• Main Page
  • Related Pages
  • Modules
  • Data Structures
  • Files
  • Examples
  • File List
  • Globals

libavcodec/x86/dsputil_mmx.c

Go to the documentation of this file.
00001 /*
00002  * MMX optimized DSP utils
00003  * Copyright (c) 2000, 2001 Fabrice Bellard
00004  * Copyright (c) 2002-2004 Michael Niedermayer <michaelni@gmx.at>
00005  *
00006  * This file is part of FFmpeg.
00007  *
00008  * FFmpeg is free software; you can redistribute it and/or
00009  * modify it under the terms of the GNU Lesser General Public
00010  * License as published by the Free Software Foundation; either
00011  * version 2.1 of the License, or (at your option) any later version.
00012  *
00013  * FFmpeg is distributed in the hope that it will be useful,
00014  * but WITHOUT ANY WARRANTY; without even the implied warranty of
00015  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
00016  * Lesser General Public License for more details.
00017  *
00018  * You should have received a copy of the GNU Lesser General Public
00019  * License along with FFmpeg; if not, write to the Free Software
00020  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
00021  *
00022  * MMX optimization by Nick Kurshev <nickols_k@mail.ru>
00023  */
00024 
00025 #include "libavutil/cpu.h"
00026 #include "libavutil/x86_cpu.h"
00027 #include "libavcodec/dsputil.h"
00028 #include "libavcodec/h264dsp.h"
00029 #include "libavcodec/mpegvideo.h"
00030 #include "libavcodec/simple_idct.h"
00031 #include "libavcodec/ac3dec.h"
00032 #include "dsputil_mmx.h"
00033 #include "idct_xvid.h"
00034 
00035 //#undef NDEBUG
00036 //#include <assert.h>
00037 
00038 /* pixel operations */
00039 DECLARE_ALIGNED(8,  const uint64_t, ff_bone) = 0x0101010101010101ULL;
00040 DECLARE_ALIGNED(8,  const uint64_t, ff_wtwo) = 0x0002000200020002ULL;
00041 
00042 DECLARE_ALIGNED(16, const uint64_t, ff_pdw_80000000)[2] =
00043 {0x8000000080000000ULL, 0x8000000080000000ULL};
00044 
00045 DECLARE_ALIGNED(8,  const uint64_t, ff_pw_1  ) = 0x0001000100010001ULL;
00046 DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_2  ) = {0x0002000200020002ULL, 0x0002000200020002ULL};
00047 DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_3  ) = {0x0003000300030003ULL, 0x0003000300030003ULL};
00048 DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_4  ) = {0x0004000400040004ULL, 0x0004000400040004ULL};
00049 DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_5  ) = {0x0005000500050005ULL, 0x0005000500050005ULL};
00050 DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_8  ) = {0x0008000800080008ULL, 0x0008000800080008ULL};
00051 DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_9  ) = {0x0009000900090009ULL, 0x0009000900090009ULL};
00052 DECLARE_ALIGNED(8,  const uint64_t, ff_pw_15 ) = 0x000F000F000F000FULL;
00053 DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_16 ) = {0x0010001000100010ULL, 0x0010001000100010ULL};
00054 DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_17 ) = {0x0011001100110011ULL, 0x0011001100110011ULL};
00055 DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_18 ) = {0x0012001200120012ULL, 0x0012001200120012ULL};
00056 DECLARE_ALIGNED(8,  const uint64_t, ff_pw_20 ) = 0x0014001400140014ULL;
00057 DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_27 ) = {0x001B001B001B001BULL, 0x001B001B001B001BULL};
00058 DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_28 ) = {0x001C001C001C001CULL, 0x001C001C001C001CULL};
00059 DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_32 ) = {0x0020002000200020ULL, 0x0020002000200020ULL};
00060 DECLARE_ALIGNED(8,  const uint64_t, ff_pw_42 ) = 0x002A002A002A002AULL;
00061 DECLARE_ALIGNED(8,  const uint64_t, ff_pw_53 ) = 0x0035003500350035ULL;
00062 DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_63 ) = {0x003F003F003F003FULL, 0x003F003F003F003FULL};
00063 DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_64 ) = {0x0040004000400040ULL, 0x0040004000400040ULL};
00064 DECLARE_ALIGNED(8,  const uint64_t, ff_pw_96 ) = 0x0060006000600060ULL;
00065 DECLARE_ALIGNED(8,  const uint64_t, ff_pw_128) = 0x0080008000800080ULL;
00066 DECLARE_ALIGNED(8,  const uint64_t, ff_pw_255) = 0x00ff00ff00ff00ffULL;
00067 
00068 DECLARE_ALIGNED(16, const xmm_reg,  ff_pb_0  ) = {0x0000000000000000ULL, 0x0000000000000000ULL};
00069 DECLARE_ALIGNED(16, const xmm_reg,  ff_pb_1  ) = {0x0101010101010101ULL, 0x0101010101010101ULL};
00070 DECLARE_ALIGNED(16, const xmm_reg,  ff_pb_3  ) = {0x0303030303030303ULL, 0x0303030303030303ULL};
00071 DECLARE_ALIGNED(16, const xmm_reg,  ff_pb_4  ) = {0x0404040404040404ULL, 0x0404040404040404ULL};
00072 DECLARE_ALIGNED(8,  const uint64_t, ff_pb_7  ) = 0x0707070707070707ULL;
00073 DECLARE_ALIGNED(8,  const uint64_t, ff_pb_1F ) = 0x1F1F1F1F1F1F1F1FULL;
00074 DECLARE_ALIGNED(8,  const uint64_t, ff_pb_3F ) = 0x3F3F3F3F3F3F3F3FULL;
00075 DECLARE_ALIGNED(16, const xmm_reg,  ff_pb_80 ) = {0x8080808080808080ULL, 0x8080808080808080ULL};
00076 DECLARE_ALIGNED(8,  const uint64_t, ff_pb_81 ) = 0x8181818181818181ULL;
00077 DECLARE_ALIGNED(16, const xmm_reg,  ff_pb_A1 ) = {0xA1A1A1A1A1A1A1A1ULL, 0xA1A1A1A1A1A1A1A1ULL};
00078 DECLARE_ALIGNED(16, const xmm_reg,  ff_pb_F8 ) = {0xF8F8F8F8F8F8F8F8ULL, 0xF8F8F8F8F8F8F8F8ULL};
00079 DECLARE_ALIGNED(8,  const uint64_t, ff_pb_FC ) = 0xFCFCFCFCFCFCFCFCULL;
00080 DECLARE_ALIGNED(16, const xmm_reg,  ff_pb_FE ) = {0xFEFEFEFEFEFEFEFEULL, 0xFEFEFEFEFEFEFEFEULL};
00081 
00082 DECLARE_ALIGNED(16, const double, ff_pd_1)[2] = { 1.0, 1.0 };
00083 DECLARE_ALIGNED(16, const double, ff_pd_2)[2] = { 2.0, 2.0 };
00084 
00085 #define JUMPALIGN() __asm__ volatile (".p2align 3"::)
00086 #define MOVQ_ZERO(regd)  __asm__ volatile ("pxor %%" #regd ", %%" #regd ::)
00087 
00088 #define MOVQ_BFE(regd) \
00089     __asm__ volatile ( \
00090     "pcmpeqd %%" #regd ", %%" #regd " \n\t"\
00091     "paddb %%" #regd ", %%" #regd " \n\t" ::)
00092 
00093 #ifndef PIC
00094 #define MOVQ_BONE(regd)  __asm__ volatile ("movq %0, %%" #regd " \n\t" ::"m"(ff_bone))
00095 #define MOVQ_WTWO(regd)  __asm__ volatile ("movq %0, %%" #regd " \n\t" ::"m"(ff_wtwo))
00096 #else
00097 // for shared library it's better to use this way for accessing constants
00098 // pcmpeqd -> -1
00099 #define MOVQ_BONE(regd) \
00100     __asm__ volatile ( \
00101     "pcmpeqd %%" #regd ", %%" #regd " \n\t" \
00102     "psrlw $15, %%" #regd " \n\t" \
00103     "packuswb %%" #regd ", %%" #regd " \n\t" ::)
00104 
00105 #define MOVQ_WTWO(regd) \
00106     __asm__ volatile ( \
00107     "pcmpeqd %%" #regd ", %%" #regd " \n\t" \
00108     "psrlw $15, %%" #regd " \n\t" \
00109     "psllw $1, %%" #regd " \n\t"::)
00110 
00111 #endif
00112 
00113 // using regr as temporary and for the output result
00114 // first argument is unmodifed and second is trashed
00115 // regfe is supposed to contain 0xfefefefefefefefe
00116 #define PAVGB_MMX_NO_RND(rega, regb, regr, regfe) \
00117     "movq " #rega ", " #regr "  \n\t"\
00118     "pand " #regb ", " #regr "  \n\t"\
00119     "pxor " #rega ", " #regb "  \n\t"\
00120     "pand " #regfe "," #regb "  \n\t"\
00121     "psrlq $1, " #regb "        \n\t"\
00122     "paddb " #regb ", " #regr " \n\t"
00123 
00124 #define PAVGB_MMX(rega, regb, regr, regfe) \
00125     "movq " #rega ", " #regr "  \n\t"\
00126     "por  " #regb ", " #regr "  \n\t"\
00127     "pxor " #rega ", " #regb "  \n\t"\
00128     "pand " #regfe "," #regb "  \n\t"\
00129     "psrlq $1, " #regb "        \n\t"\
00130     "psubb " #regb ", " #regr " \n\t"
00131 
00132 // mm6 is supposed to contain 0xfefefefefefefefe
00133 #define PAVGBP_MMX_NO_RND(rega, regb, regr,  regc, regd, regp) \
00134     "movq " #rega ", " #regr "  \n\t"\
00135     "movq " #regc ", " #regp "  \n\t"\
00136     "pand " #regb ", " #regr "  \n\t"\
00137     "pand " #regd ", " #regp "  \n\t"\
00138     "pxor " #rega ", " #regb "  \n\t"\
00139     "pxor " #regc ", " #regd "  \n\t"\
00140     "pand %%mm6, " #regb "      \n\t"\
00141     "pand %%mm6, " #regd "      \n\t"\
00142     "psrlq $1, " #regb "        \n\t"\
00143     "psrlq $1, " #regd "        \n\t"\
00144     "paddb " #regb ", " #regr " \n\t"\
00145     "paddb " #regd ", " #regp " \n\t"
00146 
00147 #define PAVGBP_MMX(rega, regb, regr, regc, regd, regp) \
00148     "movq " #rega ", " #regr "  \n\t"\
00149     "movq " #regc ", " #regp "  \n\t"\
00150     "por  " #regb ", " #regr "  \n\t"\
00151     "por  " #regd ", " #regp "  \n\t"\
00152     "pxor " #rega ", " #regb "  \n\t"\
00153     "pxor " #regc ", " #regd "  \n\t"\
00154     "pand %%mm6, " #regb "      \n\t"\
00155     "pand %%mm6, " #regd "      \n\t"\
00156     "psrlq $1, " #regd "        \n\t"\
00157     "psrlq $1, " #regb "        \n\t"\
00158     "psubb " #regb ", " #regr " \n\t"\
00159     "psubb " #regd ", " #regp " \n\t"
00160 
00161 /***********************************/
00162 /* MMX no rounding */
00163 #define DEF(x, y) x ## _no_rnd_ ## y ##_mmx
00164 #define SET_RND  MOVQ_WONE
00165 #define PAVGBP(a, b, c, d, e, f)        PAVGBP_MMX_NO_RND(a, b, c, d, e, f)
00166 #define PAVGB(a, b, c, e)               PAVGB_MMX_NO_RND(a, b, c, e)
00167 #define OP_AVG(a, b, c, e)              PAVGB_MMX(a, b, c, e)
00168 
00169 #include "dsputil_mmx_rnd_template.c"
00170 
00171 #undef DEF
00172 #undef SET_RND
00173 #undef PAVGBP
00174 #undef PAVGB
00175 /***********************************/
00176 /* MMX rounding */
00177 
00178 #define DEF(x, y) x ## _ ## y ##_mmx
00179 #define SET_RND  MOVQ_WTWO
00180 #define PAVGBP(a, b, c, d, e, f)        PAVGBP_MMX(a, b, c, d, e, f)
00181 #define PAVGB(a, b, c, e)               PAVGB_MMX(a, b, c, e)
00182 
00183 #include "dsputil_mmx_rnd_template.c"
00184 
00185 #undef DEF
00186 #undef SET_RND
00187 #undef PAVGBP
00188 #undef PAVGB
00189 #undef OP_AVG
00190 
00191 /***********************************/
00192 /* 3Dnow specific */
00193 
00194 #define DEF(x) x ## _3dnow
00195 #define PAVGB "pavgusb"
00196 #define OP_AVG PAVGB
00197 
00198 #include "dsputil_mmx_avg_template.c"
00199 
00200 #undef DEF
00201 #undef PAVGB
00202 #undef OP_AVG
00203 
00204 /***********************************/
00205 /* MMX2 specific */
00206 
00207 #define DEF(x) x ## _mmx2
00208 
00209 /* Introduced only in MMX2 set */
00210 #define PAVGB "pavgb"
00211 #define OP_AVG PAVGB
00212 
00213 #include "dsputil_mmx_avg_template.c"
00214 
00215 #undef DEF
00216 #undef PAVGB
00217 #undef OP_AVG
00218 
00219 #define put_no_rnd_pixels16_mmx put_pixels16_mmx
00220 #define put_no_rnd_pixels8_mmx put_pixels8_mmx
00221 #define put_pixels16_mmx2 put_pixels16_mmx
00222 #define put_pixels8_mmx2 put_pixels8_mmx
00223 #define put_pixels4_mmx2 put_pixels4_mmx
00224 #define put_no_rnd_pixels16_mmx2 put_no_rnd_pixels16_mmx
00225 #define put_no_rnd_pixels8_mmx2 put_no_rnd_pixels8_mmx
00226 #define put_pixels16_3dnow put_pixels16_mmx
00227 #define put_pixels8_3dnow put_pixels8_mmx
00228 #define put_pixels4_3dnow put_pixels4_mmx
00229 #define put_no_rnd_pixels16_3dnow put_no_rnd_pixels16_mmx
00230 #define put_no_rnd_pixels8_3dnow put_no_rnd_pixels8_mmx
00231 
00232 /***********************************/
00233 /* standard MMX */
00234 
00235 void ff_put_pixels_clamped_mmx(const DCTELEM *block, uint8_t *pixels, int line_size)
00236 {
00237     const DCTELEM *p;
00238     uint8_t *pix;
00239 
00240     /* read the pixels */
00241     p = block;
00242     pix = pixels;
00243     /* unrolled loop */
00244         __asm__ volatile(
00245                 "movq   %3, %%mm0               \n\t"
00246                 "movq   8%3, %%mm1              \n\t"
00247                 "movq   16%3, %%mm2             \n\t"
00248                 "movq   24%3, %%mm3             \n\t"
00249                 "movq   32%3, %%mm4             \n\t"
00250                 "movq   40%3, %%mm5             \n\t"
00251                 "movq   48%3, %%mm6             \n\t"
00252                 "movq   56%3, %%mm7             \n\t"
00253                 "packuswb %%mm1, %%mm0          \n\t"
00254                 "packuswb %%mm3, %%mm2          \n\t"
00255                 "packuswb %%mm5, %%mm4          \n\t"
00256                 "packuswb %%mm7, %%mm6          \n\t"
00257                 "movq   %%mm0, (%0)             \n\t"
00258                 "movq   %%mm2, (%0, %1)         \n\t"
00259                 "movq   %%mm4, (%0, %1, 2)      \n\t"
00260                 "movq   %%mm6, (%0, %2)         \n\t"
00261                 ::"r" (pix), "r" ((x86_reg)line_size), "r" ((x86_reg)line_size*3), "m"(*p)
00262                 :"memory");
00263         pix += line_size*4;
00264         p += 32;
00265 
00266     // if here would be an exact copy of the code above
00267     // compiler would generate some very strange code
00268     // thus using "r"
00269     __asm__ volatile(
00270             "movq       (%3), %%mm0             \n\t"
00271             "movq       8(%3), %%mm1            \n\t"
00272             "movq       16(%3), %%mm2           \n\t"
00273             "movq       24(%3), %%mm3           \n\t"
00274             "movq       32(%3), %%mm4           \n\t"
00275             "movq       40(%3), %%mm5           \n\t"
00276             "movq       48(%3), %%mm6           \n\t"
00277             "movq       56(%3), %%mm7           \n\t"
00278             "packuswb %%mm1, %%mm0              \n\t"
00279             "packuswb %%mm3, %%mm2              \n\t"
00280             "packuswb %%mm5, %%mm4              \n\t"
00281             "packuswb %%mm7, %%mm6              \n\t"
00282             "movq       %%mm0, (%0)             \n\t"
00283             "movq       %%mm2, (%0, %1)         \n\t"
00284             "movq       %%mm4, (%0, %1, 2)      \n\t"
00285             "movq       %%mm6, (%0, %2)         \n\t"
00286             ::"r" (pix), "r" ((x86_reg)line_size), "r" ((x86_reg)line_size*3), "r"(p)
00287             :"memory");
00288 }
00289 
00290 #define put_signed_pixels_clamped_mmx_half(off) \
00291             "movq    "#off"(%2), %%mm1          \n\t"\
00292             "movq 16+"#off"(%2), %%mm2          \n\t"\
00293             "movq 32+"#off"(%2), %%mm3          \n\t"\
00294             "movq 48+"#off"(%2), %%mm4          \n\t"\
00295             "packsswb  8+"#off"(%2), %%mm1      \n\t"\
00296             "packsswb 24+"#off"(%2), %%mm2      \n\t"\
00297             "packsswb 40+"#off"(%2), %%mm3      \n\t"\
00298             "packsswb 56+"#off"(%2), %%mm4      \n\t"\
00299             "paddb %%mm0, %%mm1                 \n\t"\
00300             "paddb %%mm0, %%mm2                 \n\t"\
00301             "paddb %%mm0, %%mm3                 \n\t"\
00302             "paddb %%mm0, %%mm4                 \n\t"\
00303             "movq %%mm1, (%0)                   \n\t"\
00304             "movq %%mm2, (%0, %3)               \n\t"\
00305             "movq %%mm3, (%0, %3, 2)            \n\t"\
00306             "movq %%mm4, (%0, %1)               \n\t"
00307 
00308 void ff_put_signed_pixels_clamped_mmx(const DCTELEM *block, uint8_t *pixels, int line_size)
00309 {
00310     x86_reg line_skip = line_size;
00311     x86_reg line_skip3;
00312 
00313     __asm__ volatile (
00314             "movq "MANGLE(ff_pb_80)", %%mm0 \n\t"
00315             "lea (%3, %3, 2), %1                \n\t"
00316             put_signed_pixels_clamped_mmx_half(0)
00317             "lea (%0, %3, 4), %0                \n\t"
00318             put_signed_pixels_clamped_mmx_half(64)
00319             :"+&r" (pixels), "=&r" (line_skip3)
00320             :"r" (block), "r"(line_skip)
00321             :"memory");
00322 }
00323 
00324 void ff_add_pixels_clamped_mmx(const DCTELEM *block, uint8_t *pixels, int line_size)
00325 {
00326     const DCTELEM *p;
00327     uint8_t *pix;
00328     int i;
00329 
00330     /* read the pixels */
00331     p = block;
00332     pix = pixels;
00333     MOVQ_ZERO(mm7);
00334     i = 4;
00335     do {
00336         __asm__ volatile(
00337                 "movq   (%2), %%mm0     \n\t"
00338                 "movq   8(%2), %%mm1    \n\t"
00339                 "movq   16(%2), %%mm2   \n\t"
00340                 "movq   24(%2), %%mm3   \n\t"
00341                 "movq   %0, %%mm4       \n\t"
00342                 "movq   %1, %%mm6       \n\t"
00343                 "movq   %%mm4, %%mm5    \n\t"
00344                 "punpcklbw %%mm7, %%mm4 \n\t"
00345                 "punpckhbw %%mm7, %%mm5 \n\t"
00346                 "paddsw %%mm4, %%mm0    \n\t"
00347                 "paddsw %%mm5, %%mm1    \n\t"
00348                 "movq   %%mm6, %%mm5    \n\t"
00349                 "punpcklbw %%mm7, %%mm6 \n\t"
00350                 "punpckhbw %%mm7, %%mm5 \n\t"
00351                 "paddsw %%mm6, %%mm2    \n\t"
00352                 "paddsw %%mm5, %%mm3    \n\t"
00353                 "packuswb %%mm1, %%mm0  \n\t"
00354                 "packuswb %%mm3, %%mm2  \n\t"
00355                 "movq   %%mm0, %0       \n\t"
00356                 "movq   %%mm2, %1       \n\t"
00357                 :"+m"(*pix), "+m"(*(pix+line_size))
00358                 :"r"(p)
00359                 :"memory");
00360         pix += line_size*2;
00361         p += 16;
00362     } while (--i);
00363 }
00364 
00365 static void put_pixels4_mmx(uint8_t *block, const uint8_t *pixels, int line_size, int h)
00366 {
00367     __asm__ volatile(
00368          "lea (%3, %3), %%"REG_a"       \n\t"
00369          ".p2align 3                    \n\t"
00370          "1:                            \n\t"
00371          "movd (%1), %%mm0              \n\t"
00372          "movd (%1, %3), %%mm1          \n\t"
00373          "movd %%mm0, (%2)              \n\t"
00374          "movd %%mm1, (%2, %3)          \n\t"
00375          "add %%"REG_a", %1             \n\t"
00376          "add %%"REG_a", %2             \n\t"
00377          "movd (%1), %%mm0              \n\t"
00378          "movd (%1, %3), %%mm1          \n\t"
00379          "movd %%mm0, (%2)              \n\t"
00380          "movd %%mm1, (%2, %3)          \n\t"
00381          "add %%"REG_a", %1             \n\t"
00382          "add %%"REG_a", %2             \n\t"
00383          "subl $4, %0                   \n\t"
00384          "jnz 1b                        \n\t"
00385          : "+g"(h), "+r" (pixels),  "+r" (block)
00386          : "r"((x86_reg)line_size)
00387          : "%"REG_a, "memory"
00388         );
00389 }
00390 
00391 static void put_pixels8_mmx(uint8_t *block, const uint8_t *pixels, int line_size, int h)
00392 {
00393     __asm__ volatile(
00394          "lea (%3, %3), %%"REG_a"       \n\t"
00395          ".p2align 3                    \n\t"
00396          "1:                            \n\t"
00397          "movq (%1), %%mm0              \n\t"
00398          "movq (%1, %3), %%mm1          \n\t"
00399          "movq %%mm0, (%2)              \n\t"
00400          "movq %%mm1, (%2, %3)          \n\t"
00401          "add %%"REG_a", %1             \n\t"
00402          "add %%"REG_a", %2             \n\t"
00403          "movq (%1), %%mm0              \n\t"
00404          "movq (%1, %3), %%mm1          \n\t"
00405          "movq %%mm0, (%2)              \n\t"
00406          "movq %%mm1, (%2, %3)          \n\t"
00407          "add %%"REG_a", %1             \n\t"
00408          "add %%"REG_a", %2             \n\t"
00409          "subl $4, %0                   \n\t"
00410          "jnz 1b                        \n\t"
00411          : "+g"(h), "+r" (pixels),  "+r" (block)
00412          : "r"((x86_reg)line_size)
00413          : "%"REG_a, "memory"
00414         );
00415 }
00416 
00417 static void put_pixels16_mmx(uint8_t *block, const uint8_t *pixels, int line_size, int h)
00418 {
00419     __asm__ volatile(
00420          "lea (%3, %3), %%"REG_a"       \n\t"
00421          ".p2align 3                    \n\t"
00422          "1:                            \n\t"
00423          "movq (%1), %%mm0              \n\t"
00424          "movq 8(%1), %%mm4             \n\t"
00425          "movq (%1, %3), %%mm1          \n\t"
00426          "movq 8(%1, %3), %%mm5         \n\t"
00427          "movq %%mm0, (%2)              \n\t"
00428          "movq %%mm4, 8(%2)             \n\t"
00429          "movq %%mm1, (%2, %3)          \n\t"
00430          "movq %%mm5, 8(%2, %3)         \n\t"
00431          "add %%"REG_a", %1             \n\t"
00432          "add %%"REG_a", %2             \n\t"
00433          "movq (%1), %%mm0              \n\t"
00434          "movq 8(%1), %%mm4             \n\t"
00435          "movq (%1, %3), %%mm1          \n\t"
00436          "movq 8(%1, %3), %%mm5         \n\t"
00437          "movq %%mm0, (%2)              \n\t"
00438          "movq %%mm4, 8(%2)             \n\t"
00439          "movq %%mm1, (%2, %3)          \n\t"
00440          "movq %%mm5, 8(%2, %3)         \n\t"
00441          "add %%"REG_a", %1             \n\t"
00442          "add %%"REG_a", %2             \n\t"
00443          "subl $4, %0                   \n\t"
00444          "jnz 1b                        \n\t"
00445          : "+g"(h), "+r" (pixels),  "+r" (block)
00446          : "r"((x86_reg)line_size)
00447          : "%"REG_a, "memory"
00448         );
00449 }
00450 
00451 static void put_pixels16_sse2(uint8_t *block, const uint8_t *pixels, int line_size, int h)
00452 {
00453     __asm__ volatile(
00454          "1:                            \n\t"
00455          "movdqu (%1), %%xmm0           \n\t"
00456          "movdqu (%1,%3), %%xmm1        \n\t"
00457          "movdqu (%1,%3,2), %%xmm2      \n\t"
00458          "movdqu (%1,%4), %%xmm3        \n\t"
00459          "movdqa %%xmm0, (%2)           \n\t"
00460          "movdqa %%xmm1, (%2,%3)        \n\t"
00461          "movdqa %%xmm2, (%2,%3,2)      \n\t"
00462          "movdqa %%xmm3, (%2,%4)        \n\t"
00463          "subl $4, %0                   \n\t"
00464          "lea (%1,%3,4), %1             \n\t"
00465          "lea (%2,%3,4), %2             \n\t"
00466          "jnz 1b                        \n\t"
00467          : "+g"(h), "+r" (pixels),  "+r" (block)
00468          : "r"((x86_reg)line_size), "r"((x86_reg)3L*line_size)
00469          : "memory"
00470         );
00471 }
00472 
00473 static void avg_pixels16_sse2(uint8_t *block, const uint8_t *pixels, int line_size, int h)
00474 {
00475     __asm__ volatile(
00476          "1:                            \n\t"
00477          "movdqu (%1), %%xmm0           \n\t"
00478          "movdqu (%1,%3), %%xmm1        \n\t"
00479          "movdqu (%1,%3,2), %%xmm2      \n\t"
00480          "movdqu (%1,%4), %%xmm3        \n\t"
00481          "pavgb  (%2), %%xmm0           \n\t"
00482          "pavgb  (%2,%3), %%xmm1        \n\t"
00483          "pavgb  (%2,%3,2), %%xmm2      \n\t"
00484          "pavgb  (%2,%4), %%xmm3        \n\t"
00485          "movdqa %%xmm0, (%2)           \n\t"
00486          "movdqa %%xmm1, (%2,%3)        \n\t"
00487          "movdqa %%xmm2, (%2,%3,2)      \n\t"
00488          "movdqa %%xmm3, (%2,%4)        \n\t"
00489          "subl $4, %0                   \n\t"
00490          "lea (%1,%3,4), %1             \n\t"
00491          "lea (%2,%3,4), %2             \n\t"
00492          "jnz 1b                        \n\t"
00493          : "+g"(h), "+r" (pixels),  "+r" (block)
00494          : "r"((x86_reg)line_size), "r"((x86_reg)3L*line_size)
00495          : "memory"
00496         );
00497 }
00498 
00499 #define CLEAR_BLOCKS(name,n) \
00500 static void name(DCTELEM *blocks)\
00501 {\
00502     __asm__ volatile(\
00503                 "pxor %%mm7, %%mm7              \n\t"\
00504                 "mov     %1, %%"REG_a"          \n\t"\
00505                 "1:                             \n\t"\
00506                 "movq %%mm7, (%0, %%"REG_a")    \n\t"\
00507                 "movq %%mm7, 8(%0, %%"REG_a")   \n\t"\
00508                 "movq %%mm7, 16(%0, %%"REG_a")  \n\t"\
00509                 "movq %%mm7, 24(%0, %%"REG_a")  \n\t"\
00510                 "add $32, %%"REG_a"             \n\t"\
00511                 " js 1b                         \n\t"\
00512                 : : "r" (((uint8_t *)blocks)+128*n),\
00513                     "i" (-128*n)\
00514                 : "%"REG_a\
00515         );\
00516 }
00517 CLEAR_BLOCKS(clear_blocks_mmx, 6)
00518 CLEAR_BLOCKS(clear_block_mmx, 1)
00519 
00520 static void clear_block_sse(DCTELEM *block)
00521 {
00522     __asm__ volatile(
00523         "xorps  %%xmm0, %%xmm0  \n"
00524         "movaps %%xmm0,    (%0) \n"
00525         "movaps %%xmm0,  16(%0) \n"
00526         "movaps %%xmm0,  32(%0) \n"
00527         "movaps %%xmm0,  48(%0) \n"
00528         "movaps %%xmm0,  64(%0) \n"
00529         "movaps %%xmm0,  80(%0) \n"
00530         "movaps %%xmm0,  96(%0) \n"
00531         "movaps %%xmm0, 112(%0) \n"
00532         :: "r"(block)
00533         : "memory"
00534     );
00535 }
00536 
00537 static void clear_blocks_sse(DCTELEM *blocks)
00538 {\
00539     __asm__ volatile(
00540         "xorps  %%xmm0, %%xmm0  \n"
00541         "mov     %1, %%"REG_a"  \n"
00542         "1:                     \n"
00543         "movaps %%xmm0,    (%0, %%"REG_a") \n"
00544         "movaps %%xmm0,  16(%0, %%"REG_a") \n"
00545         "movaps %%xmm0,  32(%0, %%"REG_a") \n"
00546         "movaps %%xmm0,  48(%0, %%"REG_a") \n"
00547         "movaps %%xmm0,  64(%0, %%"REG_a") \n"
00548         "movaps %%xmm0,  80(%0, %%"REG_a") \n"
00549         "movaps %%xmm0,  96(%0, %%"REG_a") \n"
00550         "movaps %%xmm0, 112(%0, %%"REG_a") \n"
00551         "add $128, %%"REG_a"    \n"
00552         " js 1b                 \n"
00553         : : "r" (((uint8_t *)blocks)+128*6),
00554             "i" (-128*6)
00555         : "%"REG_a
00556     );
00557 }
00558 
00559 static void add_bytes_mmx(uint8_t *dst, uint8_t *src, int w){
00560     x86_reg i=0;
00561     __asm__ volatile(
00562         "jmp 2f                         \n\t"
00563         "1:                             \n\t"
00564         "movq  (%1, %0), %%mm0          \n\t"
00565         "movq  (%2, %0), %%mm1          \n\t"
00566         "paddb %%mm0, %%mm1             \n\t"
00567         "movq %%mm1, (%2, %0)           \n\t"
00568         "movq 8(%1, %0), %%mm0          \n\t"
00569         "movq 8(%2, %0), %%mm1          \n\t"
00570         "paddb %%mm0, %%mm1             \n\t"
00571         "movq %%mm1, 8(%2, %0)          \n\t"
00572         "add $16, %0                    \n\t"
00573         "2:                             \n\t"
00574         "cmp %3, %0                     \n\t"
00575         " js 1b                         \n\t"
00576         : "+r" (i)
00577         : "r"(src), "r"(dst), "r"((x86_reg)w-15)
00578     );
00579     for(; i<w; i++)
00580         dst[i+0] += src[i+0];
00581 }
00582 
00583 #if HAVE_7REGS && HAVE_TEN_OPERANDS
00584 static void add_hfyu_median_prediction_cmov(uint8_t *dst, const uint8_t *top, const uint8_t *diff, int w, int *left, int *left_top) {
00585     x86_reg w2 = -w;
00586     x86_reg x;
00587     int l = *left & 0xff;
00588     int tl = *left_top & 0xff;
00589     int t;
00590     __asm__ volatile(
00591         "mov    %7, %3 \n"
00592         "1: \n"
00593         "movzbl (%3,%4), %2 \n"
00594         "mov    %2, %k3 \n"
00595         "sub   %b1, %b3 \n"
00596         "add   %b0, %b3 \n"
00597         "mov    %2, %1 \n"
00598         "cmp    %0, %2 \n"
00599         "cmovg  %0, %2 \n"
00600         "cmovg  %1, %0 \n"
00601         "cmp   %k3, %0 \n"
00602         "cmovg %k3, %0 \n"
00603         "mov    %7, %3 \n"
00604         "cmp    %2, %0 \n"
00605         "cmovl  %2, %0 \n"
00606         "add (%6,%4), %b0 \n"
00607         "mov   %b0, (%5,%4) \n"
00608         "inc    %4 \n"
00609         "jl 1b \n"
00610         :"+&q"(l), "+&q"(tl), "=&r"(t), "=&q"(x), "+&r"(w2)
00611         :"r"(dst+w), "r"(diff+w), "rm"(top+w)
00612     );
00613     *left = l;
00614     *left_top = tl;
00615 }
00616 #endif
00617 
00618 #define H263_LOOP_FILTER \
00619         "pxor %%mm7, %%mm7              \n\t"\
00620         "movq  %0, %%mm0                \n\t"\
00621         "movq  %0, %%mm1                \n\t"\
00622         "movq  %3, %%mm2                \n\t"\
00623         "movq  %3, %%mm3                \n\t"\
00624         "punpcklbw %%mm7, %%mm0         \n\t"\
00625         "punpckhbw %%mm7, %%mm1         \n\t"\
00626         "punpcklbw %%mm7, %%mm2         \n\t"\
00627         "punpckhbw %%mm7, %%mm3         \n\t"\
00628         "psubw %%mm2, %%mm0             \n\t"\
00629         "psubw %%mm3, %%mm1             \n\t"\
00630         "movq  %1, %%mm2                \n\t"\
00631         "movq  %1, %%mm3                \n\t"\
00632         "movq  %2, %%mm4                \n\t"\
00633         "movq  %2, %%mm5                \n\t"\
00634         "punpcklbw %%mm7, %%mm2         \n\t"\
00635         "punpckhbw %%mm7, %%mm3         \n\t"\
00636         "punpcklbw %%mm7, %%mm4         \n\t"\
00637         "punpckhbw %%mm7, %%mm5         \n\t"\
00638         "psubw %%mm2, %%mm4             \n\t"\
00639         "psubw %%mm3, %%mm5             \n\t"\
00640         "psllw $2, %%mm4                \n\t"\
00641         "psllw $2, %%mm5                \n\t"\
00642         "paddw %%mm0, %%mm4             \n\t"\
00643         "paddw %%mm1, %%mm5             \n\t"\
00644         "pxor %%mm6, %%mm6              \n\t"\
00645         "pcmpgtw %%mm4, %%mm6           \n\t"\
00646         "pcmpgtw %%mm5, %%mm7           \n\t"\
00647         "pxor %%mm6, %%mm4              \n\t"\
00648         "pxor %%mm7, %%mm5              \n\t"\
00649         "psubw %%mm6, %%mm4             \n\t"\
00650         "psubw %%mm7, %%mm5             \n\t"\
00651         "psrlw $3, %%mm4                \n\t"\
00652         "psrlw $3, %%mm5                \n\t"\
00653         "packuswb %%mm5, %%mm4          \n\t"\
00654         "packsswb %%mm7, %%mm6          \n\t"\
00655         "pxor %%mm7, %%mm7              \n\t"\
00656         "movd %4, %%mm2                 \n\t"\
00657         "punpcklbw %%mm2, %%mm2         \n\t"\
00658         "punpcklbw %%mm2, %%mm2         \n\t"\
00659         "punpcklbw %%mm2, %%mm2         \n\t"\
00660         "psubusb %%mm4, %%mm2           \n\t"\
00661         "movq %%mm2, %%mm3              \n\t"\
00662         "psubusb %%mm4, %%mm3           \n\t"\
00663         "psubb %%mm3, %%mm2             \n\t"\
00664         "movq %1, %%mm3                 \n\t"\
00665         "movq %2, %%mm4                 \n\t"\
00666         "pxor %%mm6, %%mm3              \n\t"\
00667         "pxor %%mm6, %%mm4              \n\t"\
00668         "paddusb %%mm2, %%mm3           \n\t"\
00669         "psubusb %%mm2, %%mm4           \n\t"\
00670         "pxor %%mm6, %%mm3              \n\t"\
00671         "pxor %%mm6, %%mm4              \n\t"\
00672         "paddusb %%mm2, %%mm2           \n\t"\
00673         "packsswb %%mm1, %%mm0          \n\t"\
00674         "pcmpgtb %%mm0, %%mm7           \n\t"\
00675         "pxor %%mm7, %%mm0              \n\t"\
00676         "psubb %%mm7, %%mm0             \n\t"\
00677         "movq %%mm0, %%mm1              \n\t"\
00678         "psubusb %%mm2, %%mm0           \n\t"\
00679         "psubb %%mm0, %%mm1             \n\t"\
00680         "pand %5, %%mm1                 \n\t"\
00681         "psrlw $2, %%mm1                \n\t"\
00682         "pxor %%mm7, %%mm1              \n\t"\
00683         "psubb %%mm7, %%mm1             \n\t"\
00684         "movq %0, %%mm5                 \n\t"\
00685         "movq %3, %%mm6                 \n\t"\
00686         "psubb %%mm1, %%mm5             \n\t"\
00687         "paddb %%mm1, %%mm6             \n\t"
00688 
00689 static void h263_v_loop_filter_mmx(uint8_t *src, int stride, int qscale){
00690     if(CONFIG_H263_DECODER || CONFIG_H263_ENCODER) {
00691     const int strength= ff_h263_loop_filter_strength[qscale];
00692 
00693     __asm__ volatile(
00694 
00695         H263_LOOP_FILTER
00696 
00697         "movq %%mm3, %1                 \n\t"
00698         "movq %%mm4, %2                 \n\t"
00699         "movq %%mm5, %0                 \n\t"
00700         "movq %%mm6, %3                 \n\t"
00701         : "+m" (*(uint64_t*)(src - 2*stride)),
00702           "+m" (*(uint64_t*)(src - 1*stride)),
00703           "+m" (*(uint64_t*)(src + 0*stride)),
00704           "+m" (*(uint64_t*)(src + 1*stride))
00705         : "g" (2*strength), "m"(ff_pb_FC)
00706     );
00707     }
00708 }
00709 
00710 static void h263_h_loop_filter_mmx(uint8_t *src, int stride, int qscale){
00711     if(CONFIG_H263_DECODER || CONFIG_H263_ENCODER) {
00712     const int strength= ff_h263_loop_filter_strength[qscale];
00713     DECLARE_ALIGNED(8, uint64_t, temp)[4];
00714     uint8_t *btemp= (uint8_t*)temp;
00715 
00716     src -= 2;
00717 
00718     transpose4x4(btemp  , src           , 8, stride);
00719     transpose4x4(btemp+4, src + 4*stride, 8, stride);
00720     __asm__ volatile(
00721         H263_LOOP_FILTER // 5 3 4 6
00722 
00723         : "+m" (temp[0]),
00724           "+m" (temp[1]),
00725           "+m" (temp[2]),
00726           "+m" (temp[3])
00727         : "g" (2*strength), "m"(ff_pb_FC)
00728     );
00729 
00730     __asm__ volatile(
00731         "movq %%mm5, %%mm1              \n\t"
00732         "movq %%mm4, %%mm0              \n\t"
00733         "punpcklbw %%mm3, %%mm5         \n\t"
00734         "punpcklbw %%mm6, %%mm4         \n\t"
00735         "punpckhbw %%mm3, %%mm1         \n\t"
00736         "punpckhbw %%mm6, %%mm0         \n\t"
00737         "movq %%mm5, %%mm3              \n\t"
00738         "movq %%mm1, %%mm6              \n\t"
00739         "punpcklwd %%mm4, %%mm5         \n\t"
00740         "punpcklwd %%mm0, %%mm1         \n\t"
00741         "punpckhwd %%mm4, %%mm3         \n\t"
00742         "punpckhwd %%mm0, %%mm6         \n\t"
00743         "movd %%mm5, (%0)               \n\t"
00744         "punpckhdq %%mm5, %%mm5         \n\t"
00745         "movd %%mm5, (%0,%2)            \n\t"
00746         "movd %%mm3, (%0,%2,2)          \n\t"
00747         "punpckhdq %%mm3, %%mm3         \n\t"
00748         "movd %%mm3, (%0,%3)            \n\t"
00749         "movd %%mm1, (%1)               \n\t"
00750         "punpckhdq %%mm1, %%mm1         \n\t"
00751         "movd %%mm1, (%1,%2)            \n\t"
00752         "movd %%mm6, (%1,%2,2)          \n\t"
00753         "punpckhdq %%mm6, %%mm6         \n\t"
00754         "movd %%mm6, (%1,%3)            \n\t"
00755         :: "r" (src),
00756            "r" (src + 4*stride),
00757            "r" ((x86_reg)   stride ),
00758            "r" ((x86_reg)(3*stride))
00759     );
00760     }
00761 }
00762 
00763 /* draw the edges of width 'w' of an image of size width, height
00764    this mmx version can only handle w==8 || w==16 */
00765 static void draw_edges_mmx(uint8_t *buf, int wrap, int width, int height, int w, int h, int sides)
00766 {
00767     uint8_t *ptr, *last_line;
00768     int i;
00769 
00770     last_line = buf + (height - 1) * wrap;
00771     /* left and right */
00772     ptr = buf;
00773     if(w==8)
00774     {
00775         __asm__ volatile(
00776                 "1:                             \n\t"
00777                 "movd (%0), %%mm0               \n\t"
00778                 "punpcklbw %%mm0, %%mm0         \n\t"
00779                 "punpcklwd %%mm0, %%mm0         \n\t"
00780                 "punpckldq %%mm0, %%mm0         \n\t"
00781                 "movq %%mm0, -8(%0)             \n\t"
00782                 "movq -8(%0, %2), %%mm1         \n\t"
00783                 "punpckhbw %%mm1, %%mm1         \n\t"
00784                 "punpckhwd %%mm1, %%mm1         \n\t"
00785                 "punpckhdq %%mm1, %%mm1         \n\t"
00786                 "movq %%mm1, (%0, %2)           \n\t"
00787                 "add %1, %0                     \n\t"
00788                 "cmp %3, %0                     \n\t"
00789                 " jb 1b                         \n\t"
00790                 : "+r" (ptr)
00791                 : "r" ((x86_reg)wrap), "r" ((x86_reg)width), "r" (ptr + wrap*height)
00792         );
00793     }
00794     else
00795     {
00796         __asm__ volatile(
00797                 "1:                             \n\t"
00798                 "movd (%0), %%mm0               \n\t"
00799                 "punpcklbw %%mm0, %%mm0         \n\t"
00800                 "punpcklwd %%mm0, %%mm0         \n\t"
00801                 "punpckldq %%mm0, %%mm0         \n\t"
00802                 "movq %%mm0, -8(%0)             \n\t"
00803                 "movq %%mm0, -16(%0)            \n\t"
00804                 "movq -8(%0, %2), %%mm1         \n\t"
00805                 "punpckhbw %%mm1, %%mm1         \n\t"
00806                 "punpckhwd %%mm1, %%mm1         \n\t"
00807                 "punpckhdq %%mm1, %%mm1         \n\t"
00808                 "movq %%mm1, (%0, %2)           \n\t"
00809                 "movq %%mm1, 8(%0, %2)          \n\t"
00810                 "add %1, %0                     \n\t"
00811                 "cmp %3, %0                     \n\t"
00812                 " jb 1b                         \n\t"
00813                 : "+r" (ptr)
00814                 : "r" ((x86_reg)wrap), "r" ((x86_reg)width), "r" (ptr + wrap*height)
00815         );
00816     }
00817 
00818     /* top and bottom (and hopefully also the corners) */
00819     if (sides&EDGE_TOP) {
00820         for(i = 0; i < h; i += 4) {
00821             ptr= buf - (i + 1) * wrap - w;
00822             __asm__ volatile(
00823                     "1:                             \n\t"
00824                     "movq (%1, %0), %%mm0           \n\t"
00825                     "movq %%mm0, (%0)               \n\t"
00826                     "movq %%mm0, (%0, %2)           \n\t"
00827                     "movq %%mm0, (%0, %2, 2)        \n\t"
00828                     "movq %%mm0, (%0, %3)           \n\t"
00829                     "add $8, %0                     \n\t"
00830                     "cmp %4, %0                     \n\t"
00831                     " jb 1b                         \n\t"
00832                     : "+r" (ptr)
00833                     : "r" ((x86_reg)buf - (x86_reg)ptr - w), "r" ((x86_reg)-wrap), "r" ((x86_reg)-wrap*3), "r" (ptr+width+2*w)
00834             );
00835         }
00836     }
00837 
00838     if (sides&EDGE_BOTTOM) {
00839         for(i = 0; i < w; i += 4) {
00840             ptr= last_line + (i + 1) * wrap - w;
00841             __asm__ volatile(
00842                     "1:                             \n\t"
00843                     "movq (%1, %0), %%mm0           \n\t"
00844                     "movq %%mm0, (%0)               \n\t"
00845                     "movq %%mm0, (%0, %2)           \n\t"
00846                     "movq %%mm0, (%0, %2, 2)        \n\t"
00847                     "movq %%mm0, (%0, %3)           \n\t"
00848                     "add $8, %0                     \n\t"
00849                     "cmp %4, %0                     \n\t"
00850                     " jb 1b                         \n\t"
00851                     : "+r" (ptr)
00852                     : "r" ((x86_reg)last_line - (x86_reg)ptr - w), "r" ((x86_reg)wrap), "r" ((x86_reg)wrap*3), "r" (ptr+width+2*w)
00853             );
00854         }
00855     }
00856 }
00857 
00858 #define QPEL_V_LOW(m3,m4,m5,m6, pw_20, pw_3, rnd, in0, in1, in2, in7, out, OP)\
00859         "paddw " #m4 ", " #m3 "           \n\t" /* x1 */\
00860         "movq "MANGLE(ff_pw_20)", %%mm4   \n\t" /* 20 */\
00861         "pmullw " #m3 ", %%mm4            \n\t" /* 20x1 */\
00862         "movq "#in7", " #m3 "             \n\t" /* d */\
00863         "movq "#in0", %%mm5               \n\t" /* D */\
00864         "paddw " #m3 ", %%mm5             \n\t" /* x4 */\
00865         "psubw %%mm5, %%mm4               \n\t" /* 20x1 - x4 */\
00866         "movq "#in1", %%mm5               \n\t" /* C */\
00867         "movq "#in2", %%mm6               \n\t" /* B */\
00868         "paddw " #m6 ", %%mm5             \n\t" /* x3 */\
00869         "paddw " #m5 ", %%mm6             \n\t" /* x2 */\
00870         "paddw %%mm6, %%mm6               \n\t" /* 2x2 */\
00871         "psubw %%mm6, %%mm5               \n\t" /* -2x2 + x3 */\
00872         "pmullw "MANGLE(ff_pw_3)", %%mm5  \n\t" /* -6x2 + 3x3 */\
00873         "paddw " #rnd ", %%mm4            \n\t" /* x2 */\
00874         "paddw %%mm4, %%mm5               \n\t" /* 20x1 - 6x2 + 3x3 - x4 */\
00875         "psraw $5, %%mm5                  \n\t"\
00876         "packuswb %%mm5, %%mm5            \n\t"\
00877         OP(%%mm5, out, %%mm7, d)
00878 
00879 #define QPEL_BASE(OPNAME, ROUNDER, RND, OP_MMX2, OP_3DNOW)\
00880 static void OPNAME ## mpeg4_qpel16_h_lowpass_mmx2(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h){\
00881     uint64_t temp;\
00882 \
00883     __asm__ volatile(\
00884         "pxor %%mm7, %%mm7                \n\t"\
00885         "1:                               \n\t"\
00886         "movq  (%0), %%mm0                \n\t" /* ABCDEFGH */\
00887         "movq %%mm0, %%mm1                \n\t" /* ABCDEFGH */\
00888         "movq %%mm0, %%mm2                \n\t" /* ABCDEFGH */\
00889         "punpcklbw %%mm7, %%mm0           \n\t" /* 0A0B0C0D */\
00890         "punpckhbw %%mm7, %%mm1           \n\t" /* 0E0F0G0H */\
00891         "pshufw $0x90, %%mm0, %%mm5       \n\t" /* 0A0A0B0C */\
00892         "pshufw $0x41, %%mm0, %%mm6       \n\t" /* 0B0A0A0B */\
00893         "movq %%mm2, %%mm3                \n\t" /* ABCDEFGH */\
00894         "movq %%mm2, %%mm4                \n\t" /* ABCDEFGH */\
00895         "psllq $8, %%mm2                  \n\t" /* 0ABCDEFG */\
00896         "psllq $16, %%mm3                 \n\t" /* 00ABCDEF */\
00897         "psllq $24, %%mm4                 \n\t" /* 000ABCDE */\
00898         "punpckhbw %%mm7, %%mm2           \n\t" /* 0D0E0F0G */\
00899         "punpckhbw %%mm7, %%mm3           \n\t" /* 0C0D0E0F */\
00900         "punpckhbw %%mm7, %%mm4           \n\t" /* 0B0C0D0E */\
00901         "paddw %%mm3, %%mm5               \n\t" /* b */\
00902         "paddw %%mm2, %%mm6               \n\t" /* c */\
00903         "paddw %%mm5, %%mm5               \n\t" /* 2b */\
00904         "psubw %%mm5, %%mm6               \n\t" /* c - 2b */\
00905         "pshufw $0x06, %%mm0, %%mm5       \n\t" /* 0C0B0A0A */\
00906         "pmullw "MANGLE(ff_pw_3)", %%mm6  \n\t" /* 3c - 6b */\
00907         "paddw %%mm4, %%mm0               \n\t" /* a */\
00908         "paddw %%mm1, %%mm5               \n\t" /* d */\
00909         "pmullw "MANGLE(ff_pw_20)", %%mm0 \n\t" /* 20a */\
00910         "psubw %%mm5, %%mm0               \n\t" /* 20a - d */\
00911         "paddw %6, %%mm6                  \n\t"\
00912         "paddw %%mm6, %%mm0               \n\t" /* 20a - 6b + 3c - d */\
00913         "psraw $5, %%mm0                  \n\t"\
00914         "movq %%mm0, %5                   \n\t"\
00915         /* mm1=EFGH, mm2=DEFG, mm3=CDEF, mm4=BCDE, mm7=0 */\
00916         \
00917         "movq 5(%0), %%mm0                \n\t" /* FGHIJKLM */\
00918         "movq %%mm0, %%mm5                \n\t" /* FGHIJKLM */\
00919         "movq %%mm0, %%mm6                \n\t" /* FGHIJKLM */\
00920         "psrlq $8, %%mm0                  \n\t" /* GHIJKLM0 */\
00921         "psrlq $16, %%mm5                 \n\t" /* HIJKLM00 */\
00922         "punpcklbw %%mm7, %%mm0           \n\t" /* 0G0H0I0J */\
00923         "punpcklbw %%mm7, %%mm5           \n\t" /* 0H0I0J0K */\
00924         "paddw %%mm0, %%mm2               \n\t" /* b */\
00925         "paddw %%mm5, %%mm3               \n\t" /* c */\
00926         "paddw %%mm2, %%mm2               \n\t" /* 2b */\
00927         "psubw %%mm2, %%mm3               \n\t" /* c - 2b */\
00928         "movq %%mm6, %%mm2                \n\t" /* FGHIJKLM */\
00929         "psrlq $24, %%mm6                 \n\t" /* IJKLM000 */\
00930         "punpcklbw %%mm7, %%mm2           \n\t" /* 0F0G0H0I */\
00931         "punpcklbw %%mm7, %%mm6           \n\t" /* 0I0J0K0L */\
00932         "pmullw "MANGLE(ff_pw_3)", %%mm3  \n\t" /* 3c - 6b */\
00933         "paddw %%mm2, %%mm1               \n\t" /* a */\
00934         "paddw %%mm6, %%mm4               \n\t" /* d */\
00935         "pmullw "MANGLE(ff_pw_20)", %%mm1 \n\t" /* 20a */\
00936         "psubw %%mm4, %%mm3               \n\t" /* - 6b +3c - d */\
00937         "paddw %6, %%mm1                  \n\t"\
00938         "paddw %%mm1, %%mm3               \n\t" /* 20a - 6b +3c - d */\
00939         "psraw $5, %%mm3                  \n\t"\
00940         "movq %5, %%mm1                   \n\t"\
00941         "packuswb %%mm3, %%mm1            \n\t"\
00942         OP_MMX2(%%mm1, (%1),%%mm4, q)\
00943         /* mm0= GHIJ, mm2=FGHI, mm5=HIJK, mm6=IJKL, mm7=0 */\
00944         \
00945         "movq 9(%0), %%mm1                \n\t" /* JKLMNOPQ */\
00946         "movq %%mm1, %%mm4                \n\t" /* JKLMNOPQ */\
00947         "movq %%mm1, %%mm3                \n\t" /* JKLMNOPQ */\
00948         "psrlq $8, %%mm1                  \n\t" /* KLMNOPQ0 */\
00949         "psrlq $16, %%mm4                 \n\t" /* LMNOPQ00 */\
00950         "punpcklbw %%mm7, %%mm1           \n\t" /* 0K0L0M0N */\
00951         "punpcklbw %%mm7, %%mm4           \n\t" /* 0L0M0N0O */\
00952         "paddw %%mm1, %%mm5               \n\t" /* b */\
00953         "paddw %%mm4, %%mm0               \n\t" /* c */\
00954         "paddw %%mm5, %%mm5               \n\t" /* 2b */\
00955         "psubw %%mm5, %%mm0               \n\t" /* c - 2b */\
00956         "movq %%mm3, %%mm5                \n\t" /* JKLMNOPQ */\
00957         "psrlq $24, %%mm3                 \n\t" /* MNOPQ000 */\
00958         "pmullw "MANGLE(ff_pw_3)", %%mm0  \n\t" /* 3c - 6b */\
00959         "punpcklbw %%mm7, %%mm3           \n\t" /* 0M0N0O0P */\
00960         "paddw %%mm3, %%mm2               \n\t" /* d */\
00961         "psubw %%mm2, %%mm0               \n\t" /* -6b + 3c - d */\
00962         "movq %%mm5, %%mm2                \n\t" /* JKLMNOPQ */\
00963         "punpcklbw %%mm7, %%mm2           \n\t" /* 0J0K0L0M */\
00964         "punpckhbw %%mm7, %%mm5           \n\t" /* 0N0O0P0Q */\
00965         "paddw %%mm2, %%mm6               \n\t" /* a */\
00966         "pmullw "MANGLE(ff_pw_20)", %%mm6 \n\t" /* 20a */\
00967         "paddw %6, %%mm0                  \n\t"\
00968         "paddw %%mm6, %%mm0               \n\t" /* 20a - 6b + 3c - d */\
00969         "psraw $5, %%mm0                  \n\t"\
00970         /* mm1=KLMN, mm2=JKLM, mm3=MNOP, mm4=LMNO, mm5=NOPQ mm7=0 */\
00971         \
00972         "paddw %%mm5, %%mm3               \n\t" /* a */\
00973         "pshufw $0xF9, %%mm5, %%mm6       \n\t" /* 0O0P0Q0Q */\
00974         "paddw %%mm4, %%mm6               \n\t" /* b */\
00975         "pshufw $0xBE, %%mm5, %%mm4       \n\t" /* 0P0Q0Q0P */\
00976         "pshufw $0x6F, %%mm5, %%mm5       \n\t" /* 0Q0Q0P0O */\
00977         "paddw %%mm1, %%mm4               \n\t" /* c */\
00978         "paddw %%mm2, %%mm5               \n\t" /* d */\
00979         "paddw %%mm6, %%mm6               \n\t" /* 2b */\
00980         "psubw %%mm6, %%mm4               \n\t" /* c - 2b */\
00981         "pmullw "MANGLE(ff_pw_20)", %%mm3 \n\t" /* 20a */\
00982         "pmullw "MANGLE(ff_pw_3)", %%mm4  \n\t" /* 3c - 6b */\
00983         "psubw %%mm5, %%mm3               \n\t" /* -6b + 3c - d */\
00984         "paddw %6, %%mm4                  \n\t"\
00985         "paddw %%mm3, %%mm4               \n\t" /* 20a - 6b + 3c - d */\
00986         "psraw $5, %%mm4                  \n\t"\
00987         "packuswb %%mm4, %%mm0            \n\t"\
00988         OP_MMX2(%%mm0, 8(%1), %%mm4, q)\
00989         \
00990         "add %3, %0                       \n\t"\
00991         "add %4, %1                       \n\t"\
00992         "decl %2                          \n\t"\
00993         " jnz 1b                          \n\t"\
00994         : "+a"(src), "+c"(dst), "+D"(h)\
00995         : "d"((x86_reg)srcStride), "S"((x86_reg)dstStride), /*"m"(ff_pw_20), "m"(ff_pw_3),*/ "m"(temp), "m"(ROUNDER)\
00996         : "memory"\
00997     );\
00998 }\
00999 \
01000 static void OPNAME ## mpeg4_qpel16_h_lowpass_3dnow(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h){\
01001     int i;\
01002     int16_t temp[16];\
01003     /* quick HACK, XXX FIXME MUST be optimized */\
01004     for(i=0; i<h; i++)\
01005     {\
01006         temp[ 0]= (src[ 0]+src[ 1])*20 - (src[ 0]+src[ 2])*6 + (src[ 1]+src[ 3])*3 - (src[ 2]+src[ 4]);\
01007         temp[ 1]= (src[ 1]+src[ 2])*20 - (src[ 0]+src[ 3])*6 + (src[ 0]+src[ 4])*3 - (src[ 1]+src[ 5]);\
01008         temp[ 2]= (src[ 2]+src[ 3])*20 - (src[ 1]+src[ 4])*6 + (src[ 0]+src[ 5])*3 - (src[ 0]+src[ 6]);\
01009         temp[ 3]= (src[ 3]+src[ 4])*20 - (src[ 2]+src[ 5])*6 + (src[ 1]+src[ 6])*3 - (src[ 0]+src[ 7]);\
01010         temp[ 4]= (src[ 4]+src[ 5])*20 - (src[ 3]+src[ 6])*6 + (src[ 2]+src[ 7])*3 - (src[ 1]+src[ 8]);\
01011         temp[ 5]= (src[ 5]+src[ 6])*20 - (src[ 4]+src[ 7])*6 + (src[ 3]+src[ 8])*3 - (src[ 2]+src[ 9]);\
01012         temp[ 6]= (src[ 6]+src[ 7])*20 - (src[ 5]+src[ 8])*6 + (src[ 4]+src[ 9])*3 - (src[ 3]+src[10]);\
01013         temp[ 7]= (src[ 7]+src[ 8])*20 - (src[ 6]+src[ 9])*6 + (src[ 5]+src[10])*3 - (src[ 4]+src[11]);\
01014         temp[ 8]= (src[ 8]+src[ 9])*20 - (src[ 7]+src[10])*6 + (src[ 6]+src[11])*3 - (src[ 5]+src[12]);\
01015         temp[ 9]= (src[ 9]+src[10])*20 - (src[ 8]+src[11])*6 + (src[ 7]+src[12])*3 - (src[ 6]+src[13]);\
01016         temp[10]= (src[10]+src[11])*20 - (src[ 9]+src[12])*6 + (src[ 8]+src[13])*3 - (src[ 7]+src[14]);\
01017         temp[11]= (src[11]+src[12])*20 - (src[10]+src[13])*6 + (src[ 9]+src[14])*3 - (src[ 8]+src[15]);\
01018         temp[12]= (src[12]+src[13])*20 - (src[11]+src[14])*6 + (src[10]+src[15])*3 - (src[ 9]+src[16]);\
01019         temp[13]= (src[13]+src[14])*20 - (src[12]+src[15])*6 + (src[11]+src[16])*3 - (src[10]+src[16]);\
01020         temp[14]= (src[14]+src[15])*20 - (src[13]+src[16])*6 + (src[12]+src[16])*3 - (src[11]+src[15]);\
01021         temp[15]= (src[15]+src[16])*20 - (src[14]+src[16])*6 + (src[13]+src[15])*3 - (src[12]+src[14]);\
01022         __asm__ volatile(\
01023             "movq (%0), %%mm0               \n\t"\
01024             "movq 8(%0), %%mm1              \n\t"\
01025             "paddw %2, %%mm0                \n\t"\
01026             "paddw %2, %%mm1                \n\t"\
01027             "psraw $5, %%mm0                \n\t"\
01028             "psraw $5, %%mm1                \n\t"\
01029             "packuswb %%mm1, %%mm0          \n\t"\
01030             OP_3DNOW(%%mm0, (%1), %%mm1, q)\
01031             "movq 16(%0), %%mm0             \n\t"\
01032             "movq 24(%0), %%mm1             \n\t"\
01033             "paddw %2, %%mm0                \n\t"\
01034             "paddw %2, %%mm1                \n\t"\
01035             "psraw $5, %%mm0                \n\t"\
01036             "psraw $5, %%mm1                \n\t"\
01037             "packuswb %%mm1, %%mm0          \n\t"\
01038             OP_3DNOW(%%mm0, 8(%1), %%mm1, q)\
01039             :: "r"(temp), "r"(dst), "m"(ROUNDER)\
01040             : "memory"\
01041         );\
01042         dst+=dstStride;\
01043         src+=srcStride;\
01044     }\
01045 }\
01046 \
01047 static void OPNAME ## mpeg4_qpel8_h_lowpass_mmx2(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h){\
01048     __asm__ volatile(\
01049         "pxor %%mm7, %%mm7                \n\t"\
01050         "1:                               \n\t"\
01051         "movq  (%0), %%mm0                \n\t" /* ABCDEFGH */\
01052         "movq %%mm0, %%mm1                \n\t" /* ABCDEFGH */\
01053         "movq %%mm0, %%mm2                \n\t" /* ABCDEFGH */\
01054         "punpcklbw %%mm7, %%mm0           \n\t" /* 0A0B0C0D */\
01055         "punpckhbw %%mm7, %%mm1           \n\t" /* 0E0F0G0H */\
01056         "pshufw $0x90, %%mm0, %%mm5       \n\t" /* 0A0A0B0C */\
01057         "pshufw $0x41, %%mm0, %%mm6       \n\t" /* 0B0A0A0B */\
01058         "movq %%mm2, %%mm3                \n\t" /* ABCDEFGH */\
01059         "movq %%mm2, %%mm4                \n\t" /* ABCDEFGH */\
01060         "psllq $8, %%mm2                  \n\t" /* 0ABCDEFG */\
01061         "psllq $16, %%mm3                 \n\t" /* 00ABCDEF */\
01062         "psllq $24, %%mm4                 \n\t" /* 000ABCDE */\
01063         "punpckhbw %%mm7, %%mm2           \n\t" /* 0D0E0F0G */\
01064         "punpckhbw %%mm7, %%mm3           \n\t" /* 0C0D0E0F */\
01065         "punpckhbw %%mm7, %%mm4           \n\t" /* 0B0C0D0E */\
01066         "paddw %%mm3, %%mm5               \n\t" /* b */\
01067         "paddw %%mm2, %%mm6               \n\t" /* c */\
01068         "paddw %%mm5, %%mm5               \n\t" /* 2b */\
01069         "psubw %%mm5, %%mm6               \n\t" /* c - 2b */\
01070         "pshufw $0x06, %%mm0, %%mm5       \n\t" /* 0C0B0A0A */\
01071         "pmullw "MANGLE(ff_pw_3)", %%mm6  \n\t" /* 3c - 6b */\
01072         "paddw %%mm4, %%mm0               \n\t" /* a */\
01073         "paddw %%mm1, %%mm5               \n\t" /* d */\
01074         "pmullw "MANGLE(ff_pw_20)", %%mm0 \n\t" /* 20a */\
01075         "psubw %%mm5, %%mm0               \n\t" /* 20a - d */\
01076         "paddw %5, %%mm6                  \n\t"\
01077         "paddw %%mm6, %%mm0               \n\t" /* 20a - 6b + 3c - d */\
01078         "psraw $5, %%mm0                  \n\t"\
01079         /* mm1=EFGH, mm2=DEFG, mm3=CDEF, mm4=BCDE, mm7=0 */\
01080         \
01081         "movd 5(%0), %%mm5                \n\t" /* FGHI */\
01082         "punpcklbw %%mm7, %%mm5           \n\t" /* 0F0G0H0I */\
01083         "pshufw $0xF9, %%mm5, %%mm6       \n\t" /* 0G0H0I0I */\
01084         "paddw %%mm5, %%mm1               \n\t" /* a */\
01085         "paddw %%mm6, %%mm2               \n\t" /* b */\
01086         "pshufw $0xBE, %%mm5, %%mm6       \n\t" /* 0H0I0I0H */\
01087         "pshufw $0x6F, %%mm5, %%mm5       \n\t" /* 0I0I0H0G */\
01088         "paddw %%mm6, %%mm3               \n\t" /* c */\
01089         "paddw %%mm5, %%mm4               \n\t" /* d */\
01090         "paddw %%mm2, %%mm2               \n\t" /* 2b */\
01091         "psubw %%mm2, %%mm3               \n\t" /* c - 2b */\
01092         "pmullw "MANGLE(ff_pw_20)", %%mm1 \n\t" /* 20a */\
01093         "pmullw "MANGLE(ff_pw_3)", %%mm3  \n\t" /* 3c - 6b */\
01094         "psubw %%mm4, %%mm3               \n\t" /* -6b + 3c - d */\
01095         "paddw %5, %%mm1                  \n\t"\
01096         "paddw %%mm1, %%mm3               \n\t" /* 20a - 6b + 3c - d */\
01097         "psraw $5, %%mm3                  \n\t"\
01098         "packuswb %%mm3, %%mm0            \n\t"\
01099         OP_MMX2(%%mm0, (%1), %%mm4, q)\
01100         \
01101         "add %3, %0                       \n\t"\
01102         "add %4, %1                       \n\t"\
01103         "decl %2                          \n\t"\
01104         " jnz 1b                          \n\t"\
01105         : "+a"(src), "+c"(dst), "+d"(h)\
01106         : "S"((x86_reg)srcStride), "D"((x86_reg)dstStride), /*"m"(ff_pw_20), "m"(ff_pw_3),*/ "m"(ROUNDER)\
01107         : "memory"\
01108     );\
01109 }\
01110 \
01111 static void OPNAME ## mpeg4_qpel8_h_lowpass_3dnow(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h){\
01112     int i;\
01113     int16_t temp[8];\
01114     /* quick HACK, XXX FIXME MUST be optimized */\
01115     for(i=0; i<h; i++)\
01116     {\
01117         temp[ 0]= (src[ 0]+src[ 1])*20 - (src[ 0]+src[ 2])*6 + (src[ 1]+src[ 3])*3 - (src[ 2]+src[ 4]);\
01118         temp[ 1]= (src[ 1]+src[ 2])*20 - (src[ 0]+src[ 3])*6 + (src[ 0]+src[ 4])*3 - (src[ 1]+src[ 5]);\
01119         temp[ 2]= (src[ 2]+src[ 3])*20 - (src[ 1]+src[ 4])*6 + (src[ 0]+src[ 5])*3 - (src[ 0]+src[ 6]);\
01120         temp[ 3]= (src[ 3]+src[ 4])*20 - (src[ 2]+src[ 5])*6 + (src[ 1]+src[ 6])*3 - (src[ 0]+src[ 7]);\
01121         temp[ 4]= (src[ 4]+src[ 5])*20 - (src[ 3]+src[ 6])*6 + (src[ 2]+src[ 7])*3 - (src[ 1]+src[ 8]);\
01122         temp[ 5]= (src[ 5]+src[ 6])*20 - (src[ 4]+src[ 7])*6 + (src[ 3]+src[ 8])*3 - (src[ 2]+src[ 8]);\
01123         temp[ 6]= (src[ 6]+src[ 7])*20 - (src[ 5]+src[ 8])*6 + (src[ 4]+src[ 8])*3 - (src[ 3]+src[ 7]);\
01124         temp[ 7]= (src[ 7]+src[ 8])*20 - (src[ 6]+src[ 8])*6 + (src[ 5]+src[ 7])*3 - (src[ 4]+src[ 6]);\
01125         __asm__ volatile(\
01126             "movq (%0), %%mm0           \n\t"\
01127             "movq 8(%0), %%mm1          \n\t"\
01128             "paddw %2, %%mm0            \n\t"\
01129             "paddw %2, %%mm1            \n\t"\
01130             "psraw $5, %%mm0            \n\t"\
01131             "psraw $5, %%mm1            \n\t"\
01132             "packuswb %%mm1, %%mm0      \n\t"\
01133             OP_3DNOW(%%mm0, (%1), %%mm1, q)\
01134             :: "r"(temp), "r"(dst), "m"(ROUNDER)\
01135             :"memory"\
01136         );\
01137         dst+=dstStride;\
01138         src+=srcStride;\
01139     }\
01140 }
01141 
01142 #define QPEL_OP(OPNAME, ROUNDER, RND, OP, MMX)\
01143 \
01144 static void OPNAME ## mpeg4_qpel16_v_lowpass_ ## MMX(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
01145     uint64_t temp[17*4];\
01146     uint64_t *temp_ptr= temp;\
01147     int count= 17;\
01148 \
01149     /*FIXME unroll */\
01150     __asm__ volatile(\
01151         "pxor %%mm7, %%mm7              \n\t"\
01152         "1:                             \n\t"\
01153         "movq (%0), %%mm0               \n\t"\
01154         "movq (%0), %%mm1               \n\t"\
01155         "movq 8(%0), %%mm2              \n\t"\
01156         "movq 8(%0), %%mm3              \n\t"\
01157         "punpcklbw %%mm7, %%mm0         \n\t"\
01158         "punpckhbw %%mm7, %%mm1         \n\t"\
01159         "punpcklbw %%mm7, %%mm2         \n\t"\
01160         "punpckhbw %%mm7, %%mm3         \n\t"\
01161         "movq %%mm0, (%1)               \n\t"\
01162         "movq %%mm1, 17*8(%1)           \n\t"\
01163         "movq %%mm2, 2*17*8(%1)         \n\t"\
01164         "movq %%mm3, 3*17*8(%1)         \n\t"\
01165         "add $8, %1                     \n\t"\
01166         "add %3, %0                     \n\t"\
01167         "decl %2                        \n\t"\
01168         " jnz 1b                        \n\t"\
01169         : "+r" (src), "+r" (temp_ptr), "+r"(count)\
01170         : "r" ((x86_reg)srcStride)\
01171         : "memory"\
01172     );\
01173     \
01174     temp_ptr= temp;\
01175     count=4;\
01176     \
01177 /*FIXME reorder for speed */\
01178     __asm__ volatile(\
01179         /*"pxor %%mm7, %%mm7              \n\t"*/\
01180         "1:                             \n\t"\
01181         "movq (%0), %%mm0               \n\t"\
01182         "movq 8(%0), %%mm1              \n\t"\
01183         "movq 16(%0), %%mm2             \n\t"\
01184         "movq 24(%0), %%mm3             \n\t"\
01185         QPEL_V_LOW(%%mm0, %%mm1, %%mm2, %%mm3, %5, %6, %5, 16(%0),  8(%0),   (%0), 32(%0), (%1), OP)\
01186         QPEL_V_LOW(%%mm1, %%mm2, %%mm3, %%mm0, %5, %6, %5,  8(%0),   (%0),   (%0), 40(%0), (%1, %3), OP)\
01187         "add %4, %1                     \n\t"\
01188         QPEL_V_LOW(%%mm2, %%mm3, %%mm0, %%mm1, %5, %6, %5,   (%0),   (%0),  8(%0), 48(%0), (%1), OP)\
01189         \
01190         QPEL_V_LOW(%%mm3, %%mm0, %%mm1, %%mm2, %5, %6, %5,   (%0),  8(%0), 16(%0), 56(%0), (%1, %3), OP)\
01191         "add %4, %1                     \n\t"\
01192         QPEL_V_LOW(%%mm0, %%mm1, %%mm2, %%mm3, %5, %6, %5,  8(%0), 16(%0), 24(%0), 64(%0), (%1), OP)\
01193         QPEL_V_LOW(%%mm1, %%mm2, %%mm3, %%mm0, %5, %6, %5, 16(%0), 24(%0), 32(%0), 72(%0), (%1, %3), OP)\
01194         "add %4, %1                     \n\t"\
01195         QPEL_V_LOW(%%mm2, %%mm3, %%mm0, %%mm1, %5, %6, %5, 24(%0), 32(%0), 40(%0), 80(%0), (%1), OP)\
01196         QPEL_V_LOW(%%mm3, %%mm0, %%mm1, %%mm2, %5, %6, %5, 32(%0), 40(%0), 48(%0), 88(%0), (%1, %3), OP)\
01197         "add %4, %1                     \n\t"\
01198         QPEL_V_LOW(%%mm0, %%mm1, %%mm2, %%mm3, %5, %6, %5, 40(%0), 48(%0), 56(%0), 96(%0), (%1), OP)\
01199         QPEL_V_LOW(%%mm1, %%mm2, %%mm3, %%mm0, %5, %6, %5, 48(%0), 56(%0), 64(%0),104(%0), (%1, %3), OP)\
01200         "add %4, %1                     \n\t"\
01201         QPEL_V_LOW(%%mm2, %%mm3, %%mm0, %%mm1, %5, %6, %5, 56(%0), 64(%0), 72(%0),112(%0), (%1), OP)\
01202         QPEL_V_LOW(%%mm3, %%mm0, %%mm1, %%mm2, %5, %6, %5, 64(%0), 72(%0), 80(%0),120(%0), (%1, %3), OP)\
01203         "add %4, %1                     \n\t"\
01204         QPEL_V_LOW(%%mm0, %%mm1, %%mm2, %%mm3, %5, %6, %5, 72(%0), 80(%0), 88(%0),128(%0), (%1), OP)\
01205         \
01206         QPEL_V_LOW(%%mm1, %%mm2, %%mm3, %%mm0, %5, %6, %5, 80(%0), 88(%0), 96(%0),128(%0), (%1, %3), OP)\
01207         "add %4, %1                     \n\t"  \
01208         QPEL_V_LOW(%%mm2, %%mm3, %%mm0, %%mm1, %5, %6, %5, 88(%0), 96(%0),104(%0),120(%0), (%1), OP)\
01209         QPEL_V_LOW(%%mm3, %%mm0, %%mm1, %%mm2, %5, %6, %5, 96(%0),104(%0),112(%0),112(%0), (%1, %3), OP)\
01210         \
01211         "add $136, %0                   \n\t"\
01212         "add %6, %1                     \n\t"\
01213         "decl %2                        \n\t"\
01214         " jnz 1b                        \n\t"\
01215         \
01216         : "+r"(temp_ptr), "+r"(dst), "+g"(count)\
01217         : "r"((x86_reg)dstStride), "r"(2*(x86_reg)dstStride), /*"m"(ff_pw_20), "m"(ff_pw_3),*/ "m"(ROUNDER), "g"(4-14*(x86_reg)dstStride)\
01218         :"memory"\
01219     );\
01220 }\
01221 \
01222 static void OPNAME ## mpeg4_qpel8_v_lowpass_ ## MMX(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
01223     uint64_t temp[9*2];\
01224     uint64_t *temp_ptr= temp;\
01225     int count= 9;\
01226 \
01227     /*FIXME unroll */\
01228     __asm__ volatile(\
01229         "pxor %%mm7, %%mm7              \n\t"\
01230         "1:                             \n\t"\
01231         "movq (%0), %%mm0               \n\t"\
01232         "movq (%0), %%mm1               \n\t"\
01233         "punpcklbw %%mm7, %%mm0         \n\t"\
01234         "punpckhbw %%mm7, %%mm1         \n\t"\
01235         "movq %%mm0, (%1)               \n\t"\
01236         "movq %%mm1, 9*8(%1)            \n\t"\
01237         "add $8, %1                     \n\t"\
01238         "add %3, %0                     \n\t"\
01239         "decl %2                        \n\t"\
01240         " jnz 1b                        \n\t"\
01241         : "+r" (src), "+r" (temp_ptr), "+r"(count)\
01242         : "r" ((x86_reg)srcStride)\
01243         : "memory"\
01244     );\
01245     \
01246     temp_ptr= temp;\
01247     count=2;\
01248     \
01249 /*FIXME reorder for speed */\
01250     __asm__ volatile(\
01251         /*"pxor %%mm7, %%mm7              \n\t"*/\
01252         "1:                             \n\t"\
01253         "movq (%0), %%mm0               \n\t"\
01254         "movq 8(%0), %%mm1              \n\t"\
01255         "movq 16(%0), %%mm2             \n\t"\
01256         "movq 24(%0), %%mm3             \n\t"\
01257         QPEL_V_LOW(%%mm0, %%mm1, %%mm2, %%mm3, %5, %6, %5, 16(%0),  8(%0),   (%0), 32(%0), (%1), OP)\
01258         QPEL_V_LOW(%%mm1, %%mm2, %%mm3, %%mm0, %5, %6, %5,  8(%0),   (%0),   (%0), 40(%0), (%1, %3), OP)\
01259         "add %4, %1                     \n\t"\
01260         QPEL_V_LOW(%%mm2, %%mm3, %%mm0, %%mm1, %5, %6, %5,   (%0),   (%0),  8(%0), 48(%0), (%1), OP)\
01261         \
01262         QPEL_V_LOW(%%mm3, %%mm0, %%mm1, %%mm2, %5, %6, %5,   (%0),  8(%0), 16(%0), 56(%0), (%1, %3), OP)\
01263         "add %4, %1                     \n\t"\
01264         QPEL_V_LOW(%%mm0, %%mm1, %%mm2, %%mm3, %5, %6, %5,  8(%0), 16(%0), 24(%0), 64(%0), (%1), OP)\
01265         \
01266         QPEL_V_LOW(%%mm1, %%mm2, %%mm3, %%mm0, %5, %6, %5, 16(%0), 24(%0), 32(%0), 64(%0), (%1, %3), OP)\
01267         "add %4, %1                     \n\t"\
01268         QPEL_V_LOW(%%mm2, %%mm3, %%mm0, %%mm1, %5, %6, %5, 24(%0), 32(%0), 40(%0), 56(%0), (%1), OP)\
01269         QPEL_V_LOW(%%mm3, %%mm0, %%mm1, %%mm2, %5, %6, %5, 32(%0), 40(%0), 48(%0), 48(%0), (%1, %3), OP)\
01270                 \
01271         "add $72, %0                    \n\t"\
01272         "add %6, %1                     \n\t"\
01273         "decl %2                        \n\t"\
01274         " jnz 1b                        \n\t"\
01275          \
01276         : "+r"(temp_ptr), "+r"(dst), "+g"(count)\
01277         : "r"((x86_reg)dstStride), "r"(2*(x86_reg)dstStride), /*"m"(ff_pw_20), "m"(ff_pw_3),*/ "m"(ROUNDER), "g"(4-6*(x86_reg)dstStride)\
01278         : "memory"\
01279    );\
01280 }\
01281 \
01282 static void OPNAME ## qpel8_mc00_ ## MMX (uint8_t *dst, uint8_t *src, int stride){\
01283     OPNAME ## pixels8_ ## MMX(dst, src, stride, 8);\
01284 }\
01285 \
01286 static void OPNAME ## qpel8_mc10_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\
01287     uint64_t temp[8];\
01288     uint8_t * const half= (uint8_t*)temp;\
01289     put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(half, src, 8, stride, 8);\
01290     OPNAME ## pixels8_l2_ ## MMX(dst, src, half, stride, stride, 8);\
01291 }\
01292 \
01293 static void OPNAME ## qpel8_mc20_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\
01294     OPNAME ## mpeg4_qpel8_h_lowpass_ ## MMX(dst, src, stride, stride, 8);\
01295 }\
01296 \
01297 static void OPNAME ## qpel8_mc30_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\
01298     uint64_t temp[8];\
01299     uint8_t * const half= (uint8_t*)temp;\
01300     put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(half, src, 8, stride, 8);\
01301     OPNAME ## pixels8_l2_ ## MMX(dst, src+1, half, stride, stride, 8);\
01302 }\
01303 \
01304 static void OPNAME ## qpel8_mc01_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\
01305     uint64_t temp[8];\
01306     uint8_t * const half= (uint8_t*)temp;\
01307     put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(half, src, 8, stride);\
01308     OPNAME ## pixels8_l2_ ## MMX(dst, src, half, stride, stride, 8);\
01309 }\
01310 \
01311 static void OPNAME ## qpel8_mc02_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\
01312     OPNAME ## mpeg4_qpel8_v_lowpass_ ## MMX(dst, src, stride, stride);\
01313 }\
01314 \
01315 static void OPNAME ## qpel8_mc03_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\
01316     uint64_t temp[8];\
01317     uint8_t * const half= (uint8_t*)temp;\
01318     put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(half, src, 8, stride);\
01319     OPNAME ## pixels8_l2_ ## MMX(dst, src+stride, half, stride, stride, 8);\
01320 }\
01321 static void OPNAME ## qpel8_mc11_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\
01322     uint64_t half[8 + 9];\
01323     uint8_t * const halfH= ((uint8_t*)half) + 64;\
01324     uint8_t * const halfHV= ((uint8_t*)half);\
01325     put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8, stride, 9);\
01326     put ## RND ## pixels8_l2_ ## MMX(halfH, src, halfH, 8, stride, 9);\
01327     put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(halfHV, halfH, 8, 8);\
01328     OPNAME ## pixels8_l2_ ## MMX(dst, halfH, halfHV, stride, 8, 8);\
01329 }\
01330 static void OPNAME ## qpel8_mc31_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\
01331     uint64_t half[8 + 9];\
01332     uint8_t * const halfH= ((uint8_t*)half) + 64;\
01333     uint8_t * const halfHV= ((uint8_t*)half);\
01334     put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8, stride, 9);\
01335     put ## RND ## pixels8_l2_ ## MMX(halfH, src+1, halfH, 8, stride, 9);\
01336     put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(halfHV, halfH, 8, 8);\
01337     OPNAME ## pixels8_l2_ ## MMX(dst, halfH, halfHV, stride, 8, 8);\
01338 }\
01339 static void OPNAME ## qpel8_mc13_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\
01340     uint64_t half[8 + 9];\
01341     uint8_t * const halfH= ((uint8_t*)half) + 64;\
01342     uint8_t * const halfHV= ((uint8_t*)half);\
01343     put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8, stride, 9);\
01344     put ## RND ## pixels8_l2_ ## MMX(halfH, src, halfH, 8, stride, 9);\
01345     put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(halfHV, halfH, 8, 8);\
01346     OPNAME ## pixels8_l2_ ## MMX(dst, halfH+8, halfHV, stride, 8, 8);\
01347 }\
01348 static void OPNAME ## qpel8_mc33_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\
01349     uint64_t half[8 + 9];\
01350     uint8_t * const halfH= ((uint8_t*)half) + 64;\
01351     uint8_t * const halfHV= ((uint8_t*)half);\
01352     put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8, stride, 9);\
01353     put ## RND ## pixels8_l2_ ## MMX(halfH, src+1, halfH, 8, stride, 9);\
01354     put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(halfHV, halfH, 8, 8);\
01355     OPNAME ## pixels8_l2_ ## MMX(dst, halfH+8, halfHV, stride, 8, 8);\
01356 }\
01357 static void OPNAME ## qpel8_mc21_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\
01358     uint64_t half[8 + 9];\
01359     uint8_t * const halfH= ((uint8_t*)half) + 64;\
01360     uint8_t * const halfHV= ((uint8_t*)half);\
01361     put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8, stride, 9);\
01362     put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(halfHV, halfH, 8, 8);\
01363     OPNAME ## pixels8_l2_ ## MMX(dst, halfH, halfHV, stride, 8, 8);\
01364 }\
01365 static void OPNAME ## qpel8_mc23_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\
01366     uint64_t half[8 + 9];\
01367     uint8_t * const halfH= ((uint8_t*)half) + 64;\
01368     uint8_t * const halfHV= ((uint8_t*)half);\
01369     put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8, stride, 9);\
01370     put ## RND ## mpeg4_qpel8_v_lowpass_ ## MMX(halfHV, halfH, 8, 8);\
01371     OPNAME ## pixels8_l2_ ## MMX(dst, halfH+8, halfHV, stride, 8, 8);\
01372 }\
01373 static void OPNAME ## qpel8_mc12_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\
01374     uint64_t half[8 + 9];\
01375     uint8_t * const halfH= ((uint8_t*)half);\
01376     put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8, stride, 9);\
01377     put ## RND ## pixels8_l2_ ## MMX(halfH, src, halfH, 8, stride, 9);\
01378     OPNAME ## mpeg4_qpel8_v_lowpass_ ## MMX(dst, halfH, stride, 8);\
01379 }\
01380 static void OPNAME ## qpel8_mc32_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\
01381     uint64_t half[8 + 9];\
01382     uint8_t * const halfH= ((uint8_t*)half);\
01383     put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8, stride, 9);\
01384     put ## RND ## pixels8_l2_ ## MMX(halfH, src+1, halfH, 8, stride, 9);\
01385     OPNAME ## mpeg4_qpel8_v_lowpass_ ## MMX(dst, halfH, stride, 8);\
01386 }\
01387 static void OPNAME ## qpel8_mc22_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\
01388     uint64_t half[9];\
01389     uint8_t * const halfH= ((uint8_t*)half);\
01390     put ## RND ## mpeg4_qpel8_h_lowpass_ ## MMX(halfH, src, 8, stride, 9);\
01391     OPNAME ## mpeg4_qpel8_v_lowpass_ ## MMX(dst, halfH, stride, 8);\
01392 }\
01393 static void OPNAME ## qpel16_mc00_ ## MMX (uint8_t *dst, uint8_t *src, int stride){\
01394     OPNAME ## pixels16_ ## MMX(dst, src, stride, 16);\
01395 }\
01396 \
01397 static void OPNAME ## qpel16_mc10_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\
01398     uint64_t temp[32];\
01399     uint8_t * const half= (uint8_t*)temp;\
01400     put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(half, src, 16, stride, 16);\
01401     OPNAME ## pixels16_l2_ ## MMX(dst, src, half, stride, stride, 16);\
01402 }\
01403 \
01404 static void OPNAME ## qpel16_mc20_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\
01405     OPNAME ## mpeg4_qpel16_h_lowpass_ ## MMX(dst, src, stride, stride, 16);\
01406 }\
01407 \
01408 static void OPNAME ## qpel16_mc30_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\
01409     uint64_t temp[32];\
01410     uint8_t * const half= (uint8_t*)temp;\
01411     put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(half, src, 16, stride, 16);\
01412     OPNAME ## pixels16_l2_ ## MMX(dst, src+1, half, stride, stride, 16);\
01413 }\
01414 \
01415 static void OPNAME ## qpel16_mc01_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\
01416     uint64_t temp[32];\
01417     uint8_t * const half= (uint8_t*)temp;\
01418     put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(half, src, 16, stride);\
01419     OPNAME ## pixels16_l2_ ## MMX(dst, src, half, stride, stride, 16);\
01420 }\
01421 \
01422 static void OPNAME ## qpel16_mc02_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\
01423     OPNAME ## mpeg4_qpel16_v_lowpass_ ## MMX(dst, src, stride, stride);\
01424 }\
01425 \
01426 static void OPNAME ## qpel16_mc03_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\
01427     uint64_t temp[32];\
01428     uint8_t * const half= (uint8_t*)temp;\
01429     put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(half, src, 16, stride);\
01430     OPNAME ## pixels16_l2_ ## MMX(dst, src+stride, half, stride, stride, 16);\
01431 }\
01432 static void OPNAME ## qpel16_mc11_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\
01433     uint64_t half[16*2 + 17*2];\
01434     uint8_t * const halfH= ((uint8_t*)half) + 256;\
01435     uint8_t * const halfHV= ((uint8_t*)half);\
01436     put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16, stride, 17);\
01437     put ## RND ## pixels16_l2_ ## MMX(halfH, src, halfH, 16, stride, 17);\
01438     put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(halfHV, halfH, 16, 16);\
01439     OPNAME ## pixels16_l2_ ## MMX(dst, halfH, halfHV, stride, 16, 16);\
01440 }\
01441 static void OPNAME ## qpel16_mc31_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\
01442     uint64_t half[16*2 + 17*2];\
01443     uint8_t * const halfH= ((uint8_t*)half) + 256;\
01444     uint8_t * const halfHV= ((uint8_t*)half);\
01445     put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16, stride, 17);\
01446     put ## RND ## pixels16_l2_ ## MMX(halfH, src+1, halfH, 16, stride, 17);\
01447     put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(halfHV, halfH, 16, 16);\
01448     OPNAME ## pixels16_l2_ ## MMX(dst, halfH, halfHV, stride, 16, 16);\
01449 }\
01450 static void OPNAME ## qpel16_mc13_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\
01451     uint64_t half[16*2 + 17*2];\
01452     uint8_t * const halfH= ((uint8_t*)half) + 256;\
01453     uint8_t * const halfHV= ((uint8_t*)half);\
01454     put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16, stride, 17);\
01455     put ## RND ## pixels16_l2_ ## MMX(halfH, src, halfH, 16, stride, 17);\
01456     put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(halfHV, halfH, 16, 16);\
01457     OPNAME ## pixels16_l2_ ## MMX(dst, halfH+16, halfHV, stride, 16, 16);\
01458 }\
01459 static void OPNAME ## qpel16_mc33_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\
01460     uint64_t half[16*2 + 17*2];\
01461     uint8_t * const halfH= ((uint8_t*)half) + 256;\
01462     uint8_t * const halfHV= ((uint8_t*)half);\
01463     put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16, stride, 17);\
01464     put ## RND ## pixels16_l2_ ## MMX(halfH, src+1, halfH, 16, stride, 17);\
01465     put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(halfHV, halfH, 16, 16);\
01466     OPNAME ## pixels16_l2_ ## MMX(dst, halfH+16, halfHV, stride, 16, 16);\
01467 }\
01468 static void OPNAME ## qpel16_mc21_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\
01469     uint64_t half[16*2 + 17*2];\
01470     uint8_t * const halfH= ((uint8_t*)half) + 256;\
01471     uint8_t * const halfHV= ((uint8_t*)half);\
01472     put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16, stride, 17);\
01473     put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(halfHV, halfH, 16, 16);\
01474     OPNAME ## pixels16_l2_ ## MMX(dst, halfH, halfHV, stride, 16, 16);\
01475 }\
01476 static void OPNAME ## qpel16_mc23_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\
01477     uint64_t half[16*2 + 17*2];\
01478     uint8_t * const halfH= ((uint8_t*)half) + 256;\
01479     uint8_t * const halfHV= ((uint8_t*)half);\
01480     put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16, stride, 17);\
01481     put ## RND ## mpeg4_qpel16_v_lowpass_ ## MMX(halfHV, halfH, 16, 16);\
01482     OPNAME ## pixels16_l2_ ## MMX(dst, halfH+16, halfHV, stride, 16, 16);\
01483 }\
01484 static void OPNAME ## qpel16_mc12_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\
01485     uint64_t half[17*2];\
01486     uint8_t * const halfH= ((uint8_t*)half);\
01487     put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16, stride, 17);\
01488     put ## RND ## pixels16_l2_ ## MMX(halfH, src, halfH, 16, stride, 17);\
01489     OPNAME ## mpeg4_qpel16_v_lowpass_ ## MMX(dst, halfH, stride, 16);\
01490 }\
01491 static void OPNAME ## qpel16_mc32_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\
01492     uint64_t half[17*2];\
01493     uint8_t * const halfH= ((uint8_t*)half);\
01494     put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16, stride, 17);\
01495     put ## RND ## pixels16_l2_ ## MMX(halfH, src+1, halfH, 16, stride, 17);\
01496     OPNAME ## mpeg4_qpel16_v_lowpass_ ## MMX(dst, halfH, stride, 16);\
01497 }\
01498 static void OPNAME ## qpel16_mc22_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\
01499     uint64_t half[17*2];\
01500     uint8_t * const halfH= ((uint8_t*)half);\
01501     put ## RND ## mpeg4_qpel16_h_lowpass_ ## MMX(halfH, src, 16, stride, 17);\
01502     OPNAME ## mpeg4_qpel16_v_lowpass_ ## MMX(dst, halfH, stride, 16);\
01503 }
01504 
01505 #define PUT_OP(a,b,temp, size) "mov" #size " " #a ", " #b "        \n\t"
01506 #define AVG_3DNOW_OP(a,b,temp, size) \
01507 "mov" #size " " #b ", " #temp "   \n\t"\
01508 "pavgusb " #temp ", " #a "        \n\t"\
01509 "mov" #size " " #a ", " #b "      \n\t"
01510 #define AVG_MMX2_OP(a,b,temp, size) \
01511 "mov" #size " " #b ", " #temp "   \n\t"\
01512 "pavgb " #temp ", " #a "          \n\t"\
01513 "mov" #size " " #a ", " #b "      \n\t"
01514 
01515 QPEL_BASE(put_       , ff_pw_16, _       , PUT_OP, PUT_OP)
01516 QPEL_BASE(avg_       , ff_pw_16, _       , AVG_MMX2_OP, AVG_3DNOW_OP)
01517 QPEL_BASE(put_no_rnd_, ff_pw_15, _no_rnd_, PUT_OP, PUT_OP)
01518 QPEL_OP(put_       , ff_pw_16, _       , PUT_OP, 3dnow)
01519 QPEL_OP(avg_       , ff_pw_16, _       , AVG_3DNOW_OP, 3dnow)
01520 QPEL_OP(put_no_rnd_, ff_pw_15, _no_rnd_, PUT_OP, 3dnow)
01521 QPEL_OP(put_       , ff_pw_16, _       , PUT_OP, mmx2)
01522 QPEL_OP(avg_       , ff_pw_16, _       , AVG_MMX2_OP, mmx2)
01523 QPEL_OP(put_no_rnd_, ff_pw_15, _no_rnd_, PUT_OP, mmx2)
01524 
01525 /***********************************/
01526 /* bilinear qpel: not compliant to any spec, only for -lavdopts fast */
01527 
01528 #define QPEL_2TAP_XY(OPNAME, SIZE, MMX, XY, HPEL)\
01529 static void OPNAME ## 2tap_qpel ## SIZE ## _mc ## XY ## _ ## MMX(uint8_t *dst, uint8_t *src, int stride){\
01530     OPNAME ## pixels ## SIZE ## HPEL(dst, src, stride, SIZE);\
01531 }
01532 #define QPEL_2TAP_L3(OPNAME, SIZE, MMX, XY, S0, S1, S2)\
01533 static void OPNAME ## 2tap_qpel ## SIZE ## _mc ## XY ## _ ## MMX(uint8_t *dst, uint8_t *src, int stride){\
01534     OPNAME ## 2tap_qpel ## SIZE ## _l3_ ## MMX(dst, src+S0, stride, SIZE, S1, S2);\
01535 }
01536 
01537 #define QPEL_2TAP(OPNAME, SIZE, MMX)\
01538 QPEL_2TAP_XY(OPNAME, SIZE, MMX, 20, _x2_ ## MMX)\
01539 QPEL_2TAP_XY(OPNAME, SIZE, MMX, 02, _y2_ ## MMX)\
01540 QPEL_2TAP_XY(OPNAME, SIZE, MMX, 22, _xy2_mmx)\
01541 static const qpel_mc_func OPNAME ## 2tap_qpel ## SIZE ## _mc00_ ## MMX =\
01542                           OPNAME ## qpel ## SIZE ## _mc00_ ## MMX;\
01543 static const qpel_mc_func OPNAME ## 2tap_qpel ## SIZE ## _mc21_ ## MMX =\
01544                           OPNAME ## 2tap_qpel ## SIZE ## _mc20_ ## MMX;\
01545 static const qpel_mc_func OPNAME ## 2tap_qpel ## SIZE ## _mc12_ ## MMX =\
01546                           OPNAME ## 2tap_qpel ## SIZE ## _mc02_ ## MMX;\
01547 static void OPNAME ## 2tap_qpel ## SIZE ## _mc32_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\
01548     OPNAME ## pixels ## SIZE ## _y2_ ## MMX(dst, src+1, stride, SIZE);\
01549 }\
01550 static void OPNAME ## 2tap_qpel ## SIZE ## _mc23_ ## MMX(uint8_t *dst, uint8_t *src, int stride){\
01551     OPNAME ## pixels ## SIZE ## _x2_ ## MMX(dst, src+stride, stride, SIZE);\
01552 }\
01553 QPEL_2TAP_L3(OPNAME, SIZE, MMX, 10, 0,         1,       0)\
01554 QPEL_2TAP_L3(OPNAME, SIZE, MMX, 30, 1,        -1,       0)\
01555 QPEL_2TAP_L3(OPNAME, SIZE, MMX, 01, 0,         stride,  0)\
01556 QPEL_2TAP_L3(OPNAME, SIZE, MMX, 03, stride,   -stride,  0)\
01557 QPEL_2TAP_L3(OPNAME, SIZE, MMX, 11, 0,         stride,  1)\
01558 QPEL_2TAP_L3(OPNAME, SIZE, MMX, 31, 1,         stride, -1)\
01559 QPEL_2TAP_L3(OPNAME, SIZE, MMX, 13, stride,   -stride,  1)\
01560 QPEL_2TAP_L3(OPNAME, SIZE, MMX, 33, stride+1, -stride, -1)\
01561 
01562 QPEL_2TAP(put_, 16, mmx2)
01563 QPEL_2TAP(avg_, 16, mmx2)
01564 QPEL_2TAP(put_,  8, mmx2)
01565 QPEL_2TAP(avg_,  8, mmx2)
01566 QPEL_2TAP(put_, 16, 3dnow)
01567 QPEL_2TAP(avg_, 16, 3dnow)
01568 QPEL_2TAP(put_,  8, 3dnow)
01569 QPEL_2TAP(avg_,  8, 3dnow)
01570 
01571 
01572 #if 0
01573 static void just_return(void) { return; }
01574 #endif
01575 
01576 #if HAVE_YASM
01577 typedef void emu_edge_core_func (uint8_t *buf, const uint8_t *src,
01578                                  x86_reg linesize, x86_reg start_y,
01579                                  x86_reg end_y, x86_reg block_h,
01580                                  x86_reg start_x, x86_reg end_x,
01581                                  x86_reg block_w);
01582 extern emu_edge_core_func ff_emu_edge_core_mmx;
01583 extern emu_edge_core_func ff_emu_edge_core_sse;
01584 
01585 static av_always_inline
01586 void emulated_edge_mc(uint8_t *buf, const uint8_t *src, int linesize,
01587                       int block_w, int block_h,
01588                       int src_x, int src_y, int w, int h,
01589                       emu_edge_core_func *core_fn)
01590 {
01591     int start_y, start_x, end_y, end_x, src_y_add=0;
01592 
01593     if(src_y>= h){
01594         src_y_add = h-1-src_y;
01595         src_y=h-1;
01596     }else if(src_y<=-block_h){
01597         src_y_add = 1-block_h-src_y;
01598         src_y=1-block_h;
01599     }
01600     if(src_x>= w){
01601         src+= (w-1-src_x);
01602         src_x=w-1;
01603     }else if(src_x<=-block_w){
01604         src+= (1-block_w-src_x);
01605         src_x=1-block_w;
01606     }
01607 
01608     start_y= FFMAX(0, -src_y);
01609     start_x= FFMAX(0, -src_x);
01610     end_y= FFMIN(block_h, h-src_y);
01611     end_x= FFMIN(block_w, w-src_x);
01612     assert(start_x < end_x && block_w > 0);
01613     assert(start_y < end_y && block_h > 0);
01614 
01615     // fill in the to-be-copied part plus all above/below
01616     src += (src_y_add+start_y)*linesize + start_x;
01617     buf += start_x;
01618     core_fn(buf, src, linesize, start_y, end_y, block_h, start_x, end_x, block_w);
01619 }
01620 
01621 #if ARCH_X86_32
01622 static av_noinline
01623 void emulated_edge_mc_mmx(uint8_t *buf, const uint8_t *src, int linesize,
01624                           int block_w, int block_h,
01625                           int src_x, int src_y, int w, int h)
01626 {
01627     emulated_edge_mc(buf, src, linesize, block_w, block_h, src_x, src_y,
01628                      w, h, &ff_emu_edge_core_mmx);
01629 }
01630 #endif
01631 static av_noinline
01632 void emulated_edge_mc_sse(uint8_t *buf, const uint8_t *src, int linesize,
01633                           int block_w, int block_h,
01634                           int src_x, int src_y, int w, int h)
01635 {
01636     emulated_edge_mc(buf, src, linesize, block_w, block_h, src_x, src_y,
01637                      w, h, &ff_emu_edge_core_sse);
01638 }
01639 #endif /* HAVE_YASM */
01640 
01641 typedef void emulated_edge_mc_func (uint8_t *dst, const uint8_t *src,
01642                                     int linesize, int block_w, int block_h,
01643                                     int src_x, int src_y, int w, int h);
01644 
01645 static av_always_inline
01646 void gmc(uint8_t *dst, uint8_t *src, int stride, int h, int ox, int oy,
01647          int dxx, int dxy, int dyx, int dyy, int shift, int r, int width, int height,
01648          emulated_edge_mc_func *emu_edge_fn)
01649 {
01650     const int w = 8;
01651     const int ix = ox>>(16+shift);
01652     const int iy = oy>>(16+shift);
01653     const int oxs = ox>>4;
01654     const int oys = oy>>4;
01655     const int dxxs = dxx>>4;
01656     const int dxys = dxy>>4;
01657     const int dyxs = dyx>>4;
01658     const int dyys = dyy>>4;
01659     const uint16_t r4[4] = {r,r,r,r};
01660     const uint16_t dxy4[4] = {dxys,dxys,dxys,dxys};
01661     const uint16_t dyy4[4] = {dyys,dyys,dyys,dyys};
01662     const uint64_t shift2 = 2*shift;
01663     uint8_t edge_buf[(h+1)*stride];
01664     int x, y;
01665 
01666     const int dxw = (dxx-(1<<(16+shift)))*(w-1);
01667     const int dyh = (dyy-(1<<(16+shift)))*(h-1);
01668     const int dxh = dxy*(h-1);
01669     const int dyw = dyx*(w-1);
01670     if( // non-constant fullpel offset (3% of blocks)
01671         ((ox^(ox+dxw)) | (ox^(ox+dxh)) | (ox^(ox+dxw+dxh)) |
01672          (oy^(oy+dyw)) | (oy^(oy+dyh)) | (oy^(oy+dyw+dyh))) >> (16+shift)
01673         // uses more than 16 bits of subpel mv (only at huge resolution)
01674         || (dxx|dxy|dyx|dyy)&15 )
01675     {
01676         //FIXME could still use mmx for some of the rows
01677         ff_gmc_c(dst, src, stride, h, ox, oy, dxx, dxy, dyx, dyy, shift, r, width, height);
01678         return;
01679     }
01680 
01681     src += ix + iy*stride;
01682     if( (unsigned)ix >= width-w ||
01683         (unsigned)iy >= height-h )
01684     {
01685         emu_edge_fn(edge_buf, src, stride, w+1, h+1, ix, iy, width, height);
01686         src = edge_buf;
01687     }
01688 
01689     __asm__ volatile(
01690         "movd         %0, %%mm6 \n\t"
01691         "pxor      %%mm7, %%mm7 \n\t"
01692         "punpcklwd %%mm6, %%mm6 \n\t"
01693         "punpcklwd %%mm6, %%mm6 \n\t"
01694         :: "r"(1<<shift)
01695     );
01696 
01697     for(x=0; x<w; x+=4){
01698         uint16_t dx4[4] = { oxs - dxys + dxxs*(x+0),
01699                             oxs - dxys + dxxs*(x+1),
01700                             oxs - dxys + dxxs*(x+2),
01701                             oxs - dxys + dxxs*(x+3) };
01702         uint16_t dy4[4] = { oys - dyys + dyxs*(x+0),
01703                             oys - dyys + dyxs*(x+1),
01704                             oys - dyys + dyxs*(x+2),
01705                             oys - dyys + dyxs*(x+3) };
01706 
01707         for(y=0; y<h; y++){
01708             __asm__ volatile(
01709                 "movq   %0,  %%mm4 \n\t"
01710                 "movq   %1,  %%mm5 \n\t"
01711                 "paddw  %2,  %%mm4 \n\t"
01712                 "paddw  %3,  %%mm5 \n\t"
01713                 "movq   %%mm4, %0  \n\t"
01714                 "movq   %%mm5, %1  \n\t"
01715                 "psrlw  $12, %%mm4 \n\t"
01716                 "psrlw  $12, %%mm5 \n\t"
01717                 : "+m"(*dx4), "+m"(*dy4)
01718                 : "m"(*dxy4), "m"(*dyy4)
01719             );
01720 
01721             __asm__ volatile(
01722                 "movq   %%mm6, %%mm2 \n\t"
01723                 "movq   %%mm6, %%mm1 \n\t"
01724                 "psubw  %%mm4, %%mm2 \n\t"
01725                 "psubw  %%mm5, %%mm1 \n\t"
01726                 "movq   %%mm2, %%mm0 \n\t"
01727                 "movq   %%mm4, %%mm3 \n\t"
01728                 "pmullw %%mm1, %%mm0 \n\t" // (s-dx)*(s-dy)
01729                 "pmullw %%mm5, %%mm3 \n\t" // dx*dy
01730                 "pmullw %%mm5, %%mm2 \n\t" // (s-dx)*dy
01731                 "pmullw %%mm4, %%mm1 \n\t" // dx*(s-dy)
01732 
01733                 "movd   %4,    %%mm5 \n\t"
01734                 "movd   %3,    %%mm4 \n\t"
01735                 "punpcklbw %%mm7, %%mm5 \n\t"
01736                 "punpcklbw %%mm7, %%mm4 \n\t"
01737                 "pmullw %%mm5, %%mm3 \n\t" // src[1,1] * dx*dy
01738                 "pmullw %%mm4, %%mm2 \n\t" // src[0,1] * (s-dx)*dy
01739 
01740                 "movd   %2,    %%mm5 \n\t"
01741                 "movd   %1,    %%mm4 \n\t"
01742                 "punpcklbw %%mm7, %%mm5 \n\t"
01743                 "punpcklbw %%mm7, %%mm4 \n\t"
01744                 "pmullw %%mm5, %%mm1 \n\t" // src[1,0] * dx*(s-dy)
01745                 "pmullw %%mm4, %%mm0 \n\t" // src[0,0] * (s-dx)*(s-dy)
01746                 "paddw  %5,    %%mm1 \n\t"
01747                 "paddw  %%mm3, %%mm2 \n\t"
01748                 "paddw  %%mm1, %%mm0 \n\t"
01749                 "paddw  %%mm2, %%mm0 \n\t"
01750 
01751                 "psrlw    %6,    %%mm0 \n\t"
01752                 "packuswb %%mm0, %%mm0 \n\t"
01753                 "movd     %%mm0, %0    \n\t"
01754 
01755                 : "=m"(dst[x+y*stride])
01756                 : "m"(src[0]), "m"(src[1]),
01757                   "m"(src[stride]), "m"(src[stride+1]),
01758                   "m"(*r4), "m"(shift2)
01759             );
01760             src += stride;
01761         }
01762         src += 4-h*stride;
01763     }
01764 }
01765 
01766 #if HAVE_YASM
01767 #if ARCH_X86_32
01768 static void gmc_mmx(uint8_t *dst, uint8_t *src, int stride, int h, int ox, int oy,
01769                     int dxx, int dxy, int dyx, int dyy, int shift, int r, int width, int height)
01770 {
01771     gmc(dst, src, stride, h, ox, oy, dxx, dxy, dyx, dyy, shift, r,
01772         width, height, &emulated_edge_mc_mmx);
01773 }
01774 #endif
01775 static void gmc_sse(uint8_t *dst, uint8_t *src, int stride, int h, int ox, int oy,
01776                     int dxx, int dxy, int dyx, int dyy, int shift, int r, int width, int height)
01777 {
01778     gmc(dst, src, stride, h, ox, oy, dxx, dxy, dyx, dyy, shift, r,
01779         width, height, &emulated_edge_mc_sse);
01780 }
01781 #else
01782 static void gmc_mmx(uint8_t *dst, uint8_t *src, int stride, int h, int ox, int oy,
01783                     int dxx, int dxy, int dyx, int dyy, int shift, int r, int width, int height)
01784 {
01785     gmc(dst, src, stride, h, ox, oy, dxx, dxy, dyx, dyy, shift, r,
01786         width, height, &ff_emulated_edge_mc);
01787 }
01788 #endif
01789 
01790 #define PREFETCH(name, op) \
01791 static void name(void *mem, int stride, int h){\
01792     const uint8_t *p= mem;\
01793     do{\
01794         __asm__ volatile(#op" %0" :: "m"(*p));\
01795         p+= stride;\
01796     }while(--h);\
01797 }
01798 PREFETCH(prefetch_mmx2,  prefetcht0)
01799 PREFETCH(prefetch_3dnow, prefetch)
01800 #undef PREFETCH
01801 
01802 #include "h264_qpel_mmx.c"
01803 
01804 void ff_put_h264_chroma_mc8_mmx_rnd   (uint8_t *dst, uint8_t *src,
01805                                        int stride, int h, int x, int y);
01806 void ff_put_rv40_chroma_mc8_mmx       (uint8_t *dst, uint8_t *src,
01807                                        int stride, int h, int x, int y);
01808 void ff_avg_h264_chroma_mc8_mmx2_rnd  (uint8_t *dst, uint8_t *src,
01809                                        int stride, int h, int x, int y);
01810 void ff_avg_rv40_chroma_mc8_mmx2      (uint8_t *dst, uint8_t *src,
01811                                        int stride, int h, int x, int y);
01812 void ff_avg_h264_chroma_mc8_3dnow_rnd (uint8_t *dst, uint8_t *src,
01813                                        int stride, int h, int x, int y);
01814 void ff_avg_rv40_chroma_mc8_3dnow     (uint8_t *dst, uint8_t *src,
01815                                        int stride, int h, int x, int y);
01816 
01817 void ff_put_h264_chroma_mc4_mmx       (uint8_t *dst, uint8_t *src,
01818                                        int stride, int h, int x, int y);
01819 void ff_put_rv40_chroma_mc4_mmx       (uint8_t *dst, uint8_t *src,
01820                                        int stride, int h, int x, int y);
01821 void ff_avg_h264_chroma_mc4_mmx2      (uint8_t *dst, uint8_t *src,
01822                                        int stride, int h, int x, int y);
01823 void ff_avg_rv40_chroma_mc4_mmx2      (uint8_t *dst, uint8_t *src,
01824                                        int stride, int h, int x, int y);
01825 void ff_avg_h264_chroma_mc4_3dnow     (uint8_t *dst, uint8_t *src,
01826                                        int stride, int h, int x, int y);
01827 void ff_avg_rv40_chroma_mc4_3dnow     (uint8_t *dst, uint8_t *src,
01828                                        int stride, int h, int x, int y);
01829 
01830 void ff_put_h264_chroma_mc2_mmx2      (uint8_t *dst, uint8_t *src,
01831                                        int stride, int h, int x, int y);
01832 void ff_avg_h264_chroma_mc2_mmx2      (uint8_t *dst, uint8_t *src,
01833                                        int stride, int h, int x, int y);
01834 
01835 void ff_put_h264_chroma_mc8_ssse3_rnd (uint8_t *dst, uint8_t *src,
01836                                        int stride, int h, int x, int y);
01837 void ff_put_h264_chroma_mc4_ssse3     (uint8_t *dst, uint8_t *src,
01838                                        int stride, int h, int x, int y);
01839 
01840 void ff_avg_h264_chroma_mc8_ssse3_rnd (uint8_t *dst, uint8_t *src,
01841                                        int stride, int h, int x, int y);
01842 void ff_avg_h264_chroma_mc4_ssse3     (uint8_t *dst, uint8_t *src,
01843                                        int stride, int h, int x, int y);
01844 
01845 #define CHROMA_MC(OP, NUM, DEPTH, OPT) \
01846 void ff_ ## OP ## _h264_chroma_mc ## NUM ## _ ## DEPTH ## _ ## OPT \
01847                                       (uint8_t *dst, uint8_t *src,\
01848                                        int stride, int h, int x, int y);
01849 
01850 CHROMA_MC(put, 2, 10, mmxext)
01851 CHROMA_MC(avg, 2, 10, mmxext)
01852 CHROMA_MC(put, 4, 10, mmxext)
01853 CHROMA_MC(avg, 4, 10, mmxext)
01854 CHROMA_MC(put, 8, 10, sse2)
01855 CHROMA_MC(avg, 8, 10, sse2)
01856 CHROMA_MC(put, 8, 10, avx)
01857 CHROMA_MC(avg, 8, 10, avx)
01858 
01859 /* CAVS specific */
01860 void ff_put_cavs_qpel8_mc00_mmx2(uint8_t *dst, uint8_t *src, int stride) {
01861     put_pixels8_mmx(dst, src, stride, 8);
01862 }
01863 void ff_avg_cavs_qpel8_mc00_mmx2(uint8_t *dst, uint8_t *src, int stride) {
01864     avg_pixels8_mmx(dst, src, stride, 8);
01865 }
01866 void ff_put_cavs_qpel16_mc00_mmx2(uint8_t *dst, uint8_t *src, int stride) {
01867     put_pixels16_mmx(dst, src, stride, 16);
01868 }
01869 void ff_avg_cavs_qpel16_mc00_mmx2(uint8_t *dst, uint8_t *src, int stride) {
01870     avg_pixels16_mmx(dst, src, stride, 16);
01871 }
01872 
01873 /* VC1 specific */
01874 void ff_put_vc1_mspel_mc00_mmx(uint8_t *dst, const uint8_t *src, int stride, int rnd) {
01875     put_pixels8_mmx(dst, src, stride, 8);
01876 }
01877 void ff_avg_vc1_mspel_mc00_mmx2(uint8_t *dst, const uint8_t *src, int stride, int rnd) {
01878     avg_pixels8_mmx2(dst, src, stride, 8);
01879 }
01880 
01881 /* XXX: those functions should be suppressed ASAP when all IDCTs are
01882    converted */
01883 #if CONFIG_GPL
01884 static void ff_libmpeg2mmx_idct_put(uint8_t *dest, int line_size, DCTELEM *block)
01885 {
01886     ff_mmx_idct (block);
01887     ff_put_pixels_clamped_mmx(block, dest, line_size);
01888 }
01889 static void ff_libmpeg2mmx_idct_add(uint8_t *dest, int line_size, DCTELEM *block)
01890 {
01891     ff_mmx_idct (block);
01892     ff_add_pixels_clamped_mmx(block, dest, line_size);
01893 }
01894 static void ff_libmpeg2mmx2_idct_put(uint8_t *dest, int line_size, DCTELEM *block)
01895 {
01896     ff_mmxext_idct (block);
01897     ff_put_pixels_clamped_mmx(block, dest, line_size);
01898 }
01899 static void ff_libmpeg2mmx2_idct_add(uint8_t *dest, int line_size, DCTELEM *block)
01900 {
01901     ff_mmxext_idct (block);
01902     ff_add_pixels_clamped_mmx(block, dest, line_size);
01903 }
01904 #endif
01905 static void ff_idct_xvid_mmx_put(uint8_t *dest, int line_size, DCTELEM *block)
01906 {
01907     ff_idct_xvid_mmx (block);
01908     ff_put_pixels_clamped_mmx(block, dest, line_size);
01909 }
01910 static void ff_idct_xvid_mmx_add(uint8_t *dest, int line_size, DCTELEM *block)
01911 {
01912     ff_idct_xvid_mmx (block);
01913     ff_add_pixels_clamped_mmx(block, dest, line_size);
01914 }
01915 static void ff_idct_xvid_mmx2_put(uint8_t *dest, int line_size, DCTELEM *block)
01916 {
01917     ff_idct_xvid_mmx2 (block);
01918     ff_put_pixels_clamped_mmx(block, dest, line_size);
01919 }
01920 static void ff_idct_xvid_mmx2_add(uint8_t *dest, int line_size, DCTELEM *block)
01921 {
01922     ff_idct_xvid_mmx2 (block);
01923     ff_add_pixels_clamped_mmx(block, dest, line_size);
01924 }
01925 
01926 static void vorbis_inverse_coupling_3dnow(float *mag, float *ang, int blocksize)
01927 {
01928     int i;
01929     __asm__ volatile("pxor %%mm7, %%mm7":);
01930     for(i=0; i<blocksize; i+=2) {
01931         __asm__ volatile(
01932             "movq    %0,    %%mm0 \n\t"
01933             "movq    %1,    %%mm1 \n\t"
01934             "movq    %%mm0, %%mm2 \n\t"
01935             "movq    %%mm1, %%mm3 \n\t"
01936             "pfcmpge %%mm7, %%mm2 \n\t" // m <= 0.0
01937             "pfcmpge %%mm7, %%mm3 \n\t" // a <= 0.0
01938             "pslld   $31,   %%mm2 \n\t" // keep only the sign bit
01939             "pxor    %%mm2, %%mm1 \n\t"
01940             "movq    %%mm3, %%mm4 \n\t"
01941             "pand    %%mm1, %%mm3 \n\t"
01942             "pandn   %%mm1, %%mm4 \n\t"
01943             "pfadd   %%mm0, %%mm3 \n\t" // a = m + ((a<0) & (a ^ sign(m)))
01944             "pfsub   %%mm4, %%mm0 \n\t" // m = m + ((a>0) & (a ^ sign(m)))
01945             "movq    %%mm3, %1    \n\t"
01946             "movq    %%mm0, %0    \n\t"
01947             :"+m"(mag[i]), "+m"(ang[i])
01948             ::"memory"
01949         );
01950     }
01951     __asm__ volatile("femms");
01952 }
01953 static void vorbis_inverse_coupling_sse(float *mag, float *ang, int blocksize)
01954 {
01955     int i;
01956 
01957     __asm__ volatile(
01958             "movaps  %0,     %%xmm5 \n\t"
01959         ::"m"(ff_pdw_80000000[0])
01960     );
01961     for(i=0; i<blocksize; i+=4) {
01962         __asm__ volatile(
01963             "movaps  %0,     %%xmm0 \n\t"
01964             "movaps  %1,     %%xmm1 \n\t"
01965             "xorps   %%xmm2, %%xmm2 \n\t"
01966             "xorps   %%xmm3, %%xmm3 \n\t"
01967             "cmpleps %%xmm0, %%xmm2 \n\t" // m <= 0.0
01968             "cmpleps %%xmm1, %%xmm3 \n\t" // a <= 0.0
01969             "andps   %%xmm5, %%xmm2 \n\t" // keep only the sign bit
01970             "xorps   %%xmm2, %%xmm1 \n\t"
01971             "movaps  %%xmm3, %%xmm4 \n\t"
01972             "andps   %%xmm1, %%xmm3 \n\t"
01973             "andnps  %%xmm1, %%xmm4 \n\t"
01974             "addps   %%xmm0, %%xmm3 \n\t" // a = m + ((a<0) & (a ^ sign(m)))
01975             "subps   %%xmm4, %%xmm0 \n\t" // m = m + ((a>0) & (a ^ sign(m)))
01976             "movaps  %%xmm3, %1     \n\t"
01977             "movaps  %%xmm0, %0     \n\t"
01978             :"+m"(mag[i]), "+m"(ang[i])
01979             ::"memory"
01980         );
01981     }
01982 }
01983 
01984 #define IF1(x) x
01985 #define IF0(x)
01986 
01987 #define MIX5(mono,stereo)\
01988     __asm__ volatile(\
01989         "movss          0(%2), %%xmm5 \n"\
01990         "movss          8(%2), %%xmm6 \n"\
01991         "movss         24(%2), %%xmm7 \n"\
01992         "shufps    $0, %%xmm5, %%xmm5 \n"\
01993         "shufps    $0, %%xmm6, %%xmm6 \n"\
01994         "shufps    $0, %%xmm7, %%xmm7 \n"\
01995         "1: \n"\
01996         "movaps       (%0,%1), %%xmm0 \n"\
01997         "movaps  0x400(%0,%1), %%xmm1 \n"\
01998         "movaps  0x800(%0,%1), %%xmm2 \n"\
01999         "movaps  0xc00(%0,%1), %%xmm3 \n"\
02000         "movaps 0x1000(%0,%1), %%xmm4 \n"\
02001         "mulps         %%xmm5, %%xmm0 \n"\
02002         "mulps         %%xmm6, %%xmm1 \n"\
02003         "mulps         %%xmm5, %%xmm2 \n"\
02004         "mulps         %%xmm7, %%xmm3 \n"\
02005         "mulps         %%xmm7, %%xmm4 \n"\
02006  stereo("addps         %%xmm1, %%xmm0 \n")\
02007         "addps         %%xmm1, %%xmm2 \n"\
02008         "addps         %%xmm3, %%xmm0 \n"\
02009         "addps         %%xmm4, %%xmm2 \n"\
02010    mono("addps         %%xmm2, %%xmm0 \n")\
02011         "movaps  %%xmm0,      (%0,%1) \n"\
02012  stereo("movaps  %%xmm2, 0x400(%0,%1) \n")\
02013         "add $16, %0 \n"\
02014         "jl 1b \n"\
02015         :"+&r"(i)\
02016         :"r"(samples[0]+len), "r"(matrix)\
02017         :XMM_CLOBBERS("%xmm0", "%xmm1", "%xmm2", "%xmm3", \
02018                       "%xmm4", "%xmm5", "%xmm6", "%xmm7",)\
02019          "memory"\
02020     );
02021 
02022 #define MIX_MISC(stereo)\
02023     __asm__ volatile(\
02024         "1: \n"\
02025         "movaps  (%3,%0), %%xmm0 \n"\
02026  stereo("movaps   %%xmm0, %%xmm1 \n")\
02027         "mulps    %%xmm4, %%xmm0 \n"\
02028  stereo("mulps    %%xmm5, %%xmm1 \n")\
02029         "lea 1024(%3,%0), %1 \n"\
02030         "mov %5, %2 \n"\
02031         "2: \n"\
02032         "movaps   (%1),   %%xmm2 \n"\
02033  stereo("movaps   %%xmm2, %%xmm3 \n")\
02034         "mulps   (%4,%2), %%xmm2 \n"\
02035  stereo("mulps 16(%4,%2), %%xmm3 \n")\
02036         "addps    %%xmm2, %%xmm0 \n"\
02037  stereo("addps    %%xmm3, %%xmm1 \n")\
02038         "add $1024, %1 \n"\
02039         "add $32, %2 \n"\
02040         "jl 2b \n"\
02041         "movaps   %%xmm0,     (%3,%0) \n"\
02042  stereo("movaps   %%xmm1, 1024(%3,%0) \n")\
02043         "add $16, %0 \n"\
02044         "jl 1b \n"\
02045         :"+&r"(i), "=&r"(j), "=&r"(k)\
02046         :"r"(samples[0]+len), "r"(matrix_simd+in_ch), "g"((intptr_t)-32*(in_ch-1))\
02047         :"memory"\
02048     );
02049 
02050 static void ac3_downmix_sse(float (*samples)[256], float (*matrix)[2], int out_ch, int in_ch, int len)
02051 {
02052     int (*matrix_cmp)[2] = (int(*)[2])matrix;
02053     intptr_t i,j,k;
02054 
02055     i = -len*sizeof(float);
02056     if(in_ch == 5 && out_ch == 2 && !(matrix_cmp[0][1]|matrix_cmp[2][0]|matrix_cmp[3][1]|matrix_cmp[4][0]|(matrix_cmp[1][0]^matrix_cmp[1][1])|(matrix_cmp[0][0]^matrix_cmp[2][1]))) {
02057         MIX5(IF0,IF1);
02058     } else if(in_ch == 5 && out_ch == 1 && matrix_cmp[0][0]==matrix_cmp[2][0] && matrix_cmp[3][0]==matrix_cmp[4][0]) {
02059         MIX5(IF1,IF0);
02060     } else {
02061         DECLARE_ALIGNED(16, float, matrix_simd)[AC3_MAX_CHANNELS][2][4];
02062         j = 2*in_ch*sizeof(float);
02063         __asm__ volatile(
02064             "1: \n"
02065             "sub $8, %0 \n"
02066             "movss     (%2,%0), %%xmm4 \n"
02067             "movss    4(%2,%0), %%xmm5 \n"
02068             "shufps $0, %%xmm4, %%xmm4 \n"
02069             "shufps $0, %%xmm5, %%xmm5 \n"
02070             "movaps %%xmm4,   (%1,%0,4) \n"
02071             "movaps %%xmm5, 16(%1,%0,4) \n"
02072             "jg 1b \n"
02073             :"+&r"(j)
02074             :"r"(matrix_simd), "r"(matrix)
02075             :"memory"
02076         );
02077         if(out_ch == 2) {
02078             MIX_MISC(IF1);
02079         } else {
02080             MIX_MISC(IF0);
02081         }
02082     }
02083 }
02084 
02085 static void vector_fmul_3dnow(float *dst, const float *src0, const float *src1, int len){
02086     x86_reg i = (len-4)*4;
02087     __asm__ volatile(
02088         "1: \n\t"
02089         "movq    (%2,%0), %%mm0 \n\t"
02090         "movq   8(%2,%0), %%mm1 \n\t"
02091         "pfmul   (%3,%0), %%mm0 \n\t"
02092         "pfmul  8(%3,%0), %%mm1 \n\t"
02093         "movq   %%mm0,  (%1,%0) \n\t"
02094         "movq   %%mm1, 8(%1,%0) \n\t"
02095         "sub  $16, %0 \n\t"
02096         "jge 1b \n\t"
02097         "femms  \n\t"
02098         :"+r"(i)
02099         :"r"(dst), "r"(src0), "r"(src1)
02100         :"memory"
02101     );
02102 }
02103 static void vector_fmul_sse(float *dst, const float *src0, const float *src1, int len){
02104     x86_reg i = (len-8)*4;
02105     __asm__ volatile(
02106         "1: \n\t"
02107         "movaps    (%2,%0), %%xmm0 \n\t"
02108         "movaps  16(%2,%0), %%xmm1 \n\t"
02109         "mulps     (%3,%0), %%xmm0 \n\t"
02110         "mulps   16(%3,%0), %%xmm1 \n\t"
02111         "movaps  %%xmm0,   (%1,%0) \n\t"
02112         "movaps  %%xmm1, 16(%1,%0) \n\t"
02113         "sub  $32, %0 \n\t"
02114         "jge 1b \n\t"
02115         :"+r"(i)
02116         :"r"(dst), "r"(src0), "r"(src1)
02117         :"memory"
02118     );
02119 }
02120 
02121 static void vector_fmul_reverse_3dnow2(float *dst, const float *src0, const float *src1, int len){
02122     x86_reg i = len*4-16;
02123     __asm__ volatile(
02124         "1: \n\t"
02125         "pswapd   8(%1), %%mm0 \n\t"
02126         "pswapd    (%1), %%mm1 \n\t"
02127         "pfmul  (%3,%0), %%mm0 \n\t"
02128         "pfmul 8(%3,%0), %%mm1 \n\t"
02129         "movq  %%mm0,  (%2,%0) \n\t"
02130         "movq  %%mm1, 8(%2,%0) \n\t"
02131         "add   $16, %1 \n\t"
02132         "sub   $16, %0 \n\t"
02133         "jge   1b \n\t"
02134         :"+r"(i), "+r"(src1)
02135         :"r"(dst), "r"(src0)
02136     );
02137     __asm__ volatile("femms");
02138 }
02139 static void vector_fmul_reverse_sse(float *dst, const float *src0, const float *src1, int len){
02140     x86_reg i = len*4-32;
02141     __asm__ volatile(
02142         "1: \n\t"
02143         "movaps        16(%1), %%xmm0 \n\t"
02144         "movaps          (%1), %%xmm1 \n\t"
02145         "shufps $0x1b, %%xmm0, %%xmm0 \n\t"
02146         "shufps $0x1b, %%xmm1, %%xmm1 \n\t"
02147         "mulps        (%3,%0), %%xmm0 \n\t"
02148         "mulps      16(%3,%0), %%xmm1 \n\t"
02149         "movaps     %%xmm0,   (%2,%0) \n\t"
02150         "movaps     %%xmm1, 16(%2,%0) \n\t"
02151         "add    $32, %1 \n\t"
02152         "sub    $32, %0 \n\t"
02153         "jge    1b \n\t"
02154         :"+r"(i), "+r"(src1)
02155         :"r"(dst), "r"(src0)
02156     );
02157 }
02158 
02159 static void vector_fmul_add_3dnow(float *dst, const float *src0, const float *src1,
02160                                   const float *src2, int len){
02161     x86_reg i = (len-4)*4;
02162     __asm__ volatile(
02163         "1: \n\t"
02164         "movq    (%2,%0), %%mm0 \n\t"
02165         "movq   8(%2,%0), %%mm1 \n\t"
02166         "pfmul   (%3,%0), %%mm0 \n\t"
02167         "pfmul  8(%3,%0), %%mm1 \n\t"
02168         "pfadd   (%4,%0), %%mm0 \n\t"
02169         "pfadd  8(%4,%0), %%mm1 \n\t"
02170         "movq  %%mm0,   (%1,%0) \n\t"
02171         "movq  %%mm1,  8(%1,%0) \n\t"
02172         "sub  $16, %0 \n\t"
02173         "jge  1b \n\t"
02174         :"+r"(i)
02175         :"r"(dst), "r"(src0), "r"(src1), "r"(src2)
02176         :"memory"
02177     );
02178     __asm__ volatile("femms");
02179 }
02180 static void vector_fmul_add_sse(float *dst, const float *src0, const float *src1,
02181                                 const float *src2, int len){
02182     x86_reg i = (len-8)*4;
02183     __asm__ volatile(
02184         "1: \n\t"
02185         "movaps   (%2,%0), %%xmm0 \n\t"
02186         "movaps 16(%2,%0), %%xmm1 \n\t"
02187         "mulps    (%3,%0), %%xmm0 \n\t"
02188         "mulps  16(%3,%0), %%xmm1 \n\t"
02189         "addps    (%4,%0), %%xmm0 \n\t"
02190         "addps  16(%4,%0), %%xmm1 \n\t"
02191         "movaps %%xmm0,   (%1,%0) \n\t"
02192         "movaps %%xmm1, 16(%1,%0) \n\t"
02193         "sub  $32, %0 \n\t"
02194         "jge  1b \n\t"
02195         :"+r"(i)
02196         :"r"(dst), "r"(src0), "r"(src1), "r"(src2)
02197         :"memory"
02198     );
02199 }
02200 
02201 #if HAVE_6REGS
02202 static void vector_fmul_window_3dnow2(float *dst, const float *src0, const float *src1,
02203                                       const float *win, int len){
02204     x86_reg i = -len*4;
02205     x86_reg j = len*4-8;
02206     __asm__ volatile(
02207         "1: \n"
02208         "pswapd  (%5,%1), %%mm1 \n"
02209         "movq    (%5,%0), %%mm0 \n"
02210         "pswapd  (%4,%1), %%mm5 \n"
02211         "movq    (%3,%0), %%mm4 \n"
02212         "movq      %%mm0, %%mm2 \n"
02213         "movq      %%mm1, %%mm3 \n"
02214         "pfmul     %%mm4, %%mm2 \n" // src0[len+i]*win[len+i]
02215         "pfmul     %%mm5, %%mm3 \n" // src1[    j]*win[len+j]
02216         "pfmul     %%mm4, %%mm1 \n" // src0[len+i]*win[len+j]
02217         "pfmul     %%mm5, %%mm0 \n" // src1[    j]*win[len+i]
02218         "pfadd     %%mm3, %%mm2 \n"
02219         "pfsub     %%mm0, %%mm1 \n"
02220         "pswapd    %%mm2, %%mm2 \n"
02221         "movq      %%mm1, (%2,%0) \n"
02222         "movq      %%mm2, (%2,%1) \n"
02223         "sub $8, %1 \n"
02224         "add $8, %0 \n"
02225         "jl 1b \n"
02226         "femms \n"
02227         :"+r"(i), "+r"(j)
02228         :"r"(dst+len), "r"(src0+len), "r"(src1), "r"(win+len)
02229     );
02230 }
02231 
02232 static void vector_fmul_window_sse(float *dst, const float *src0, const float *src1,
02233                                    const float *win, int len){
02234     x86_reg i = -len*4;
02235     x86_reg j = len*4-16;
02236     __asm__ volatile(
02237         "1: \n"
02238         "movaps       (%5,%1), %%xmm1 \n"
02239         "movaps       (%5,%0), %%xmm0 \n"
02240         "movaps       (%4,%1), %%xmm5 \n"
02241         "movaps       (%3,%0), %%xmm4 \n"
02242         "shufps $0x1b, %%xmm1, %%xmm1 \n"
02243         "shufps $0x1b, %%xmm5, %%xmm5 \n"
02244         "movaps        %%xmm0, %%xmm2 \n"
02245         "movaps        %%xmm1, %%xmm3 \n"
02246         "mulps         %%xmm4, %%xmm2 \n" // src0[len+i]*win[len+i]
02247         "mulps         %%xmm5, %%xmm3 \n" // src1[    j]*win[len+j]
02248         "mulps         %%xmm4, %%xmm1 \n" // src0[len+i]*win[len+j]
02249         "mulps         %%xmm5, %%xmm0 \n" // src1[    j]*win[len+i]
02250         "addps         %%xmm3, %%xmm2 \n"
02251         "subps         %%xmm0, %%xmm1 \n"
02252         "shufps $0x1b, %%xmm2, %%xmm2 \n"
02253         "movaps        %%xmm1, (%2,%0) \n"
02254         "movaps        %%xmm2, (%2,%1) \n"
02255         "sub $16, %1 \n"
02256         "add $16, %0 \n"
02257         "jl 1b \n"
02258         :"+r"(i), "+r"(j)
02259         :"r"(dst+len), "r"(src0+len), "r"(src1), "r"(win+len)
02260     );
02261 }
02262 #endif /* HAVE_6REGS */
02263 
02264 static void vector_clipf_sse(float *dst, const float *src, float min, float max,
02265                              int len)
02266 {
02267     x86_reg i = (len-16)*4;
02268     __asm__ volatile(
02269         "movss  %3, %%xmm4 \n"
02270         "movss  %4, %%xmm5 \n"
02271         "shufps $0, %%xmm4, %%xmm4 \n"
02272         "shufps $0, %%xmm5, %%xmm5 \n"
02273         "1: \n\t"
02274         "movaps    (%2,%0), %%xmm0 \n\t" // 3/1 on intel
02275         "movaps  16(%2,%0), %%xmm1 \n\t"
02276         "movaps  32(%2,%0), %%xmm2 \n\t"
02277         "movaps  48(%2,%0), %%xmm3 \n\t"
02278         "maxps      %%xmm4, %%xmm0 \n\t"
02279         "maxps      %%xmm4, %%xmm1 \n\t"
02280         "maxps      %%xmm4, %%xmm2 \n\t"
02281         "maxps      %%xmm4, %%xmm3 \n\t"
02282         "minps      %%xmm5, %%xmm0 \n\t"
02283         "minps      %%xmm5, %%xmm1 \n\t"
02284         "minps      %%xmm5, %%xmm2 \n\t"
02285         "minps      %%xmm5, %%xmm3 \n\t"
02286         "movaps  %%xmm0,   (%1,%0) \n\t"
02287         "movaps  %%xmm1, 16(%1,%0) \n\t"
02288         "movaps  %%xmm2, 32(%1,%0) \n\t"
02289         "movaps  %%xmm3, 48(%1,%0) \n\t"
02290         "sub  $64, %0 \n\t"
02291         "jge 1b \n\t"
02292         :"+&r"(i)
02293         :"r"(dst), "r"(src), "m"(min), "m"(max)
02294         :"memory"
02295     );
02296 }
02297 
02298 void ff_vp3_idct_mmx(int16_t *input_data);
02299 void ff_vp3_idct_put_mmx(uint8_t *dest, int line_size, DCTELEM *block);
02300 void ff_vp3_idct_add_mmx(uint8_t *dest, int line_size, DCTELEM *block);
02301 
02302 void ff_vp3_idct_dc_add_mmx2(uint8_t *dest, int line_size, const DCTELEM *block);
02303 
02304 void ff_vp3_v_loop_filter_mmx2(uint8_t *src, int stride, int *bounding_values);
02305 void ff_vp3_h_loop_filter_mmx2(uint8_t *src, int stride, int *bounding_values);
02306 
02307 void ff_vp3_idct_sse2(int16_t *input_data);
02308 void ff_vp3_idct_put_sse2(uint8_t *dest, int line_size, DCTELEM *block);
02309 void ff_vp3_idct_add_sse2(uint8_t *dest, int line_size, DCTELEM *block);
02310 
02311 int32_t ff_scalarproduct_int16_mmx2(const int16_t *v1, const int16_t *v2, int order, int shift);
02312 int32_t ff_scalarproduct_int16_sse2(const int16_t *v1, const int16_t *v2, int order, int shift);
02313 int32_t ff_scalarproduct_and_madd_int16_mmx2(int16_t *v1, const int16_t *v2, const int16_t *v3, int order, int mul);
02314 int32_t ff_scalarproduct_and_madd_int16_sse2(int16_t *v1, const int16_t *v2, const int16_t *v3, int order, int mul);
02315 int32_t ff_scalarproduct_and_madd_int16_ssse3(int16_t *v1, const int16_t *v2, const int16_t *v3, int order, int mul);
02316 
02317 void ff_apply_window_int16_mmxext    (int16_t *output, const int16_t *input,
02318                                       const int16_t *window, unsigned int len);
02319 void ff_apply_window_int16_mmxext_ba (int16_t *output, const int16_t *input,
02320                                       const int16_t *window, unsigned int len);
02321 void ff_apply_window_int16_sse2      (int16_t *output, const int16_t *input,
02322                                       const int16_t *window, unsigned int len);
02323 void ff_apply_window_int16_sse2_ba   (int16_t *output, const int16_t *input,
02324                                       const int16_t *window, unsigned int len);
02325 void ff_apply_window_int16_ssse3     (int16_t *output, const int16_t *input,
02326                                       const int16_t *window, unsigned int len);
02327 void ff_apply_window_int16_ssse3_atom(int16_t *output, const int16_t *input,
02328                                       const int16_t *window, unsigned int len);
02329 
02330 void ff_add_hfyu_median_prediction_mmx2(uint8_t *dst, const uint8_t *top, const uint8_t *diff, int w, int *left, int *left_top);
02331 int  ff_add_hfyu_left_prediction_ssse3(uint8_t *dst, const uint8_t *src, int w, int left);
02332 int  ff_add_hfyu_left_prediction_sse4(uint8_t *dst, const uint8_t *src, int w, int left);
02333 
02334 float ff_scalarproduct_float_sse(const float *v1, const float *v2, int order);
02335 
02336 void dsputil_init_mmx(DSPContext* c, AVCodecContext *avctx)
02337 {
02338     int mm_flags = av_get_cpu_flags();
02339     const int high_bit_depth = avctx->codec_id == CODEC_ID_H264 && avctx->bits_per_raw_sample > 8;
02340     const int bit_depth = avctx->bits_per_raw_sample;
02341 
02342     if (avctx->dsp_mask) {
02343         if (avctx->dsp_mask & AV_CPU_FLAG_FORCE)
02344             mm_flags |= (avctx->dsp_mask & 0xffff);
02345         else
02346             mm_flags &= ~(avctx->dsp_mask & 0xffff);
02347     }
02348 
02349 #if 0
02350     av_log(avctx, AV_LOG_INFO, "libavcodec: CPU flags:");
02351     if (mm_flags & AV_CPU_FLAG_MMX)
02352         av_log(avctx, AV_LOG_INFO, " mmx");
02353     if (mm_flags & AV_CPU_FLAG_MMX2)
02354         av_log(avctx, AV_LOG_INFO, " mmx2");
02355     if (mm_flags & AV_CPU_FLAG_3DNOW)
02356         av_log(avctx, AV_LOG_INFO, " 3dnow");
02357     if (mm_flags & AV_CPU_FLAG_SSE)
02358         av_log(avctx, AV_LOG_INFO, " sse");
02359     if (mm_flags & AV_CPU_FLAG_SSE2)
02360         av_log(avctx, AV_LOG_INFO, " sse2");
02361     av_log(avctx, AV_LOG_INFO, "\n");
02362 #endif
02363 
02364     if (mm_flags & AV_CPU_FLAG_MMX) {
02365         const int idct_algo= avctx->idct_algo;
02366 
02367         if(avctx->lowres==0){
02368             if(idct_algo==FF_IDCT_AUTO || idct_algo==FF_IDCT_SIMPLEMMX){
02369                 c->idct_put= ff_simple_idct_put_mmx;
02370                 c->idct_add= ff_simple_idct_add_mmx;
02371                 c->idct    = ff_simple_idct_mmx;
02372                 c->idct_permutation_type= FF_SIMPLE_IDCT_PERM;
02373 #if CONFIG_GPL
02374             }else if(idct_algo==FF_IDCT_LIBMPEG2MMX){
02375                 if(mm_flags & AV_CPU_FLAG_MMX2){
02376                     c->idct_put= ff_libmpeg2mmx2_idct_put;
02377                     c->idct_add= ff_libmpeg2mmx2_idct_add;
02378                     c->idct    = ff_mmxext_idct;
02379                 }else{
02380                     c->idct_put= ff_libmpeg2mmx_idct_put;
02381                     c->idct_add= ff_libmpeg2mmx_idct_add;
02382                     c->idct    = ff_mmx_idct;
02383                 }
02384                 c->idct_permutation_type= FF_LIBMPEG2_IDCT_PERM;
02385 #endif
02386             }else if((CONFIG_VP3_DECODER || CONFIG_VP5_DECODER || CONFIG_VP6_DECODER) &&
02387                      idct_algo==FF_IDCT_VP3 && HAVE_YASM){
02388                 if(mm_flags & AV_CPU_FLAG_SSE2){
02389                     c->idct_put= ff_vp3_idct_put_sse2;
02390                     c->idct_add= ff_vp3_idct_add_sse2;
02391                     c->idct    = ff_vp3_idct_sse2;
02392                     c->idct_permutation_type= FF_TRANSPOSE_IDCT_PERM;
02393                 }else{
02394                     c->idct_put= ff_vp3_idct_put_mmx;
02395                     c->idct_add= ff_vp3_idct_add_mmx;
02396                     c->idct    = ff_vp3_idct_mmx;
02397                     c->idct_permutation_type= FF_PARTTRANS_IDCT_PERM;
02398                 }
02399             }else if(idct_algo==FF_IDCT_CAVS){
02400                     c->idct_permutation_type= FF_TRANSPOSE_IDCT_PERM;
02401             }else if(idct_algo==FF_IDCT_XVIDMMX){
02402                 if(mm_flags & AV_CPU_FLAG_SSE2){
02403                     c->idct_put= ff_idct_xvid_sse2_put;
02404                     c->idct_add= ff_idct_xvid_sse2_add;
02405                     c->idct    = ff_idct_xvid_sse2;
02406                     c->idct_permutation_type= FF_SSE2_IDCT_PERM;
02407                 }else if(mm_flags & AV_CPU_FLAG_MMX2){
02408                     c->idct_put= ff_idct_xvid_mmx2_put;
02409                     c->idct_add= ff_idct_xvid_mmx2_add;
02410                     c->idct    = ff_idct_xvid_mmx2;
02411                 }else{
02412                     c->idct_put= ff_idct_xvid_mmx_put;
02413                     c->idct_add= ff_idct_xvid_mmx_add;
02414                     c->idct    = ff_idct_xvid_mmx;
02415                 }
02416             }
02417         }
02418 
02419         c->put_pixels_clamped = ff_put_pixels_clamped_mmx;
02420         c->put_signed_pixels_clamped = ff_put_signed_pixels_clamped_mmx;
02421         c->add_pixels_clamped = ff_add_pixels_clamped_mmx;
02422         if (!high_bit_depth) {
02423         c->clear_block  = clear_block_mmx;
02424         c->clear_blocks = clear_blocks_mmx;
02425         if ((mm_flags & AV_CPU_FLAG_SSE) &&
02426             !(CONFIG_MPEG_XVMC_DECODER && avctx->xvmc_acceleration > 1)){
02427             /* XvMCCreateBlocks() may not allocate 16-byte aligned blocks */
02428             c->clear_block  = clear_block_sse;
02429             c->clear_blocks = clear_blocks_sse;
02430         }
02431         }
02432 
02433 #define SET_HPEL_FUNCS(PFX, IDX, SIZE, CPU) \
02434         c->PFX ## _pixels_tab[IDX][0] = PFX ## _pixels ## SIZE ## _ ## CPU; \
02435         c->PFX ## _pixels_tab[IDX][1] = PFX ## _pixels ## SIZE ## _x2_ ## CPU; \
02436         c->PFX ## _pixels_tab[IDX][2] = PFX ## _pixels ## SIZE ## _y2_ ## CPU; \
02437         c->PFX ## _pixels_tab[IDX][3] = PFX ## _pixels ## SIZE ## _xy2_ ## CPU
02438 
02439         if (!high_bit_depth) {
02440         SET_HPEL_FUNCS(put, 0, 16, mmx);
02441         SET_HPEL_FUNCS(put_no_rnd, 0, 16, mmx);
02442         SET_HPEL_FUNCS(avg, 0, 16, mmx);
02443         SET_HPEL_FUNCS(avg_no_rnd, 0, 16, mmx);
02444         SET_HPEL_FUNCS(put, 1, 8, mmx);
02445         SET_HPEL_FUNCS(put_no_rnd, 1, 8, mmx);
02446         SET_HPEL_FUNCS(avg, 1, 8, mmx);
02447         SET_HPEL_FUNCS(avg_no_rnd, 1, 8, mmx);
02448         }
02449 
02450 #if ARCH_X86_32 || !HAVE_YASM
02451         c->gmc= gmc_mmx;
02452 #endif
02453 #if ARCH_X86_32 && HAVE_YASM
02454         if (!high_bit_depth)
02455         c->emulated_edge_mc = emulated_edge_mc_mmx;
02456 #endif
02457 
02458         c->add_bytes= add_bytes_mmx;
02459 
02460         if (!high_bit_depth)
02461         c->draw_edges = draw_edges_mmx;
02462 
02463         if (CONFIG_H263_DECODER || CONFIG_H263_ENCODER) {
02464             c->h263_v_loop_filter= h263_v_loop_filter_mmx;
02465             c->h263_h_loop_filter= h263_h_loop_filter_mmx;
02466         }
02467 
02468 #if HAVE_YASM
02469         if (!high_bit_depth) {
02470         c->put_h264_chroma_pixels_tab[0]= ff_put_h264_chroma_mc8_mmx_rnd;
02471         c->put_h264_chroma_pixels_tab[1]= ff_put_h264_chroma_mc4_mmx;
02472         }
02473 
02474         c->put_rv40_chroma_pixels_tab[0]= ff_put_rv40_chroma_mc8_mmx;
02475         c->put_rv40_chroma_pixels_tab[1]= ff_put_rv40_chroma_mc4_mmx;
02476 #endif
02477 
02478         if (mm_flags & AV_CPU_FLAG_MMX2) {
02479             c->prefetch = prefetch_mmx2;
02480 
02481             if (!high_bit_depth) {
02482             c->put_pixels_tab[0][1] = put_pixels16_x2_mmx2;
02483             c->put_pixels_tab[0][2] = put_pixels16_y2_mmx2;
02484 
02485             c->avg_pixels_tab[0][0] = avg_pixels16_mmx2;
02486             c->avg_pixels_tab[0][1] = avg_pixels16_x2_mmx2;
02487             c->avg_pixels_tab[0][2] = avg_pixels16_y2_mmx2;
02488 
02489             c->put_pixels_tab[1][1] = put_pixels8_x2_mmx2;
02490             c->put_pixels_tab[1][2] = put_pixels8_y2_mmx2;
02491 
02492             c->avg_pixels_tab[1][0] = avg_pixels8_mmx2;
02493             c->avg_pixels_tab[1][1] = avg_pixels8_x2_mmx2;
02494             c->avg_pixels_tab[1][2] = avg_pixels8_y2_mmx2;
02495             }
02496 
02497             if(!(avctx->flags & CODEC_FLAG_BITEXACT)){
02498                 if (!high_bit_depth) {
02499                 c->put_no_rnd_pixels_tab[0][1] = put_no_rnd_pixels16_x2_mmx2;
02500                 c->put_no_rnd_pixels_tab[0][2] = put_no_rnd_pixels16_y2_mmx2;
02501                 c->put_no_rnd_pixels_tab[1][1] = put_no_rnd_pixels8_x2_mmx2;
02502                 c->put_no_rnd_pixels_tab[1][2] = put_no_rnd_pixels8_y2_mmx2;
02503                 c->avg_pixels_tab[0][3] = avg_pixels16_xy2_mmx2;
02504                 c->avg_pixels_tab[1][3] = avg_pixels8_xy2_mmx2;
02505                 }
02506 
02507                 if (CONFIG_VP3_DECODER && HAVE_YASM) {
02508                     c->vp3_v_loop_filter= ff_vp3_v_loop_filter_mmx2;
02509                     c->vp3_h_loop_filter= ff_vp3_h_loop_filter_mmx2;
02510                 }
02511             }
02512             if (CONFIG_VP3_DECODER && HAVE_YASM) {
02513                 c->vp3_idct_dc_add = ff_vp3_idct_dc_add_mmx2;
02514             }
02515 
02516             if (CONFIG_VP3_DECODER
02517                 && (avctx->codec_id == CODEC_ID_VP3 || avctx->codec_id == CODEC_ID_THEORA)) {
02518                 c->put_no_rnd_pixels_tab[1][1] = put_no_rnd_pixels8_x2_exact_mmx2;
02519                 c->put_no_rnd_pixels_tab[1][2] = put_no_rnd_pixels8_y2_exact_mmx2;
02520             }
02521 
02522 #define SET_QPEL_FUNCS(PFX, IDX, SIZE, CPU) \
02523             c->PFX ## _pixels_tab[IDX][ 0] = PFX ## SIZE ## _mc00_ ## CPU; \
02524             c->PFX ## _pixels_tab[IDX][ 1] = PFX ## SIZE ## _mc10_ ## CPU; \
02525             c->PFX ## _pixels_tab[IDX][ 2] = PFX ## SIZE ## _mc20_ ## CPU; \
02526             c->PFX ## _pixels_tab[IDX][ 3] = PFX ## SIZE ## _mc30_ ## CPU; \
02527             c->PFX ## _pixels_tab[IDX][ 4] = PFX ## SIZE ## _mc01_ ## CPU; \
02528             c->PFX ## _pixels_tab[IDX][ 5] = PFX ## SIZE ## _mc11_ ## CPU; \
02529             c->PFX ## _pixels_tab[IDX][ 6] = PFX ## SIZE ## _mc21_ ## CPU; \
02530             c->PFX ## _pixels_tab[IDX][ 7] = PFX ## SIZE ## _mc31_ ## CPU; \
02531             c->PFX ## _pixels_tab[IDX][ 8] = PFX ## SIZE ## _mc02_ ## CPU; \
02532             c->PFX ## _pixels_tab[IDX][ 9] = PFX ## SIZE ## _mc12_ ## CPU; \
02533             c->PFX ## _pixels_tab[IDX][10] = PFX ## SIZE ## _mc22_ ## CPU; \
02534             c->PFX ## _pixels_tab[IDX][11] = PFX ## SIZE ## _mc32_ ## CPU; \
02535             c->PFX ## _pixels_tab[IDX][12] = PFX ## SIZE ## _mc03_ ## CPU; \
02536             c->PFX ## _pixels_tab[IDX][13] = PFX ## SIZE ## _mc13_ ## CPU; \
02537             c->PFX ## _pixels_tab[IDX][14] = PFX ## SIZE ## _mc23_ ## CPU; \
02538             c->PFX ## _pixels_tab[IDX][15] = PFX ## SIZE ## _mc33_ ## CPU
02539 
02540             SET_QPEL_FUNCS(put_qpel, 0, 16, mmx2);
02541             SET_QPEL_FUNCS(put_qpel, 1, 8, mmx2);
02542             SET_QPEL_FUNCS(put_no_rnd_qpel, 0, 16, mmx2);
02543             SET_QPEL_FUNCS(put_no_rnd_qpel, 1, 8, mmx2);
02544             SET_QPEL_FUNCS(avg_qpel, 0, 16, mmx2);
02545             SET_QPEL_FUNCS(avg_qpel, 1, 8, mmx2);
02546 
02547             if (!high_bit_depth) {
02548             SET_QPEL_FUNCS(put_h264_qpel, 0, 16, mmx2);
02549             SET_QPEL_FUNCS(put_h264_qpel, 1, 8, mmx2);
02550             SET_QPEL_FUNCS(put_h264_qpel, 2, 4, mmx2);
02551             SET_QPEL_FUNCS(avg_h264_qpel, 0, 16, mmx2);
02552             SET_QPEL_FUNCS(avg_h264_qpel, 1, 8, mmx2);
02553             SET_QPEL_FUNCS(avg_h264_qpel, 2, 4, mmx2);
02554             }
02555 
02556             SET_QPEL_FUNCS(put_2tap_qpel, 0, 16, mmx2);
02557             SET_QPEL_FUNCS(put_2tap_qpel, 1, 8, mmx2);
02558             SET_QPEL_FUNCS(avg_2tap_qpel, 0, 16, mmx2);
02559             SET_QPEL_FUNCS(avg_2tap_qpel, 1, 8, mmx2);
02560 
02561 #if HAVE_YASM
02562             c->avg_rv40_chroma_pixels_tab[0]= ff_avg_rv40_chroma_mc8_mmx2;
02563             c->avg_rv40_chroma_pixels_tab[1]= ff_avg_rv40_chroma_mc4_mmx2;
02564 
02565             if (!high_bit_depth) {
02566             c->avg_h264_chroma_pixels_tab[0]= ff_avg_h264_chroma_mc8_mmx2_rnd;
02567             c->avg_h264_chroma_pixels_tab[1]= ff_avg_h264_chroma_mc4_mmx2;
02568             c->avg_h264_chroma_pixels_tab[2]= ff_avg_h264_chroma_mc2_mmx2;
02569             c->put_h264_chroma_pixels_tab[2]= ff_put_h264_chroma_mc2_mmx2;
02570             }
02571             if (bit_depth == 10) {
02572                 c->put_h264_chroma_pixels_tab[2]= ff_put_h264_chroma_mc2_10_mmxext;
02573                 c->avg_h264_chroma_pixels_tab[2]= ff_avg_h264_chroma_mc2_10_mmxext;
02574                 c->put_h264_chroma_pixels_tab[1]= ff_put_h264_chroma_mc4_10_mmxext;
02575                 c->avg_h264_chroma_pixels_tab[1]= ff_avg_h264_chroma_mc4_10_mmxext;
02576             }
02577 
02578             c->add_hfyu_median_prediction = ff_add_hfyu_median_prediction_mmx2;
02579 #endif
02580 #if HAVE_7REGS && HAVE_TEN_OPERANDS
02581             if( mm_flags&AV_CPU_FLAG_3DNOW )
02582                 c->add_hfyu_median_prediction = add_hfyu_median_prediction_cmov;
02583 #endif
02584 
02585         } else if (mm_flags & AV_CPU_FLAG_3DNOW) {
02586             c->prefetch = prefetch_3dnow;
02587 
02588             if (!high_bit_depth) {
02589             c->put_pixels_tab[0][1] = put_pixels16_x2_3dnow;
02590             c->put_pixels_tab[0][2] = put_pixels16_y2_3dnow;
02591 
02592             c->avg_pixels_tab[0][0] = avg_pixels16_3dnow;
02593             c->avg_pixels_tab[0][1] = avg_pixels16_x2_3dnow;
02594             c->avg_pixels_tab[0][2] = avg_pixels16_y2_3dnow;
02595 
02596             c->put_pixels_tab[1][1] = put_pixels8_x2_3dnow;
02597             c->put_pixels_tab[1][2] = put_pixels8_y2_3dnow;
02598 
02599             c->avg_pixels_tab[1][0] = avg_pixels8_3dnow;
02600             c->avg_pixels_tab[1][1] = avg_pixels8_x2_3dnow;
02601             c->avg_pixels_tab[1][2] = avg_pixels8_y2_3dnow;
02602 
02603             if(!(avctx->flags & CODEC_FLAG_BITEXACT)){
02604                 c->put_no_rnd_pixels_tab[0][1] = put_no_rnd_pixels16_x2_3dnow;
02605                 c->put_no_rnd_pixels_tab[0][2] = put_no_rnd_pixels16_y2_3dnow;
02606                 c->put_no_rnd_pixels_tab[1][1] = put_no_rnd_pixels8_x2_3dnow;
02607                 c->put_no_rnd_pixels_tab[1][2] = put_no_rnd_pixels8_y2_3dnow;
02608                 c->avg_pixels_tab[0][3] = avg_pixels16_xy2_3dnow;
02609                 c->avg_pixels_tab[1][3] = avg_pixels8_xy2_3dnow;
02610             }
02611             }
02612 
02613             if (CONFIG_VP3_DECODER
02614                 && (avctx->codec_id == CODEC_ID_VP3 || avctx->codec_id == CODEC_ID_THEORA)) {
02615                 c->put_no_rnd_pixels_tab[1][1] = put_no_rnd_pixels8_x2_exact_3dnow;
02616                 c->put_no_rnd_pixels_tab[1][2] = put_no_rnd_pixels8_y2_exact_3dnow;
02617             }
02618 
02619             SET_QPEL_FUNCS(put_qpel, 0, 16, 3dnow);
02620             SET_QPEL_FUNCS(put_qpel, 1, 8, 3dnow);
02621             SET_QPEL_FUNCS(put_no_rnd_qpel, 0, 16, 3dnow);
02622             SET_QPEL_FUNCS(put_no_rnd_qpel, 1, 8, 3dnow);
02623             SET_QPEL_FUNCS(avg_qpel, 0, 16, 3dnow);
02624             SET_QPEL_FUNCS(avg_qpel, 1, 8, 3dnow);
02625 
02626             if (!high_bit_depth) {
02627             SET_QPEL_FUNCS(put_h264_qpel, 0, 16, 3dnow);
02628             SET_QPEL_FUNCS(put_h264_qpel, 1, 8, 3dnow);
02629             SET_QPEL_FUNCS(put_h264_qpel, 2, 4, 3dnow);
02630             SET_QPEL_FUNCS(avg_h264_qpel, 0, 16, 3dnow);
02631             SET_QPEL_FUNCS(avg_h264_qpel, 1, 8, 3dnow);
02632             SET_QPEL_FUNCS(avg_h264_qpel, 2, 4, 3dnow);
02633             }
02634 
02635             SET_QPEL_FUNCS(put_2tap_qpel, 0, 16, 3dnow);
02636             SET_QPEL_FUNCS(put_2tap_qpel, 1, 8, 3dnow);
02637             SET_QPEL_FUNCS(avg_2tap_qpel, 0, 16, 3dnow);
02638             SET_QPEL_FUNCS(avg_2tap_qpel, 1, 8, 3dnow);
02639 
02640 #if HAVE_YASM
02641             if (!high_bit_depth) {
02642             c->avg_h264_chroma_pixels_tab[0]= ff_avg_h264_chroma_mc8_3dnow_rnd;
02643             c->avg_h264_chroma_pixels_tab[1]= ff_avg_h264_chroma_mc4_3dnow;
02644             }
02645 
02646             c->avg_rv40_chroma_pixels_tab[0]= ff_avg_rv40_chroma_mc8_3dnow;
02647             c->avg_rv40_chroma_pixels_tab[1]= ff_avg_rv40_chroma_mc4_3dnow;
02648 #endif
02649         }
02650 
02651 
02652 #define H264_QPEL_FUNCS(x, y, CPU)\
02653             c->put_h264_qpel_pixels_tab[0][x+y*4] = put_h264_qpel16_mc##x##y##_##CPU;\
02654             c->put_h264_qpel_pixels_tab[1][x+y*4] = put_h264_qpel8_mc##x##y##_##CPU;\
02655             c->avg_h264_qpel_pixels_tab[0][x+y*4] = avg_h264_qpel16_mc##x##y##_##CPU;\
02656             c->avg_h264_qpel_pixels_tab[1][x+y*4] = avg_h264_qpel8_mc##x##y##_##CPU;
02657         if((mm_flags & AV_CPU_FLAG_SSE2) && !(mm_flags & AV_CPU_FLAG_3DNOW)){
02658             // these functions are slower than mmx on AMD, but faster on Intel
02659             if (!high_bit_depth) {
02660             c->put_pixels_tab[0][0] = put_pixels16_sse2;
02661             c->put_no_rnd_pixels_tab[0][0] = put_pixels16_sse2;
02662             c->avg_pixels_tab[0][0] = avg_pixels16_sse2;
02663             H264_QPEL_FUNCS(0, 0, sse2);
02664             }
02665         }
02666         if(mm_flags & AV_CPU_FLAG_SSE2){
02667             if (!high_bit_depth) {
02668             H264_QPEL_FUNCS(0, 1, sse2);
02669             H264_QPEL_FUNCS(0, 2, sse2);
02670             H264_QPEL_FUNCS(0, 3, sse2);
02671             H264_QPEL_FUNCS(1, 1, sse2);
02672             H264_QPEL_FUNCS(1, 2, sse2);
02673             H264_QPEL_FUNCS(1, 3, sse2);
02674             H264_QPEL_FUNCS(2, 1, sse2);
02675             H264_QPEL_FUNCS(2, 2, sse2);
02676             H264_QPEL_FUNCS(2, 3, sse2);
02677             H264_QPEL_FUNCS(3, 1, sse2);
02678             H264_QPEL_FUNCS(3, 2, sse2);
02679             H264_QPEL_FUNCS(3, 3, sse2);
02680             }
02681 #if HAVE_YASM
02682             if (bit_depth == 10) {
02683                 c->put_h264_chroma_pixels_tab[0]= ff_put_h264_chroma_mc8_10_sse2;
02684                 c->avg_h264_chroma_pixels_tab[0]= ff_avg_h264_chroma_mc8_10_sse2;
02685             }
02686 #endif
02687         }
02688 #if HAVE_SSSE3
02689         if(mm_flags & AV_CPU_FLAG_SSSE3){
02690             if (!high_bit_depth) {
02691             H264_QPEL_FUNCS(1, 0, ssse3);
02692             H264_QPEL_FUNCS(1, 1, ssse3);
02693             H264_QPEL_FUNCS(1, 2, ssse3);
02694             H264_QPEL_FUNCS(1, 3, ssse3);
02695             H264_QPEL_FUNCS(2, 0, ssse3);
02696             H264_QPEL_FUNCS(2, 1, ssse3);
02697             H264_QPEL_FUNCS(2, 2, ssse3);
02698             H264_QPEL_FUNCS(2, 3, ssse3);
02699             H264_QPEL_FUNCS(3, 0, ssse3);
02700             H264_QPEL_FUNCS(3, 1, ssse3);
02701             H264_QPEL_FUNCS(3, 2, ssse3);
02702             H264_QPEL_FUNCS(3, 3, ssse3);
02703             }
02704 #if HAVE_YASM
02705             if (!high_bit_depth) {
02706             c->put_h264_chroma_pixels_tab[0]= ff_put_h264_chroma_mc8_ssse3_rnd;
02707             c->avg_h264_chroma_pixels_tab[0]= ff_avg_h264_chroma_mc8_ssse3_rnd;
02708             c->put_h264_chroma_pixels_tab[1]= ff_put_h264_chroma_mc4_ssse3;
02709             c->avg_h264_chroma_pixels_tab[1]= ff_avg_h264_chroma_mc4_ssse3;
02710             }
02711             c->add_hfyu_left_prediction = ff_add_hfyu_left_prediction_ssse3;
02712             if (mm_flags & AV_CPU_FLAG_SSE4) // not really sse4, just slow on Conroe
02713                 c->add_hfyu_left_prediction = ff_add_hfyu_left_prediction_sse4;
02714 #endif
02715         }
02716 #endif
02717 
02718         if(mm_flags & AV_CPU_FLAG_3DNOW){
02719             c->vorbis_inverse_coupling = vorbis_inverse_coupling_3dnow;
02720             c->vector_fmul = vector_fmul_3dnow;
02721         }
02722         if(mm_flags & AV_CPU_FLAG_3DNOWEXT){
02723             c->vector_fmul_reverse = vector_fmul_reverse_3dnow2;
02724 #if HAVE_6REGS
02725             c->vector_fmul_window = vector_fmul_window_3dnow2;
02726 #endif
02727         }
02728         if(mm_flags & AV_CPU_FLAG_MMX2){
02729 #if HAVE_YASM
02730             c->scalarproduct_int16 = ff_scalarproduct_int16_mmx2;
02731             c->scalarproduct_and_madd_int16 = ff_scalarproduct_and_madd_int16_mmx2;
02732             if (avctx->flags & CODEC_FLAG_BITEXACT) {
02733                 c->apply_window_int16 = ff_apply_window_int16_mmxext_ba;
02734             } else {
02735                 c->apply_window_int16 = ff_apply_window_int16_mmxext;
02736             }
02737 #endif
02738         }
02739         if(mm_flags & AV_CPU_FLAG_SSE){
02740             c->vorbis_inverse_coupling = vorbis_inverse_coupling_sse;
02741             c->ac3_downmix = ac3_downmix_sse;
02742             c->vector_fmul = vector_fmul_sse;
02743             c->vector_fmul_reverse = vector_fmul_reverse_sse;
02744             c->vector_fmul_add = vector_fmul_add_sse;
02745 #if HAVE_6REGS
02746             c->vector_fmul_window = vector_fmul_window_sse;
02747 #endif
02748             c->vector_clipf = vector_clipf_sse;
02749 #if HAVE_YASM
02750             c->scalarproduct_float = ff_scalarproduct_float_sse;
02751 #endif
02752         }
02753         if(mm_flags & AV_CPU_FLAG_3DNOW)
02754             c->vector_fmul_add = vector_fmul_add_3dnow; // faster than sse
02755         if(mm_flags & AV_CPU_FLAG_SSE2){
02756 #if HAVE_YASM
02757             c->scalarproduct_int16 = ff_scalarproduct_int16_sse2;
02758             c->scalarproduct_and_madd_int16 = ff_scalarproduct_and_madd_int16_sse2;
02759             if (avctx->flags & CODEC_FLAG_BITEXACT) {
02760                 c->apply_window_int16 = ff_apply_window_int16_sse2_ba;
02761             } else {
02762                 if (!(mm_flags & AV_CPU_FLAG_SSE2SLOW)) {
02763                     c->apply_window_int16 = ff_apply_window_int16_sse2;
02764                 }
02765             }
02766 
02767             if (!high_bit_depth)
02768             c->emulated_edge_mc = emulated_edge_mc_sse;
02769             c->gmc= gmc_sse;
02770 #endif
02771         }
02772         if (mm_flags & AV_CPU_FLAG_SSSE3) {
02773 #if HAVE_YASM
02774             if (mm_flags & AV_CPU_FLAG_ATOM) {
02775                 c->apply_window_int16 = ff_apply_window_int16_ssse3_atom;
02776             } else {
02777                 c->apply_window_int16 = ff_apply_window_int16_ssse3;
02778             }
02779             if (!(mm_flags & (AV_CPU_FLAG_SSE42|AV_CPU_FLAG_3DNOW))) { // cachesplit
02780                 c->scalarproduct_and_madd_int16 = ff_scalarproduct_and_madd_int16_ssse3;
02781             }
02782 #endif
02783         }
02784 #if HAVE_AVX && HAVE_YASM
02785         if (mm_flags & AV_CPU_FLAG_AVX) {
02786             if (bit_depth == 10) {
02787                 c->put_h264_chroma_pixels_tab[0]= ff_put_h264_chroma_mc8_10_avx;
02788                 c->avg_h264_chroma_pixels_tab[0]= ff_avg_h264_chroma_mc8_10_avx;
02789             }
02790         }
02791 #endif
02792     }
02793 
02794     if (CONFIG_ENCODERS)
02795         dsputilenc_init_mmx(c, avctx);
02796 
02797 #if 0
02798     // for speed testing
02799     get_pixels = just_return;
02800     put_pixels_clamped = just_return;
02801     add_pixels_clamped = just_return;
02802 
02803     pix_abs16x16 = just_return;
02804     pix_abs16x16_x2 = just_return;
02805     pix_abs16x16_y2 = just_return;
02806     pix_abs16x16_xy2 = just_return;
02807 
02808     put_pixels_tab[0] = just_return;
02809     put_pixels_tab[1] = just_return;
02810     put_pixels_tab[2] = just_return;
02811     put_pixels_tab[3] = just_return;
02812 
02813     put_no_rnd_pixels_tab[0] = just_return;
02814     put_no_rnd_pixels_tab[1] = just_return;
02815     put_no_rnd_pixels_tab[2] = just_return;
02816     put_no_rnd_pixels_tab[3] = just_return;
02817 
02818     avg_pixels_tab[0] = just_return;
02819     avg_pixels_tab[1] = just_return;
02820     avg_pixels_tab[2] = just_return;
02821     avg_pixels_tab[3] = just_return;
02822 
02823     avg_no_rnd_pixels_tab[0] = just_return;
02824     avg_no_rnd_pixels_tab[1] = just_return;
02825     avg_no_rnd_pixels_tab[2] = just_return;
02826     avg_no_rnd_pixels_tab[3] = just_return;
02827 
02828     //av_fdct = just_return;
02829     //ff_idct = just_return;
02830 #endif
02831 }

Generated on Wed Apr 11 2012 07:31:35 for FFmpeg by  doxygen 1.7.1