Libav
|
00001 /* 00002 * MLP DSP functions x86-optimized 00003 * Copyright (c) 2009 Ramiro Polla 00004 * 00005 * This file is part of FFmpeg. 00006 * 00007 * FFmpeg is free software; you can redistribute it and/or 00008 * modify it under the terms of the GNU Lesser General Public 00009 * License as published by the Free Software Foundation; either 00010 * version 2.1 of the License, or (at your option) any later version. 00011 * 00012 * FFmpeg is distributed in the hope that it will be useful, 00013 * but WITHOUT ANY WARRANTY; without even the implied warranty of 00014 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU 00015 * Lesser General Public License for more details. 00016 * 00017 * You should have received a copy of the GNU Lesser General Public 00018 * License along with FFmpeg; if not, write to the Free Software 00019 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA 00020 */ 00021 00022 #include "libavutil/x86_cpu.h" 00023 #include "libavcodec/dsputil.h" 00024 #include "libavcodec/mlp.h" 00025 00026 #if HAVE_7REGS && HAVE_TEN_OPERANDS 00027 00028 extern void ff_mlp_firorder_8; 00029 extern void ff_mlp_firorder_7; 00030 extern void ff_mlp_firorder_6; 00031 extern void ff_mlp_firorder_5; 00032 extern void ff_mlp_firorder_4; 00033 extern void ff_mlp_firorder_3; 00034 extern void ff_mlp_firorder_2; 00035 extern void ff_mlp_firorder_1; 00036 extern void ff_mlp_firorder_0; 00037 00038 extern void ff_mlp_iirorder_4; 00039 extern void ff_mlp_iirorder_3; 00040 extern void ff_mlp_iirorder_2; 00041 extern void ff_mlp_iirorder_1; 00042 extern void ff_mlp_iirorder_0; 00043 00044 static const void *firtable[9] = { &ff_mlp_firorder_0, &ff_mlp_firorder_1, 00045 &ff_mlp_firorder_2, &ff_mlp_firorder_3, 00046 &ff_mlp_firorder_4, &ff_mlp_firorder_5, 00047 &ff_mlp_firorder_6, &ff_mlp_firorder_7, 00048 &ff_mlp_firorder_8 }; 00049 static const void *iirtable[5] = { &ff_mlp_iirorder_0, &ff_mlp_iirorder_1, 00050 &ff_mlp_iirorder_2, &ff_mlp_iirorder_3, 00051 &ff_mlp_iirorder_4 }; 00052 00053 #if ARCH_X86_64 00054 00055 #define MLPMUL(label, offset, offs, offc) \ 00056 LABEL_MANGLE(label)": \n\t" \ 00057 "movslq "offset"+"offs"(%0), %%rax\n\t" \ 00058 "movslq "offset"+"offc"(%1), %%rdx\n\t" \ 00059 "imul %%rdx, %%rax\n\t" \ 00060 "add %%rax, %%rsi\n\t" 00061 00062 #define FIRMULREG(label, offset, firc)\ 00063 LABEL_MANGLE(label)": \n\t" \ 00064 "movslq "#offset"(%0), %%rax\n\t" \ 00065 "imul %"#firc", %%rax\n\t" \ 00066 "add %%rax, %%rsi\n\t" 00067 00068 #define CLEAR_ACCUM \ 00069 "xor %%rsi, %%rsi\n\t" 00070 00071 #define SHIFT_ACCUM \ 00072 "shr %%cl, %%rsi\n\t" 00073 00074 #define ACCUM "%%rdx" 00075 #define RESULT "%%rsi" 00076 #define RESULT32 "%%esi" 00077 00078 #else /* if ARCH_X86_32 */ 00079 00080 #define MLPMUL(label, offset, offs, offc) \ 00081 LABEL_MANGLE(label)": \n\t" \ 00082 "mov "offset"+"offs"(%0), %%eax\n\t" \ 00083 "imull "offset"+"offc"(%1) \n\t" \ 00084 "add %%eax , %%esi\n\t" \ 00085 "adc %%edx , %%ecx\n\t" 00086 00087 #define FIRMULREG(label, offset, firc) \ 00088 MLPMUL(label, #offset, "0", "0") 00089 00090 #define CLEAR_ACCUM \ 00091 "xor %%esi, %%esi\n\t" \ 00092 "xor %%ecx, %%ecx\n\t" 00093 00094 #define SHIFT_ACCUM \ 00095 "mov %%ecx, %%edx\n\t" \ 00096 "mov %%esi, %%eax\n\t" \ 00097 "movzbl %7 , %%ecx\n\t" \ 00098 "shrd %%cl, %%edx, %%eax\n\t" \ 00099 00100 #define ACCUM "%%edx" 00101 #define RESULT "%%eax" 00102 #define RESULT32 "%%eax" 00103 00104 #endif /* !ARCH_X86_64 */ 00105 00106 #define BINC AV_STRINGIFY(4* MAX_CHANNELS) 00107 #define IOFFS AV_STRINGIFY(4*(MAX_FIR_ORDER + MAX_BLOCKSIZE)) 00108 #define IOFFC AV_STRINGIFY(4* MAX_FIR_ORDER) 00109 00110 #define FIRMUL(label, offset) MLPMUL(label, #offset, "0", "0") 00111 #define IIRMUL(label, offset) MLPMUL(label, #offset, IOFFS, IOFFC) 00112 00113 static void mlp_filter_channel_x86(int32_t *state, const int32_t *coeff, 00114 int firorder, int iirorder, 00115 unsigned int filter_shift, int32_t mask, 00116 int blocksize, int32_t *sample_buffer) 00117 { 00118 const void *firjump = firtable[firorder]; 00119 const void *iirjump = iirtable[iirorder]; 00120 00121 blocksize = -blocksize; 00122 00123 __asm__ volatile( 00124 "1: \n\t" 00125 CLEAR_ACCUM 00126 "jmp *%5 \n\t" 00127 FIRMUL (ff_mlp_firorder_8, 0x1c ) 00128 FIRMUL (ff_mlp_firorder_7, 0x18 ) 00129 FIRMUL (ff_mlp_firorder_6, 0x14 ) 00130 FIRMUL (ff_mlp_firorder_5, 0x10 ) 00131 FIRMUL (ff_mlp_firorder_4, 0x0c ) 00132 FIRMULREG(ff_mlp_firorder_3, 0x08,10) 00133 FIRMULREG(ff_mlp_firorder_2, 0x04, 9) 00134 FIRMULREG(ff_mlp_firorder_1, 0x00, 8) 00135 LABEL_MANGLE(ff_mlp_firorder_0)":\n\t" 00136 "jmp *%6 \n\t" 00137 IIRMUL (ff_mlp_iirorder_4, 0x0c ) 00138 IIRMUL (ff_mlp_iirorder_3, 0x08 ) 00139 IIRMUL (ff_mlp_iirorder_2, 0x04 ) 00140 IIRMUL (ff_mlp_iirorder_1, 0x00 ) 00141 LABEL_MANGLE(ff_mlp_iirorder_0)":\n\t" 00142 SHIFT_ACCUM 00143 "mov "RESULT" ,"ACCUM" \n\t" 00144 "add (%2) ,"RESULT" \n\t" 00145 "and %4 ,"RESULT" \n\t" 00146 "sub $4 , %0 \n\t" 00147 "mov "RESULT32", (%0) \n\t" 00148 "mov "RESULT32", (%2) \n\t" 00149 "add $"BINC" , %2 \n\t" 00150 "sub "ACCUM" ,"RESULT" \n\t" 00151 "mov "RESULT32","IOFFS"(%0) \n\t" 00152 "incl %3 \n\t" 00153 "js 1b \n\t" 00154 : /* 0*/"+r"(state), 00155 /* 1*/"+r"(coeff), 00156 /* 2*/"+r"(sample_buffer), 00157 #if ARCH_X86_64 00158 /* 3*/"+r"(blocksize) 00159 : /* 4*/"r"((x86_reg)mask), /* 5*/"r"(firjump), 00160 /* 6*/"r"(iirjump) , /* 7*/"c"(filter_shift) 00161 , /* 8*/"r"((int64_t)coeff[0]) 00162 , /* 9*/"r"((int64_t)coeff[1]) 00163 , /*10*/"r"((int64_t)coeff[2]) 00164 : "rax", "rdx", "rsi" 00165 #else /* ARCH_X86_32 */ 00166 /* 3*/"+m"(blocksize) 00167 : /* 4*/"m"( mask), /* 5*/"m"(firjump), 00168 /* 6*/"m"(iirjump) , /* 7*/"m"(filter_shift) 00169 : "eax", "edx", "esi", "ecx" 00170 #endif /* !ARCH_X86_64 */ 00171 ); 00172 } 00173 00174 #endif /* HAVE_7REGS && HAVE_TEN_OPERANDS */ 00175 00176 void ff_mlp_init_x86(DSPContext* c, AVCodecContext *avctx) 00177 { 00178 #if HAVE_7REGS && HAVE_TEN_OPERANDS 00179 c->mlp_filter_channel = mlp_filter_channel_x86; 00180 #endif 00181 }