Mercurial > cgi-bin > hgwebdir.cgi > PR > Applications > VSs > VSs__H264__App

annotate libavcodec/x86/dsputil_mmx.h @ 4:96e628866d41

find changesets by author, revision, files, or words in the commit message
naming some tasks to help debugging
author Nina Engelhardt <nengel@mailbox.tu-berlin.de>
date Wed, 19 Dec 2012 15:40:26 +0100
parents
children
rev   line source
nengel@2 1 /*
nengel@2 2 * MMX optimized DSP utils
nengel@2 3 * Copyright (c) 2007 Aurelien Jacobs <aurel@gnuage.org>
nengel@2 4 *
nengel@2 5 * This file is part of FFmpeg.
nengel@2 6 *
nengel@2 7 * FFmpeg is free software; you can redistribute it and/or
nengel@2 8 * modify it under the terms of the GNU Lesser General Public
nengel@2 9 * License as published by the Free Software Foundation; either
nengel@2 10 * version 2.1 of the License, or (at your option) any later version.
nengel@2 11 *
nengel@2 12 * FFmpeg is distributed in the hope that it will be useful,
nengel@2 13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
nengel@2 14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
nengel@2 15 * Lesser General Public License for more details.
nengel@2 16 *
nengel@2 17 * You should have received a copy of the GNU Lesser General Public
nengel@2 18 * License along with FFmpeg; if not, write to the Free Software
nengel@2 19 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
nengel@2 20 */
nengel@2 21
nengel@2 22 #ifndef AVCODEC_X86_DSPUTIL_MMX_H
nengel@2 23 #define AVCODEC_X86_DSPUTIL_MMX_H
nengel@2 24
nengel@2 25 #include <stdint.h>
nengel@2 26 #include "libavcodec/dsputil.h"
nengel@2 27
nengel@2 28 typedef struct { uint64_t a, b; } xmm_reg;
nengel@2 29
nengel@2 30 extern const uint64_t ff_bone;
nengel@2 31 extern const uint64_t ff_wtwo;
nengel@2 32
nengel@2 33 extern const uint64_t ff_pdw_80000000[2];
nengel@2 34
nengel@2 35 extern const uint64_t ff_pw_3;
nengel@2 36 extern const uint64_t ff_pw_4;
nengel@2 37 extern const xmm_reg ff_pw_5;
nengel@2 38 extern const xmm_reg ff_pw_8;
nengel@2 39 extern const uint64_t ff_pw_15;
nengel@2 40 extern const xmm_reg ff_pw_16;
nengel@2 41 extern const uint64_t ff_pw_20;
nengel@2 42 extern const xmm_reg ff_pw_28;
nengel@2 43 extern const xmm_reg ff_pw_32;
nengel@2 44 extern const uint64_t ff_pw_42;
nengel@2 45 extern const xmm_reg ff_pw_64;
nengel@2 46 extern const uint64_t ff_pw_96;
nengel@2 47 extern const uint64_t ff_pw_128;
nengel@2 48 extern const uint64_t ff_pw_255;
nengel@2 49
nengel@2 50 extern const uint64_t ff_pb_1;
nengel@2 51 extern const uint64_t ff_pb_3;
nengel@2 52 extern const uint64_t ff_pb_7;
nengel@2 53 extern const uint64_t ff_pb_1F;
nengel@2 54 extern const uint64_t ff_pb_3F;
nengel@2 55 extern const uint64_t ff_pb_81;
nengel@2 56 extern const uint64_t ff_pb_A1;
nengel@2 57 extern const uint64_t ff_pb_FC;
nengel@2 58
nengel@2 59 extern const double ff_pd_1[2];
nengel@2 60 extern const double ff_pd_2[2];
nengel@2 61
nengel@2 62 #define LOAD4(stride,in,a,b,c,d)\
nengel@2 63 "movq 0*"#stride"+"#in", "#a"\n\t"\
nengel@2 64 "movq 1*"#stride"+"#in", "#b"\n\t"\
nengel@2 65 "movq 2*"#stride"+"#in", "#c"\n\t"\
nengel@2 66 "movq 3*"#stride"+"#in", "#d"\n\t"
nengel@2 67
nengel@2 68 #define STORE4(stride,out,a,b,c,d)\
nengel@2 69 "movq "#a", 0*"#stride"+"#out"\n\t"\
nengel@2 70 "movq "#b", 1*"#stride"+"#out"\n\t"\
nengel@2 71 "movq "#c", 2*"#stride"+"#out"\n\t"\
nengel@2 72 "movq "#d", 3*"#stride"+"#out"\n\t"
nengel@2 73
nengel@2 74 /* in/out: mma=mma+mmb, mmb=mmb-mma */
nengel@2 75 #define SUMSUB_BA( a, b ) \
nengel@2 76 "paddw "#b", "#a" \n\t"\
nengel@2 77 "paddw "#b", "#b" \n\t"\
nengel@2 78 "psubw "#a", "#b" \n\t"
nengel@2 79
nengel@2 80 #define SBUTTERFLY(a,b,t,n,m)\
nengel@2 81 "mov" #m " " #a ", " #t " \n\t" /* abcd */\
nengel@2 82 "punpckl" #n " " #b ", " #a " \n\t" /* aebf */\
nengel@2 83 "punpckh" #n " " #b ", " #t " \n\t" /* cgdh */\
nengel@2 84
nengel@2 85 #define TRANSPOSE4(a,b,c,d,t)\
nengel@2 86 SBUTTERFLY(a,b,t,wd,q) /* a=aebf t=cgdh */\
nengel@2 87 SBUTTERFLY(c,d,b,wd,q) /* c=imjn b=kolp */\
nengel@2 88 SBUTTERFLY(a,c,d,dq,q) /* a=aeim d=bfjn */\
nengel@2 89 SBUTTERFLY(t,b,c,dq,q) /* t=cgko c=dhlp */
nengel@2 90
nengel@2 91 // e,f,g,h can be memory
nengel@2 92 // out: a,d,t,c
nengel@2 93 #define TRANSPOSE8x4(a,b,c,d,e,f,g,h,t)\
nengel@2 94 "punpcklbw " #e ", " #a " \n\t" /* a0 e0 a1 e1 a2 e2 a3 e3 */\
nengel@2 95 "punpcklbw " #f ", " #b " \n\t" /* b0 f0 b1 f1 b2 f2 b3 f3 */\
nengel@2 96 "punpcklbw " #g ", " #c " \n\t" /* c0 g0 c1 g1 c2 g2 d3 g3 */\
nengel@2 97 "punpcklbw " #h ", " #d " \n\t" /* d0 h0 d1 h1 d2 h2 d3 h3 */\
nengel@2 98 SBUTTERFLY(a, b, t, bw, q) /* a= a0 b0 e0 f0 a1 b1 e1 f1 */\
nengel@2 99 /* t= a2 b2 e2 f2 a3 b3 e3 f3 */\
nengel@2 100 SBUTTERFLY(c, d, b, bw, q) /* c= c0 d0 g0 h0 c1 d1 g1 h1 */\
nengel@2 101 /* b= c2 d2 g2 h2 c3 d3 g3 h3 */\
nengel@2 102 SBUTTERFLY(a, c, d, wd, q) /* a= a0 b0 c0 d0 e0 f0 g0 h0 */\
nengel@2 103 /* d= a1 b1 c1 d1 e1 f1 g1 h1 */\
nengel@2 104 SBUTTERFLY(t, b, c, wd, q) /* t= a2 b2 c2 d2 e2 f2 g2 h2 */\
nengel@2 105 /* c= a3 b3 c3 d3 e3 f3 g3 h3 */
nengel@2 106
nengel@2 107 #if ARCH_X86_64
nengel@2 108 // permutes 01234567 -> 05736421
nengel@2 109 #define TRANSPOSE8(a,b,c,d,e,f,g,h,t)\
nengel@2 110 SBUTTERFLY(a,b,%%xmm8,wd,dqa)\
nengel@2 111 SBUTTERFLY(c,d,b,wd,dqa)\
nengel@2 112 SBUTTERFLY(e,f,d,wd,dqa)\
nengel@2 113 SBUTTERFLY(g,h,f,wd,dqa)\
nengel@2 114 SBUTTERFLY(a,c,h,dq,dqa)\
nengel@2 115 SBUTTERFLY(%%xmm8,b,c,dq,dqa)\
nengel@2 116 SBUTTERFLY(e,g,b,dq,dqa)\
nengel@2 117 SBUTTERFLY(d,f,g,dq,dqa)\
nengel@2 118 SBUTTERFLY(a,e,f,qdq,dqa)\
nengel@2 119 SBUTTERFLY(%%xmm8,d,e,qdq,dqa)\
nengel@2 120 SBUTTERFLY(h,b,d,qdq,dqa)\
nengel@2 121 SBUTTERFLY(c,g,b,qdq,dqa)\
nengel@2 122 "movdqa %%xmm8, "#g" \n\t"
nengel@2 123 #else
nengel@2 124 #define TRANSPOSE8(a,b,c,d,e,f,g,h,t)\
nengel@2 125 "movdqa "#h", "#t" \n\t"\
nengel@2 126 SBUTTERFLY(a,b,h,wd,dqa)\
nengel@2 127 "movdqa "#h", 16"#t" \n\t"\
nengel@2 128 "movdqa "#t", "#h" \n\t"\
nengel@2 129 SBUTTERFLY(c,d,b,wd,dqa)\
nengel@2 130 SBUTTERFLY(e,f,d,wd,dqa)\
nengel@2 131 SBUTTERFLY(g,h,f,wd,dqa)\
nengel@2 132 SBUTTERFLY(a,c,h,dq,dqa)\
nengel@2 133 "movdqa "#h", "#t" \n\t"\
nengel@2 134 "movdqa 16"#t", "#h" \n\t"\
nengel@2 135 SBUTTERFLY(h,b,c,dq,dqa)\
nengel@2 136 SBUTTERFLY(e,g,b,dq,dqa)\
nengel@2 137 SBUTTERFLY(d,f,g,dq,dqa)\
nengel@2 138 SBUTTERFLY(a,e,f,qdq,dqa)\
nengel@2 139 SBUTTERFLY(h,d,e,qdq,dqa)\
nengel@2 140 "movdqa "#h", 16"#t" \n\t"\
nengel@2 141 "movdqa "#t", "#h" \n\t"\
nengel@2 142 SBUTTERFLY(h,b,d,qdq,dqa)\
nengel@2 143 SBUTTERFLY(c,g,b,qdq,dqa)\
nengel@2 144 "movdqa 16"#t", "#g" \n\t"
nengel@2 145 #endif
nengel@2 146
nengel@2 147 #define MOVQ_WONE(regd) \
nengel@2 148 __asm__ volatile ( \
nengel@2 149 "pcmpeqd %%" #regd ", %%" #regd " \n\t" \
nengel@2 150 "psrlw $15, %%" #regd ::)
nengel@2 151
nengel@2 152 void add_pixels_clamped_mmx(const DCTELEM *block, uint8_t *pixels, int line_size);
nengel@2 153 void put_pixels_clamped_mmx(const DCTELEM *block, uint8_t *pixels, int line_size);
nengel@2 154 void put_signed_pixels_clamped_mmx(const DCTELEM *block, uint8_t *pixels, int line_size);
nengel@2 155
nengel@2 156 void ff_put_cavs_qpel8_mc00_mmx2(uint8_t *dst, uint8_t *src, int stride);
nengel@2 157 void ff_avg_cavs_qpel8_mc00_mmx2(uint8_t *dst, uint8_t *src, int stride);
nengel@2 158 void ff_put_cavs_qpel16_mc00_mmx2(uint8_t *dst, uint8_t *src, int stride);
nengel@2 159 void ff_avg_cavs_qpel16_mc00_mmx2(uint8_t *dst, uint8_t *src, int stride);
nengel@2 160
nengel@2 161 void ff_put_vc1_mspel_mc00_mmx(uint8_t *dst, const uint8_t *src, int stride, int rnd);
nengel@2 162 void ff_avg_vc1_mspel_mc00_mmx2(uint8_t *dst, const uint8_t *src, int stride, int rnd);
nengel@2 163
nengel@2 164 void ff_lpc_compute_autocorr_sse2(const int32_t *data, int len, int lag,
nengel@2 165 double *autoc);
nengel@2 166
nengel@2 167 void ff_mmx_idct(DCTELEM *block);
nengel@2 168 void ff_mmxext_idct(DCTELEM *block);
nengel@2 169
nengel@2 170 #endif /* AVCODEC_X86_DSPUTIL_MMX_H */