFF_BYTE"("#c"), "#index" \n\t" \ |
| 1120 | "movq (%2, "#index"), %%mm5 \n\t" /* uvbuf0[eax+2048]*/\ |
| 1121 | "movq (%3, "#index"), %%mm4 \n\t" /* uvbuf1[eax+2048]*/\ |
| 1122 | "sub "UV_OFF_BYTE"("#c"), "#index" \n\t" \ |
| 1123 | "paddw %%mm2, %%mm3 \n\t" /* uvbuf0[eax] + uvbuf1[eax]*/\ |
| 1124 | "paddw %%mm5, %%mm4 \n\t" /* uvbuf0[eax+2048] + uvbuf1[eax+2048]*/\ |
| 1125 | "psrlw $5, %%mm3 \n\t" /*FIXME might overflow*/\ |
| 1126 | "psrlw $5, %%mm4 \n\t" /*FIXME might overflow*/\ |
| 1127 | "psubw "U_OFFSET"("#c"), %%mm3 \n\t" /* (U-128)8*/\ |
| 1128 | "psubw "V_OFFSET"("#c"), %%mm4 \n\t" /* (V-128)8*/\ |
| 1129 | "movq %%mm3, %%mm2 \n\t" /* (U-128)8*/\ |
| 1130 | "movq %%mm4, %%mm5 \n\t" /* (V-128)8*/\ |
| 1131 | "pmulhw "UG_COEFF"("#c"), %%mm3 \n\t"\ |
| 1132 | "pmulhw "VG_COEFF"("#c"), %%mm4 \n\t"\ |
| 1133 | /* mm2=(U-128)8, mm3=ug, mm4=vg mm5=(V-128)8 */\ |
| 1134 | "movq (%0, "#index", 2), %%mm1 \n\t" /*buf0[eax]*/\ |
| 1135 | "movq 8(%0, "#index", 2), %%mm7 \n\t" /*buf0[eax]*/\ |
| 1136 | "psraw $4, %%mm1 \n\t" /* buf0[eax] - buf1[eax] >>4*/\ |
| 1137 | "psraw $4, %%mm7 \n\t" /* buf0[eax] - buf1[eax] >>4*/\ |
| 1138 | "pmulhw "UB_COEFF"("#c"), %%mm2 \n\t"\ |
| 1139 | "pmulhw "VR_COEFF"("#c"), %%mm5 \n\t"\ |
| 1140 | "psubw "Y_OFFSET"("#c"), %%mm1 \n\t" /* 8(Y-16)*/\ |
| 1141 | "psubw "Y_OFFSET"("#c"), %%mm7 \n\t" /* 8(Y-16)*/\ |
| 1142 | "pmulhw "Y_COEFF"("#c"), %%mm1 \n\t"\ |
| 1143 | "pmulhw "Y_COEFF"("#c"), %%mm7 \n\t"\ |
| 1144 | /* mm1= Y1, mm2=ub, mm3=ug, mm4=vg mm5=vr, mm7=Y2 */\ |
| 1145 | "paddw %%mm3, %%mm4 \n\t"\ |
| 1146 | "movq %%mm2, %%mm0 \n\t"\ |
| 1147 | "movq %%mm5, %%mm6 \n\t"\ |
| 1148 | "movq %%mm4, %%mm3 \n\t"\ |
| 1149 | "punpcklwd %%mm2, %%mm2 \n\t"\ |
| 1150 | "punpcklwd %%mm5, %%mm5 \n\t"\ |
| 1151 | "punpcklwd %%mm4, %%mm4 \n\t"\ |
| 1152 | "paddw %%mm1, %%mm2 \n\t"\ |
| 1153 | "paddw %%mm1, %%mm5 \n\t"\ |
| 1154 | "paddw %%mm1, %%mm4 \n\t"\ |
| 1155 | "punpckhwd %%mm0, %%mm0 \n\t"\ |
| 1156 | "punpckhwd %%mm6, %%mm6 \n\t"\ |
| 1157 | "punpckhwd %%mm3, %%mm3 \n\t"\ |
| 1158 | "paddw %%mm7, %%mm0 \n\t"\ |
| 1159 | "paddw %%mm7, %%mm6 \n\t"\ |
| 1160 | "paddw %%mm7, %%mm3 \n\t"\ |
| 1161 | /* mm0=B1, mm2=B2, mm3=G2, mm4=G1, mm5=R1, mm6=R2 */\ |
| 1162 | "packuswb %%mm0, %%mm2 \n\t"\ |
| 1163 | "packuswb %%mm6, %%mm5 \n\t"\ |
| 1164 | "packuswb %%mm3, %%mm4 \n\t"\ |
| 1165 | |
| 1166 | #define YSCALEYUV2RGB1b(index, c) REAL_YSCALEYUV2RGB1b(index, c) |
| 1167 | |
| 1168 | #define REAL_YSCALEYUV2RGB1_ALPHA(index) \ |
| 1169 | "movq (%1, "#index", 2), %%mm7 \n\t" /* abuf0[index ] */\ |
| 1170 | "movq 8(%1, "#index", 2), %%mm1 \n\t" /* abuf0[index+4] */\ |
| 1171 | "psraw $7, %%mm7 \n\t" /* abuf0[index ] >>7 */\ |
| 1172 | "psraw $7, %%mm1 \n\t" /* abuf0[index+4] >>7 */\ |
| 1173 | "packuswb %%mm1, %%mm7 \n\t" |
| 1174 | #define YSCALEYUV2RGB1_ALPHA(index) REAL_YSCALEYUV2RGB1_ALPHA(index) |
| 1175 | |
| 1176 | /** |
| 1177 | * YV12 to RGB without scaling or interpolating |
| 1178 | */ |
| 1179 | static void RENAME(yuv2rgb32_1)(SwsContext *c, const int16_t *buf0, |
| 1180 | const int16_t *ubuf[2], const int16_t *vbuf[2], |
| 1181 | const int16_t *abuf0, uint8_t *dest, |
| 1182 | int dstW, int uvalpha, int y) |
| 1183 | { |
| 1184 | const int16_t *ubuf0 = ubuf[0]; |
| 1185 | const int16_t *buf1= buf0; //FIXME needed for RGB1/BGR1 |
| 1186 | |
| 1187 | if (uvalpha < 2048) { // note this is not correct (shifts chrominance by 0.5 pixels) but it is a bit faster |
| 1188 | const int16_t *ubuf1 = ubuf[0]; |
| 1189 | if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) { |
| 1190 | __asm__ volatile( |
| 1191 | "mov %%"REG_b", "ESP_OFFSET"(%5) \n\t" |
| 1192 | "mov %4, %%"REG_b" \n\t" |
| 1193 | "push %%"REG_BP" \n\t" |
| 1194 | YSCALEYUV2RGB1(%%REGBP, %5) |
| 1195 | YSCALEYUV2RGB1_ALPHA(%%REGBP) |
| 1196 | WRITEBGR32(%%REGb, DSTW_OFFSET"(%5)", %%REGBP, %%mm2, %%mm4, %%mm5, %%mm7, %%mm0, %%mm1, %%mm3, %%mm6) |
| 1197 | "pop %%"REG_BP" \n\t" |
| 1198 | "mov "ESP_OFFSET"(%5), %%"REG_b" \n\t" |
| 1199 | :: "c" (buf0), "d" (abuf0), "S" (ubuf0), "D" (ubuf1), "m" (dest), |
| 1200 | "a" (&c->redDither) |
| 1201 | ); |
| 1202 | } else { |
| 1203 | __asm__ volatile( |
| 1204 | "mov %%"REG_b", "ESP_OFFSET"(%5) \n\t" |
| 1205 | "mov %4, %%"REG_b" \n\t" |
| 1206 | "push %%"REG_BP" \n\t" |
| 1207 | YSCALEYUV2RGB1(%%REGBP, %5) |
| 1208 | "pcmpeqd %%mm7, %%mm7 \n\t" |
| 1209 | WRITEBGR32(%%REGb, DSTW_OFFSET"(%5)", %%REGBP, %%mm2, %%mm4, %%mm5, %%mm7, %%mm0, %%mm1, %%mm3, %%mm6) |
| 1210 | "pop %%"REG_BP" \n\t" |
| 1211 | "mov "ESP_OFFSET"(%5), %%"REG_b" \n\t" |
| 1212 | :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest), |
| 1213 | "a" (&c->redDither) |
| 1214 | ); |
| 1215 | } |
| 1216 | } else { |
| 1217 | const int16_t *ubuf1 = ubuf[1]; |
| 1218 | if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) { |
| 1219 | __asm__ volatile( |
| 1220 | "mov %%"REG_b", "ESP_OFFSET"(%5) \n\t" |
| 1221 | "mov %4, %%"REG_b" \n\t" |
| 1222 | "push %%"REG_BP" \n\t" |
| 1223 | YSCALEYUV2RGB1b(%%REGBP, %5) |
| 1224 | YSCALEYUV2RGB1_ALPHA(%%REGBP) |
| 1225 | WRITEBGR32(%%REGb, DSTW_OFFSET"(%5)", %%REGBP, %%mm2, %%mm4, %%mm5, %%mm7, %%mm0, %%mm1, %%mm3, %%mm6) |
| 1226 | "pop %%"REG_BP" \n\t" |
| 1227 | "mov "ESP_OFFSET"(%5), %%"REG_b" \n\t" |
| 1228 | :: "c" (buf0), "d" (abuf0), "S" (ubuf0), "D" (ubuf1), "m" (dest), |
| 1229 | "a" (&c->redDither) |
| 1230 | ); |
| 1231 | } else { |
| 1232 | __asm__ volatile( |
| 1233 | "mov %%"REG_b", "ESP_OFFSET"(%5) \n\t" |
| 1234 | "mov %4, %%"REG_b" \n\t" |
| 1235 | "push %%"REG_BP" \n\t" |
| 1236 | YSCALEYUV2RGB1b(%%REGBP, %5) |
| 1237 | "pcmpeqd %%mm7, %%mm7 \n\t" |
| 1238 | WRITEBGR32(%%REGb, DSTW_OFFSET"(%5)", %%REGBP, %%mm2, %%mm4, %%mm5, %%mm7, %%mm0, %%mm1, %%mm3, %%mm6) |
| 1239 | "pop %%"REG_BP" \n\t" |
| 1240 | "mov "ESP_OFFSET"(%5), %%"REG_b" \n\t" |
| 1241 | :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest), |
| 1242 | "a" (&c->redDither) |
| 1243 | ); |
| 1244 | } |
| 1245 | } |
| 1246 | } |
| 1247 | |
| 1248 | static void RENAME(yuv2bgr24_1)(SwsContext *c, const int16_t *buf0, |
| 1249 | const int16_t *ubuf[2], const int16_t *vbuf[2], |
| 1250 | const int16_t *abuf0, uint8_t *dest, |
| 1251 | int dstW, int uvalpha, int y) |
| 1252 | { |
| 1253 | const int16_t *ubuf0 = ubuf[0]; |
| 1254 | const int16_t *buf1= buf0; //FIXME needed for RGB1/BGR1 |
| 1255 | |
| 1256 | if (uvalpha < 2048) { // note this is not correct (shifts chrominance by 0.5 pixels) but it is a bit faster |
| 1257 | const int16_t *ubuf1 = ubuf[0]; |
| 1258 | __asm__ volatile( |
| 1259 | "mov %%"REG_b", "ESP_OFFSET"(%5) \n\t" |
| 1260 | "mov %4, %%"REG_b" \n\t" |
| 1261 | "push %%"REG_BP" \n\t" |
| 1262 | YSCALEYUV2RGB1(%%REGBP, %5) |
| 1263 | "pxor %%mm7, %%mm7 \n\t" |
| 1264 | WRITEBGR24(%%REGb, DSTW_OFFSET"(%5)", %%REGBP) |
| 1265 | "pop %%"REG_BP" \n\t" |
| 1266 | "mov "ESP_OFFSET"(%5), %%"REG_b" \n\t" |
| 1267 | :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest), |
| 1268 | "a" (&c->redDither) |
| 1269 | NAMED_CONSTRAINTS_ADD(ff_M24A,ff_M24C,ff_M24B) |
| 1270 | ); |
| 1271 | } else { |
| 1272 | const int16_t *ubuf1 = ubuf[1]; |
| 1273 | __asm__ volatile( |
| 1274 | "mov %%"REG_b", "ESP_OFFSET"(%5) \n\t" |
| 1275 | "mov %4, %%"REG_b" \n\t" |
| 1276 | "push %%"REG_BP" \n\t" |
| 1277 | YSCALEYUV2RGB1b(%%REGBP, %5) |
| 1278 | "pxor %%mm7, %%mm7 \n\t" |
| 1279 | WRITEBGR24(%%REGb, DSTW_OFFSET"(%5)", %%REGBP) |
| 1280 | "pop %%"REG_BP" \n\t" |
| 1281 | "mov "ESP_OFFSET"(%5), %%"REG_b" \n\t" |
| 1282 | :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest), |
| 1283 | "a" (&c->redDither) |
| 1284 | NAMED_CONSTRAINTS_ADD(ff_M24A,ff_M24C,ff_M24B) |
| 1285 | ); |
| 1286 | } |
| 1287 | } |
| 1288 | |
| 1289 | static void RENAME(yuv2rgb555_1)(SwsContext *c, const int16_t *buf0, |
| 1290 | const int16_t *ubuf[2], const int16_t *vbuf[2], |
| 1291 | const int16_t *abuf0, uint8_t *dest, |
| 1292 | int dstW, int uvalpha, int y) |
| 1293 | { |
| 1294 | const int16_t *ubuf0 = ubuf[0]; |
| 1295 | const int16_t *buf1= buf0; //FIXME needed for RGB1/BGR1 |
| 1296 | |
| 1297 | if (uvalpha < 2048) { // note this is not correct (shifts chrominance by 0.5 pixels) but it is a bit faster |
| 1298 | const int16_t *ubuf1 = ubuf[0]; |
| 1299 | __asm__ volatile( |
| 1300 | "mov %%"REG_b", "ESP_OFFSET"(%5) \n\t" |
| 1301 | "mov %4, %%"REG_b" \n\t" |
| 1302 | "push %%"REG_BP" \n\t" |
| 1303 | YSCALEYUV2RGB1(%%REGBP, %5) |
| 1304 | "pxor %%mm7, %%mm7 \n\t" |
| 1305 | /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */ |
| 1306 | #ifdef DITHER1XBPP |
| 1307 | "paddusb "BLUE_DITHER"(%5), %%mm2 \n\t" |
| 1308 | "paddusb "GREEN_DITHER"(%5), %%mm4 \n\t" |
| 1309 | "paddusb "RED_DITHER"(%5), %%mm5 \n\t" |
| 1310 | #endif |
| 1311 | WRITERGB15(%%REGb, DSTW_OFFSET"(%5)", %%REGBP) |
| 1312 | "pop %%"REG_BP" \n\t" |
| 1313 | "mov "ESP_OFFSET"(%5), %%"REG_b" \n\t" |
| 1314 | :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest), |
| 1315 | "a" (&c->redDither) |
| 1316 | NAMED_CONSTRAINTS_ADD(bF8) |
| 1317 | ); |
| 1318 | } else { |
| 1319 | const int16_t *ubuf1 = ubuf[1]; |
| 1320 | __asm__ volatile( |
| 1321 | "mov %%"REG_b", "ESP_OFFSET"(%5) \n\t" |
| 1322 | "mov %4, %%"REG_b" \n\t" |
| 1323 | "push %%"REG_BP" \n\t" |
| 1324 | YSCALEYUV2RGB1b(%%REGBP, %5) |
| 1325 | "pxor %%mm7, %%mm7 \n\t" |
| 1326 | /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */ |
| 1327 | #ifdef DITHER1XBPP |
| 1328 | "paddusb "BLUE_DITHER"(%5), %%mm2 \n\t" |
| 1329 | "paddusb "GREEN_DITHER"(%5), %%mm4 \n\t" |
| 1330 | "paddusb "RED_DITHER"(%5), %%mm5 \n\t" |
| 1331 | #endif |
| 1332 | WRITERGB15(%%REGb, DSTW_OFFSET"(%5)", %%REGBP) |
| 1333 | "pop %%"REG_BP" \n\t" |
| 1334 | "mov "ESP_OFFSET"(%5), %%"REG_b" \n\t" |
| 1335 | :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest), |
| 1336 | "a" (&c->redDither) |
| 1337 | NAMED_CONSTRAINTS_ADD(bF8) |
| 1338 | ); |
| 1339 | } |
| 1340 | } |
| 1341 | |
| 1342 | static void RENAME(yuv2rgb565_1)(SwsContext *c, const int16_t *buf0, |
| 1343 | const int16_t *ubuf[2], const int16_t *vbuf[2], |
| 1344 | const int16_t *abuf0, uint8_t *dest, |
| 1345 | int dstW, int uvalpha, int y) |
| 1346 | { |
| 1347 | const int16_t *ubuf0 = ubuf[0]; |
| 1348 | const int16_t *buf1= buf0; //FIXME needed for RGB1/BGR1 |
| 1349 | |
| 1350 | if (uvalpha < 2048) { // note this is not correct (shifts chrominance by 0.5 pixels) but it is a bit faster |
| 1351 | const int16_t *ubuf1 = ubuf[0]; |
| 1352 | __asm__ volatile( |
| 1353 | "mov %%"REG_b", "ESP_OFFSET"(%5) \n\t" |
| 1354 | "mov %4, %%"REG_b" \n\t" |
| 1355 | "push %%"REG_BP" \n\t" |
| 1356 | YSCALEYUV2RGB1(%%REGBP, %5) |
| 1357 | "pxor %%mm7, %%mm7 \n\t" |
| 1358 | /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */ |
| 1359 | #ifdef DITHER1XBPP |
| 1360 | "paddusb "BLUE_DITHER"(%5), %%mm2 \n\t" |
| 1361 | "paddusb "GREEN_DITHER"(%5), %%mm4 \n\t" |
| 1362 | "paddusb "RED_DITHER"(%5), %%mm5 \n\t" |
| 1363 | #endif |
| 1364 | WRITERGB16(%%REGb, DSTW_OFFSET"(%5)", %%REGBP) |
| 1365 | "pop %%"REG_BP" \n\t" |
| 1366 | "mov "ESP_OFFSET"(%5), %%"REG_b" \n\t" |
| 1367 | :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest), |
| 1368 | "a" (&c->redDither) |
| 1369 | NAMED_CONSTRAINTS_ADD(bF8,bFC) |
| 1370 | ); |
| 1371 | } else { |
| 1372 | const int16_t *ubuf1 = ubuf[1]; |
| 1373 | __asm__ volatile( |
| 1374 | "mov %%"REG_b", "ESP_OFFSET"(%5) \n\t" |
| 1375 | "mov %4, %%"REG_b" \n\t" |
| 1376 | "push %%"REG_BP" \n\t" |
| 1377 | YSCALEYUV2RGB1b(%%REGBP, %5) |
| 1378 | "pxor %%mm7, %%mm7 \n\t" |
| 1379 | /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */ |
| 1380 | #ifdef DITHER1XBPP |
| 1381 | "paddusb "BLUE_DITHER"(%5), %%mm2 \n\t" |
| 1382 | "paddusb "GREEN_DITHER"(%5), %%mm4 \n\t" |
| 1383 | "paddusb "RED_DITHER"(%5), %%mm5 \n\t" |
| 1384 | #endif |
| 1385 | WRITERGB16(%%REGb, DSTW_OFFSET"(%5)", %%REGBP) |
| 1386 | "pop %%"REG_BP" \n\t" |
| 1387 | "mov "ESP_OFFSET"(%5), %%"REG_b" \n\t" |
| 1388 | :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest), |
| 1389 | "a" (&c->redDither) |
| 1390 | NAMED_CONSTRAINTS_ADD(bF8,bFC) |
| 1391 | ); |
| 1392 | } |
| 1393 | } |
| 1394 | |
| 1395 | #define REAL_YSCALEYUV2PACKED1(index, c) \ |
| 1396 | "xor "#index", "#index" \n\t"\ |
| 1397 | ".p2align 4 \n\t"\ |
| 1398 | "1: \n\t"\ |
| 1399 | "movq (%2, "#index"), %%mm3 \n\t" /* uvbuf0[eax]*/\ |
| 1400 | "add "UV_OFF_BYTE"("#c"), "#index" \n\t" \ |
| 1401 | "movq (%2, "#index"), %%mm4 \n\t" /* uvbuf0[eax+2048]*/\ |
| 1402 | "sub "UV_OFF_BYTE"("#c"), "#index" \n\t" \ |
| 1403 | "psraw $7, %%mm3 \n\t" \ |
| 1404 | "psraw $7, %%mm4 \n\t" \ |
| 1405 | "movq (%0, "#index", 2), %%mm1 \n\t" /*buf0[eax]*/\ |
| 1406 | "movq 8(%0, "#index", 2), %%mm7 \n\t" /*buf0[eax]*/\ |
| 1407 | "psraw $7, %%mm1 \n\t" \ |
| 1408 | "psraw $7, %%mm7 \n\t" \ |
| 1409 | |
| 1410 | #define YSCALEYUV2PACKED1(index, c) REAL_YSCALEYUV2PACKED1(index, c) |
| 1411 | |
| 1412 | #define REAL_YSCALEYUV2PACKED1b(index, c) \ |
| 1413 | "xor "#index", "#index" \n\t"\ |
| 1414 | ".p2align 4 \n\t"\ |
| 1415 | "1: \n\t"\ |
| 1416 | "movq (%2, "#index"), %%mm2 \n\t" /* uvbuf0[eax]*/\ |
| 1417 | "movq (%3, "#index"), %%mm3 \n\t" /* uvbuf1[eax]*/\ |
| 1418 | "add "UV_OFF_BYTE"("#c"), "#index" \n\t" \ |
| 1419 | "movq (%2, "#index"), %%mm5 \n\t" /* uvbuf0[eax+2048]*/\ |
| 1420 | "movq (%3, "#index"), %%mm4 \n\t" /* uvbuf1[eax+2048]*/\ |
| 1421 | "sub "UV_OFF_BYTE"("#c"), "#index" \n\t" \ |
| 1422 | "paddw %%mm2, %%mm3 \n\t" /* uvbuf0[eax] + uvbuf1[eax]*/\ |
| 1423 | "paddw %%mm5, %%mm4 \n\t" /* uvbuf0[eax+2048] + uvbuf1[eax+2048]*/\ |
| 1424 | "psrlw $8, %%mm3 \n\t" \ |
| 1425 | "psrlw $8, %%mm4 \n\t" \ |
| 1426 | "movq (%0, "#index", 2), %%mm1 \n\t" /*buf0[eax]*/\ |
| 1427 | "movq 8(%0, "#index", 2), %%mm7 \n\t" /*buf0[eax]*/\ |
| 1428 | "psraw $7, %%mm1 \n\t" \ |
| 1429 | "psraw $7, %%mm7 \n\t" |
| 1430 | #define YSCALEYUV2PACKED1b(index, c) REAL_YSCALEYUV2PACKED1b(index, c) |
| 1431 | |
| 1432 | static void RENAME(yuv2yuyv422_1)(SwsContext *c, const int16_t *buf0, |
| 1433 | const int16_t *ubuf[2], const int16_t *vbuf[2], |
| 1434 | const int16_t *abuf0, uint8_t *dest, |
| 1435 | int dstW, int uvalpha, int y) |
| 1436 | { |
| 1437 | const int16_t *ubuf0 = ubuf[0]; |
| 1438 | const int16_t *buf1= buf0; //FIXME needed for RGB1/BGR1 |
| 1439 | |
| 1440 | if (uvalpha < 2048) { // note this is not correct (shifts chrominance by 0.5 pixels) but it is a bit faster |
| 1441 | const int16_t *ubuf1 = ubuf[0]; |
| 1442 | __asm__ volatile( |
| 1443 | "mov %%"REG_b", "ESP_OFFSET"(%5) \n\t" |
| 1444 | "mov %4, %%"REG_b" \n\t" |
| 1445 | "push %%"REG_BP" \n\t" |
| 1446 | YSCALEYUV2PACKED1(%%REGBP, %5) |
| 1447 | WRITEYUY2(%%REGb, DSTW_OFFSET"(%5)", %%REGBP) |
| 1448 | "pop %%"REG_BP" \n\t" |
| 1449 | "mov "ESP_OFFSET"(%5), %%"REG_b" \n\t" |
| 1450 | :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest), |
| 1451 | "a" (&c->redDither) |
| 1452 | ); |
| 1453 | } else { |
| 1454 | const int16_t *ubuf1 = ubuf[1]; |
| 1455 | __asm__ volatile( |
| 1456 | "mov %%"REG_b", "ESP_OFFSET"(%5) \n\t" |
| 1457 | "mov %4, %%"REG_b" \n\t" |
| 1458 | "push %%"REG_BP" \n\t" |
| 1459 | YSCALEYUV2PACKED1b(%%REGBP, %5) |
| 1460 | WRITEYUY2(%%REGb, DSTW_OFFSET"(%5)", %%REGBP) |
| 1461 | "pop %%"REG_BP" \n\t" |
| 1462 | "mov "ESP_OFFSET"(%5), %%"REG_b" \n\t" |
| 1463 | :: "c" (buf0), "d" (buf1), "S" (ubuf0), "D" (ubuf1), "m" (dest), |
| 1464 | "a" (&c->redDither) |
| 1465 | ); |
| 1466 | } |
| 1467 | } |
| 1468 | static av_cold void RENAME(sws_init_swscale)(SwsContext *c) |
| 1469 | { |
| 1470 | enum AVPixelFormat dstFormat = c->dstFormat; |
| 1471 | |
| 1472 | c->use_mmx_vfilter= 0; |
| 1473 | if (!is16BPS(dstFormat) && !is9_OR_10BPS(dstFormat) && dstFormat != AV_PIX_FMT_NV12 |
| 1474 | && dstFormat != AV_PIX_FMT_NV21 && !(c->flags & SWS_BITEXACT)) { |
| 1475 | if (c->flags & SWS_ACCURATE_RND) { |
| 1476 | if (!(c->flags & SWS_FULL_CHR_H_INT)) { |
| 1477 | switch (c->dstFormat) { |
| 1478 | case AV_PIX_FMT_RGB32: c->yuv2packedX = RENAME(yuv2rgb32_X_ar); break; |
| 1479 | #if HAVE_6REGS |
| 1480 | case AV_PIX_FMT_BGR24: c->yuv2packedX = RENAME(yuv2bgr24_X_ar); break; |
| 1481 | #endif |
| 1482 | case AV_PIX_FMT_RGB555: c->yuv2packedX = RENAME(yuv2rgb555_X_ar); break; |
| 1483 | case AV_PIX_FMT_RGB565: c->yuv2packedX = RENAME(yuv2rgb565_X_ar); break; |
| 1484 | case AV_PIX_FMT_YUYV422: c->yuv2packedX = RENAME(yuv2yuyv422_X_ar); break; |
| 1485 | default: break; |
| 1486 | } |
| 1487 | } |
| 1488 | } else { |
| 1489 | c->use_mmx_vfilter= 1; |
| 1490 | c->yuv2planeX = RENAME(yuv2yuvX ); |
| 1491 | if (!(c->flags & SWS_FULL_CHR_H_INT)) { |
| 1492 | switch (c->dstFormat) { |
| 1493 | case AV_PIX_FMT_RGB32: c->yuv2packedX = RENAME(yuv2rgb32_X); break; |
| 1494 | #if HAVE_6REGS |
| 1495 | case AV_PIX_FMT_BGR24: c->yuv2packedX = RENAME(yuv2bgr24_X); break; |
| 1496 | #endif |
| 1497 | case AV_PIX_FMT_RGB555: c->yuv2packedX = RENAME(yuv2rgb555_X); break; |
| 1498 | case AV_PIX_FMT_RGB565: c->yuv2packedX = RENAME(yuv2rgb565_X); break; |
| 1499 | case AV_PIX_FMT_YUYV422: c->yuv2packedX = RENAME(yuv2yuyv422_X); break; |
| 1500 | default: break; |
| 1501 | } |
| 1502 | } |
| 1503 | } |
| 1504 | if (!(c->flags & SWS_FULL_CHR_H_INT)) { |
| 1505 | switch (c->dstFormat) { |
| 1506 | case AV_PIX_FMT_RGB32: |
| 1507 | c->yuv2packed1 = RENAME(yuv2rgb32_1); |
| 1508 | c->yuv2packed2 = RENAME(yuv2rgb32_2); |
| 1509 | break; |
| 1510 | case AV_PIX_FMT_BGR24: |
| 1511 | c->yuv2packed1 = RENAME(yuv2bgr24_1); |
| 1512 | c->yuv2packed2 = RENAME(yuv2bgr24_2); |
| 1513 | break; |
| 1514 | case AV_PIX_FMT_RGB555: |
| 1515 | c->yuv2packed1 = RENAME(yuv2rgb555_1); |
| 1516 | c->yuv2packed2 = RENAME(yuv2rgb555_2); |
| 1517 | break; |
| 1518 | case AV_PIX_FMT_RGB565: |
| 1519 | c->yuv2packed1 = RENAME(yuv2rgb565_1); |
| 1520 | c->yuv2packed2 = RENAME(yuv2rgb565_2); |
| 1521 | break; |
| 1522 | case AV_PIX_FMT_YUYV422: |
| 1523 | c->yuv2packed1 = RENAME(yuv2yuyv422_1); |
| 1524 | c->yuv2packed2 = RENAME(yuv2yuyv422_2); |
| 1525 | break; |
| 1526 | default: |
| 1527 | break; |
| 1528 | } |
| 1529 | } |
| 1530 | } |
| 1531 | |
| 1532 | if (c->srcBpc == 8 && c->dstBpc <= 14) { |
| 1533 | // Use the new MMX scaler if the MMXEXT one can't be used (it is faster than the x86 ASM one). |
| 1534 | #if COMPILE_TEMPLATE_MMXEXT |
| 1535 | if (c->flags & SWS_FAST_BILINEAR && c->canMMXEXTBeUsed) { |
| 1536 | c->hyscale_fast = ff_hyscale_fast_mmxext; |
| 1537 | c->hcscale_fast = ff_hcscale_fast_mmxext; |
| 1538 | } else { |
| 1539 | #endif /* COMPILE_TEMPLATE_MMXEXT */ |
| 1540 | c->hyscale_fast = NULL; |
| 1541 | c->hcscale_fast = NULL; |
| 1542 | #if COMPILE_TEMPLATE_MMXEXT |
| 1543 | } |
| 1544 | #endif /* COMPILE_TEMPLATE_MMXEXT */ |
| 1545 | } |
| 1546 | } |