1;****************************************************************************** 2;* software YUV to RGB converter 3;* 4;* Copyright (C) 2001-2007 Michael Niedermayer 5;* (c) 2010 Konstantin Shishkov 6;* 7;* This file is part of FFmpeg. 8;* 9;* FFmpeg is free software; you can redistribute it and/or 10;* modify it under the terms of the GNU Lesser General Public 11;* License as published by the Free Software Foundation; either 12;* version 2.1 of the License, or (at your option) any later version. 13;* 14;* FFmpeg is distributed in the hope that it will be useful, 15;* but WITHOUT ANY WARRANTY; without even the implied warranty of 16;* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU 17;* Lesser General Public License for more details. 18;* 19;* You should have received a copy of the GNU Lesser General Public 20;* License along with FFmpeg; if not, write to the Free Software 21;* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA 22;****************************************************************************** 23 24%include "libavutil/x86/x86util.asm" 25 26SECTION_RODATA 27 28; below variables are named like mask_dwXY, which means to preserve dword No.X & No.Y 29mask_dw036 : db -1, -1, 0, 0, 0, 0, -1, -1, 0, 0, 0, 0, -1, -1, 0, 0 30mask_dw147 : db 0, 0, -1, -1, 0, 0, 0, 0, -1, -1, 0, 0, 0, 0, -1, -1 31mask_dw25 : db 0, 0, 0, 0, -1, -1, 0, 0, 0, 0, -1, -1, 0, 0, 0, 0 32rgb24_shuf1: db 0, 1, 6, 7, 12, 13, 2, 3, 8, 9, 14, 15, 4, 5, 10, 11 33rgb24_shuf2: db 10, 11, 0, 1, 6, 7, 12, 13, 2, 3, 8, 9, 14, 15, 4, 5 34rgb24_shuf3: db 4, 5, 10, 11, 0, 1, 6, 7, 12, 13, 2, 3, 8, 9, 14, 15 35pw_00ff: times 8 dw 255 36pb_f8: times 16 db 248 37pb_e0: times 16 db 224 38pb_03: times 16 db 3 39pb_07: times 16 db 7 40 41mask_1101: dw -1, -1, 0, -1 42mask_0010: dw 0, 0, -1, 0 43mask_0110: dw 0, -1, -1, 0 44mask_1001: dw -1, 0, 0, -1 45mask_0100: dw 0, -1, 0, 0 46 47SECTION .text 48 49;----------------------------------------------------------------------------- 50; 51; YUV420/YUVA420 to RGB/BGR 15/16/24/32 52; R = Y + ((vrCoff * (v - 128)) >> 8) 53; G = Y - ((ugCoff * (u - 128) + vgCoff * (v - 128)) >> 8) 54; B = Y + ((ubCoff * (u - 128)) >> 8) 55; 56;----------------------------------------------------------------------------- 57 58%macro MOV_H2L 1 59%if mmsize == 8 60 psrlq %1, 32 61%else ; mmsize == 16 62 psrldq %1, 8 63%endif 64%endmacro 65 66%macro yuv2rgb_fn 3 67 68%if %3 == 32 69 %ifidn %1, yuva 70 %define parameters index, image, pu_index, pv_index, pointer_c_dither, py_2index, pa_2index 71 %define GPR_num 7 72 %endif 73%else 74 %define parameters index, image, pu_index, pv_index, pointer_c_dither, py_2index 75 %define GPR_num 6 76%endif 77 78%define m_green m2 79%define m_alpha m3 80%define m_y m6 81%define m_u m0 82%define m_v m1 83%ifidn %2, rgb 84%define m_red m1 85%define m_blue m0 86%else 87%define m_red m0 88%define m_blue m1 89%endif 90 91%if mmsize == 8 92%define time_num 1 93%define reg_num 8 94%define y_offset [pointer_c_ditherq + 8 * 8] 95%define u_offset [pointer_c_ditherq + 9 * 8] 96%define v_offset [pointer_c_ditherq + 10 * 8] 97%define ug_coff [pointer_c_ditherq + 7 * 8] 98%define vg_coff [pointer_c_ditherq + 6 * 8] 99%define y_coff [pointer_c_ditherq + 3 * 8] 100%define ub_coff [pointer_c_ditherq + 5 * 8] 101%define vr_coff [pointer_c_ditherq + 4 * 8] 102%elif mmsize == 16 103%define time_num 2 104%if ARCH_X86_32 105%define reg_num 8 106%define my_offset [pointer_c_ditherq + 8 * 8] 107%define mu_offset [pointer_c_ditherq + 9 * 8] 108%define mv_offset [pointer_c_ditherq + 10 * 8] 109%define mug_coff [pointer_c_ditherq + 7 * 8] 110%define mvg_coff [pointer_c_ditherq + 6 * 8] 111%define my_coff [pointer_c_ditherq + 3 * 8] 112%define mub_coff [pointer_c_ditherq + 5 * 8] 113%define mvr_coff [pointer_c_ditherq + 4 * 8] 114%else ; ARCH_X86_64 115%define reg_num 16 116%define y_offset m8 117%define u_offset m9 118%define v_offset m10 119%define ug_coff m11 120%define vg_coff m12 121%define y_coff m13 122%define ub_coff m14 123%define vr_coff m15 124%endif ; ARCH_X86_32/64 125%endif ; coeff define mmsize == 8/16 126 127cglobal %1_420_%2%3, GPR_num, GPR_num, reg_num, parameters 128 129%if ARCH_X86_64 130 movsxd indexq, indexd 131%if mmsize == 16 132 VBROADCASTSD y_offset, [pointer_c_ditherq + 8 * 8] 133 VBROADCASTSD u_offset, [pointer_c_ditherq + 9 * 8] 134 VBROADCASTSD v_offset, [pointer_c_ditherq + 10 * 8] 135 VBROADCASTSD ug_coff, [pointer_c_ditherq + 7 * 8] 136 VBROADCASTSD vg_coff, [pointer_c_ditherq + 6 * 8] 137 VBROADCASTSD y_coff, [pointer_c_ditherq + 3 * 8] 138 VBROADCASTSD ub_coff, [pointer_c_ditherq + 5 * 8] 139 VBROADCASTSD vr_coff, [pointer_c_ditherq + 4 * 8] 140%endif 141%endif 142 movu m_y, [py_2indexq + 2 * indexq] 143 movh m_u, [pu_indexq + indexq] 144 movh m_v, [pv_indexq + indexq] 145.loop0: 146 pxor m4, m4 147 mova m7, m6 148 punpcklbw m0, m4 149 punpcklbw m1, m4 150 mova m2, [pw_00ff] 151 pand m6, m2 152 psrlw m7, 8 153 psllw m0, 3 154 psllw m1, 3 155 psllw m6, 3 156 psllw m7, 3 157%if (ARCH_X86_32 && mmsize == 16) 158 VBROADCASTSD m2, mu_offset 159 VBROADCASTSD m3, mv_offset 160 VBROADCASTSD m4, my_offset 161 psubsw m0, m2 ; U = U - 128 162 psubsw m1, m3 ; V = V - 128 163 psubw m6, m4 164 psubw m7, m4 165 VBROADCASTSD m2, mug_coff 166 VBROADCASTSD m3, mvg_coff 167 VBROADCASTSD m4, my_coff 168 VBROADCASTSD m5, mub_coff 169 pmulhw m2, m0 170 pmulhw m3, m1 171 pmulhw m6, m4 172 pmulhw m7, m4 173 pmulhw m0, m5 174 VBROADCASTSD m4, mvr_coff 175 pmulhw m1, m4 176%else ; ARCH_X86_64 || mmsize == 8 177 psubsw m0, u_offset ; U = U - 128 178 psubsw m1, v_offset ; V = V - 128 179 psubw m6, y_offset 180 psubw m7, y_offset 181 mova m2, m0 182 mova m3, m1 183 pmulhw m2, ug_coff 184 pmulhw m3, vg_coff 185 pmulhw m6, y_coff 186 pmulhw m7, y_coff 187 pmulhw m0, ub_coff 188 pmulhw m1, vr_coff 189%endif 190 paddsw m2, m3 191 mova m3, m7 192 mova m5, m7 193 paddsw m3, m0 ; B1 B3 B5 B7 ... 194 paddsw m5, m1 ; R1 R3 R5 R7 ... 195 paddsw m7, m2 ; G1 G3 G4 G7 ... 196 paddsw m0, m6 ; B0 B2 B4 B6 ... 197 paddsw m1, m6 ; R0 R2 R4 R6 ... 198 paddsw m2, m6 ; G0 G2 G4 G6 ... 199 200%if %3 == 24 ; PACK RGB24 201%define depth 3 202 packuswb m0, m3 ; R0 R2 R4 R6 ... R1 R3 R5 R7 ... 203 packuswb m1, m5 ; B0 B2 B4 B6 ... B1 B3 B5 B7 ... 204 packuswb m2, m7 ; G0 G2 G4 G6 ... G1 G3 G5 G7 ... 205 mova m3, m_red 206 mova m6, m_blue 207 MOV_H2L m_red 208 punpcklbw m3, m2 ; R0 G0 R2 G2 R4 G4 R6 G6 R8 G8 ... 209 punpcklbw m6, m_red ; B0 R1 B2 R3 B4 R5 B6 R7 B8 R9 ... 210 mova m5, m3 211 punpckhbw m2, m_blue ; G1 B1 G3 B3 G5 B5 G7 B7 G9 B9 ... 212%if mmsize == 8 213 punpcklwd m3 ,m6 ; R0 G0 B0 R1 R2 G2 B2 R3 214 punpckhwd m5, m6 ; R4 G4 B4 R5 R6 G6 B6 R7 215%if cpuflag(mmxext) 216 pshufw m1, m2, 0xc6 217 pshufw m6, m3, 0x84 218 pshufw m7, m5, 0x38 219 pand m6, [mask_1101] ; R0 G0 B0 R1 -- -- R2 G2 220 movq m0, m1 221 pand m7, [mask_0110] ; -- -- R6 G6 B6 R7 -- -- 222 movq m2, m1 223 pand m1, [mask_0100] ; -- -- G3 B3 -- -- -- -- 224 psrlq m3, 48 ; B2 R3 -- -- -- -- -- -- 225 pand m0, [mask_0010] ; -- -- -- -- G1 B1 -- -- 226 psllq m5, 32 ; -- -- -- -- R4 G4 B4 R5 227 pand m2, [mask_1001] ; G5 B5 -- -- -- -- G7 B7 228 por m1, m3 229 por m0, m6 230 por m1, m5 231 por m2, m7 232 movntq [imageq], m0 233 movntq [imageq + 8], m1 234 movntq [imageq + 16], m2 235%else ; cpuflag(mmx) 236 movd [imageq], m3 ; R0 G0 R2 G2 237 movd [imageq + 4], m2 ; G1 B1 238 psrlq m3, 32 239 psrlq m2, 16 240 movd [imageq + 6], m3 ; R2 G2 B2 R3 241 movd [imageq + 10], m2 ; G3 B3 242 psrlq m2, 16 243 movd [imageq + 12], m5 ; R4 G4 B4 R5 244 movd [imageq + 16], m2 ; G5 B5 245 psrlq m5, 32 246 movd [imageq + 20], m2 ; -- -- G7 B7 247 movd [imageq + 18], m5 ; R6 G6 B6 R7 248%endif ; mmsize = 8 249%else ; mmsize == 16 250 pshufb m3, [rgb24_shuf1] ; r0 g0 r6 g6 r12 g12 r2 g2 r8 g8 r14 g14 r4 g4 r10 g10 251 pshufb m6, [rgb24_shuf2] ; b10 r11 b0 r1 b6 r7 b12 r13 b2 r3 b8 r9 b14 r15 b4 r5 252 pshufb m2, [rgb24_shuf3] ; g5 b5 g11 b11 g1 b1 g7 b7 g13 b13 g3 b3 g9 b9 g15 b15 253 mova m7, [mask_dw036] 254 mova m4, [mask_dw147] 255 mova m5, [mask_dw25] 256 pand m0, m7, m3 ; r0 g0 --- --- --- --- r2 g2 --- --- --- --- r4 g4 --- --- 257 pand m1, m4, m6 ; --- --- b0 r1 --- --- --- --- b2 r3 --- --- --- --- b4 r5 258 por m0, m1 259 pand m1, m5, m2 ; --- --- --- --- g1 b1 --- --- --- --- g3 b3 --- --- --- --- 260 por m0, m1 ; r0 g0 b0 r1 g1 b1 r2 g2 b2 r3 g3 b3 r4 g4 b4 r5 261 pand m1, m7, m2 ; g5 b5 --- --- --- --- g7 b7 --- --- --- --- g9 b9 --- --- 262 pand m7, m6 ; b10 r11 --- --- --- --- b12 r13 --- --- --- --- b14 r15 --- --- 263 pand m6, m5 ; --- --- --- --- b6 r7 --- --- --- --- b8 r9 --- --- --- --- 264 por m1, m6 265 pand m6, m4, m3 ; --- --- r6 g6 --- --- --- --- r8 g8 --- --- --- --- r10 g10 266 pand m2, m4 ; --- --- g11 b11 --- --- --- --- g13 b13 --- --- --- --- g15 b15 267 pand m3, m5 ; --- --- --- --- r12 g12 --- --- --- --- r14 g14 --- --- --- --- 268 por m2, m7 269 por m1, m6 ; g5 b5 r6 g6 b6 r7 g7 b7 r8 g8 b8 r9 g9 b9 r10 g10 270 por m2, m3 ; b10 r11 g11 b11 r12 g12 b12 r13 g13 b13 r14 g14 b14 r15 g15 b15 271 movu [imageq], m0 272 movu [imageq + 16], m1 273 movu [imageq + 32], m2 274%endif ; mmsize = 16 275%else ; PACK RGB15/16/32 276 packuswb m0, m1 277 packuswb m3, m5 278 packuswb m2, m2 279 mova m1, m0 280 packuswb m7, m7 281 punpcklbw m0, m3 ; B0 B1 B2 B3 ... B7 282 punpckhbw m1, m3 ; R0 R1 R2 R3 ... R7 283 punpcklbw m2, m7 ; G0 G1 G2 G3 ... G7 284%if %3 == 32 ; PACK RGB32 285%define depth 4 286%ifidn %1, yuv 287 pcmpeqd m3, m3 ; Set alpha empty 288%else 289 movu m3, [pa_2indexq + 2 * indexq] ; Load alpha 290%endif 291 mova m5, m_blue 292 mova m6, m_red 293 punpckhbw m5, m_green 294 punpcklbw m_blue, m_green 295 punpckhbw m6, m_alpha 296 punpcklbw m_red, m_alpha 297 mova m_green, m_blue 298 mova m_alpha, m5 299 punpcklwd m_blue, m_red 300 punpckhwd m_green, m_red 301 punpcklwd m5, m6 302 punpckhwd m_alpha, m6 303 movu [imageq + 0], m_blue 304 movu [imageq + 8 * time_num], m_green 305 movu [imageq + 16 * time_num], m5 306 movu [imageq + 24 * time_num], m_alpha 307%else ; PACK RGB15/16 308%define depth 2 309%if cpuflag(ssse3) 310 %define red_dither m3 311 %define green_dither m4 312 %define blue_dither m5 313 VBROADCASTSD red_dither, [pointer_c_ditherq + 0 * 8] 314 VBROADCASTSD green_dither, [pointer_c_ditherq + 1 * 8] 315 VBROADCASTSD blue_dither, [pointer_c_ditherq + 2 * 8] 316%else ; cpuflag(mmx/mmxext) 317%define blue_dither [pointer_c_ditherq + 2 * 8] 318%define green_dither [pointer_c_ditherq + 1 * 8] 319%define red_dither [pointer_c_ditherq + 0 * 8] 320%endif 321%if %3 == 15 322%define gmask pb_03 323%define isRGB15 1 324%else 325%define gmask pb_07 326%define isRGB15 0 327%endif 328 paddusb m0, blue_dither 329 paddusb m2, green_dither 330 paddusb m1, red_dither 331 pand m0, [pb_f8] 332 pand m1, [pb_f8] 333 mova m3, m2 334 psllw m2, 3 - isRGB15 335 psrlw m3, 5 + isRGB15 336 psrlw m0, 3 337 psrlw m1, isRGB15 338 pand m2, [pb_e0] 339 pand m3, [gmask] 340 por m0, m2 341 por m1, m3 342 mova m2, m0 343 punpcklbw m0, m1 344 punpckhbw m2, m1 345 movu [imageq], m0 346 movu [imageq + 8 * time_num], m2 347%endif ; PACK RGB15/16 348%endif ; PACK RGB15/16/32 349 350movu m_y, [py_2indexq + 2 * indexq + 8 * time_num] 351movh m_v, [pv_indexq + indexq + 4 * time_num] 352movh m_u, [pu_indexq + indexq + 4 * time_num] 353add imageq, 8 * depth * time_num 354add indexq, 4 * time_num 355js .loop0 356 357REP_RET 358 359%endmacro 360 361INIT_MMX mmx 362yuv2rgb_fn yuv, rgb, 24 363yuv2rgb_fn yuv, bgr, 24 364yuv2rgb_fn yuv, rgb, 32 365yuv2rgb_fn yuv, bgr, 32 366yuv2rgb_fn yuva, rgb, 32 367yuv2rgb_fn yuva, bgr, 32 368yuv2rgb_fn yuv, rgb, 15 369yuv2rgb_fn yuv, rgb, 16 370 371INIT_MMX mmxext 372yuv2rgb_fn yuv, rgb, 24 373yuv2rgb_fn yuv, bgr, 24 374 375INIT_XMM ssse3 376yuv2rgb_fn yuv, rgb, 24 377yuv2rgb_fn yuv, bgr, 24 378yuv2rgb_fn yuv, rgb, 32 379yuv2rgb_fn yuv, bgr, 32 380yuv2rgb_fn yuva, rgb, 32 381yuv2rgb_fn yuva, bgr, 32 382yuv2rgb_fn yuv, rgb, 15 383yuv2rgb_fn yuv, rgb, 16 384