Libav
|
00001 /* 00002 * iWMMXt optimized DSP utils 00003 * Copyright (c) 2004 AGAWA Koji 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 "libavcodec/dsputil.h" 00023 00024 #define DEF(x, y) x ## _no_rnd_ ## y ##_iwmmxt 00025 #define SET_RND(regd) __asm__ volatile ("mov r12, #1 \n\t tbcsth " #regd ", r12":::"r12"); 00026 #define WAVG2B "wavg2b" 00027 #include "dsputil_iwmmxt_rnd_template.c" 00028 #undef DEF 00029 #undef SET_RND 00030 #undef WAVG2B 00031 00032 #define DEF(x, y) x ## _ ## y ##_iwmmxt 00033 #define SET_RND(regd) __asm__ volatile ("mov r12, #2 \n\t tbcsth " #regd ", r12":::"r12"); 00034 #define WAVG2B "wavg2br" 00035 #include "dsputil_iwmmxt_rnd_template.c" 00036 #undef DEF 00037 #undef SET_RND 00038 #undef WAVG2BR 00039 00040 // need scheduling 00041 #define OP(AVG) \ 00042 __asm__ volatile ( \ 00043 /* alignment */ \ 00044 "and r12, %[pixels], #7 \n\t" \ 00045 "bic %[pixels], %[pixels], #7 \n\t" \ 00046 "tmcr wcgr1, r12 \n\t" \ 00047 \ 00048 "wldrd wr0, [%[pixels]] \n\t" \ 00049 "wldrd wr1, [%[pixels], #8] \n\t" \ 00050 "add %[pixels], %[pixels], %[line_size] \n\t" \ 00051 "walignr1 wr4, wr0, wr1 \n\t" \ 00052 \ 00053 "1: \n\t" \ 00054 \ 00055 "wldrd wr2, [%[pixels]] \n\t" \ 00056 "wldrd wr3, [%[pixels], #8] \n\t" \ 00057 "add %[pixels], %[pixels], %[line_size] \n\t" \ 00058 "pld [%[pixels]] \n\t" \ 00059 "walignr1 wr5, wr2, wr3 \n\t" \ 00060 AVG " wr6, wr4, wr5 \n\t" \ 00061 "wstrd wr6, [%[block]] \n\t" \ 00062 "add %[block], %[block], %[line_size] \n\t" \ 00063 \ 00064 "wldrd wr0, [%[pixels]] \n\t" \ 00065 "wldrd wr1, [%[pixels], #8] \n\t" \ 00066 "add %[pixels], %[pixels], %[line_size] \n\t" \ 00067 "walignr1 wr4, wr0, wr1 \n\t" \ 00068 "pld [%[pixels]] \n\t" \ 00069 AVG " wr6, wr4, wr5 \n\t" \ 00070 "wstrd wr6, [%[block]] \n\t" \ 00071 "add %[block], %[block], %[line_size] \n\t" \ 00072 \ 00073 "subs %[h], %[h], #2 \n\t" \ 00074 "bne 1b \n\t" \ 00075 : [block]"+r"(block), [pixels]"+r"(pixels), [h]"+r"(h) \ 00076 : [line_size]"r"(line_size) \ 00077 : "memory", "r12"); 00078 void put_pixels8_y2_iwmmxt(uint8_t *block, const uint8_t *pixels, const int line_size, int h) 00079 { 00080 OP("wavg2br"); 00081 } 00082 void put_no_rnd_pixels8_y2_iwmmxt(uint8_t *block, const uint8_t *pixels, const int line_size, int h) 00083 { 00084 OP("wavg2b"); 00085 } 00086 #undef OP 00087 00088 void add_pixels_clamped_iwmmxt(const DCTELEM *block, uint8_t *pixels, int line_size) 00089 { 00090 uint8_t *pixels2 = pixels + line_size; 00091 00092 __asm__ volatile ( 00093 "mov r12, #4 \n\t" 00094 "1: \n\t" 00095 "pld [%[pixels], %[line_size2]] \n\t" 00096 "pld [%[pixels2], %[line_size2]] \n\t" 00097 "wldrd wr4, [%[pixels]] \n\t" 00098 "wldrd wr5, [%[pixels2]] \n\t" 00099 "pld [%[block], #32] \n\t" 00100 "wunpckelub wr6, wr4 \n\t" 00101 "wldrd wr0, [%[block]] \n\t" 00102 "wunpckehub wr7, wr4 \n\t" 00103 "wldrd wr1, [%[block], #8] \n\t" 00104 "wunpckelub wr8, wr5 \n\t" 00105 "wldrd wr2, [%[block], #16] \n\t" 00106 "wunpckehub wr9, wr5 \n\t" 00107 "wldrd wr3, [%[block], #24] \n\t" 00108 "add %[block], %[block], #32 \n\t" 00109 "waddhss wr10, wr0, wr6 \n\t" 00110 "waddhss wr11, wr1, wr7 \n\t" 00111 "waddhss wr12, wr2, wr8 \n\t" 00112 "waddhss wr13, wr3, wr9 \n\t" 00113 "wpackhus wr14, wr10, wr11 \n\t" 00114 "wpackhus wr15, wr12, wr13 \n\t" 00115 "wstrd wr14, [%[pixels]] \n\t" 00116 "add %[pixels], %[pixels], %[line_size2] \n\t" 00117 "subs r12, r12, #1 \n\t" 00118 "wstrd wr15, [%[pixels2]] \n\t" 00119 "add %[pixels2], %[pixels2], %[line_size2] \n\t" 00120 "bne 1b \n\t" 00121 : [block]"+r"(block), [pixels]"+r"(pixels), [pixels2]"+r"(pixels2) 00122 : [line_size2]"r"(line_size << 1) 00123 : "cc", "memory", "r12"); 00124 } 00125 00126 static void clear_blocks_iwmmxt(DCTELEM *blocks) 00127 { 00128 __asm__ volatile( 00129 "wzero wr0 \n\t" 00130 "mov r1, #(128 * 6 / 32) \n\t" 00131 "1: \n\t" 00132 "wstrd wr0, [%0] \n\t" 00133 "wstrd wr0, [%0, #8] \n\t" 00134 "wstrd wr0, [%0, #16] \n\t" 00135 "wstrd wr0, [%0, #24] \n\t" 00136 "subs r1, r1, #1 \n\t" 00137 "add %0, %0, #32 \n\t" 00138 "bne 1b \n\t" 00139 : "+r"(blocks) 00140 : 00141 : "r1" 00142 ); 00143 } 00144 00145 static void nop(uint8_t *block, const uint8_t *pixels, int line_size, int h) 00146 { 00147 return; 00148 } 00149 00150 /* A run time test is not simple. If this file is compiled in 00151 * then we should install the functions 00152 */ 00153 int mm_flags = FF_MM_IWMMXT; /* multimedia extension flags */ 00154 00155 void ff_dsputil_init_iwmmxt(DSPContext* c, AVCodecContext *avctx) 00156 { 00157 if (avctx->dsp_mask) { 00158 if (avctx->dsp_mask & FF_MM_FORCE) 00159 mm_flags |= (avctx->dsp_mask & 0xffff); 00160 else 00161 mm_flags &= ~(avctx->dsp_mask & 0xffff); 00162 } 00163 00164 if (!(mm_flags & FF_MM_IWMMXT)) return; 00165 00166 c->add_pixels_clamped = add_pixels_clamped_iwmmxt; 00167 00168 c->clear_blocks = clear_blocks_iwmmxt; 00169 00170 c->put_pixels_tab[0][0] = put_pixels16_iwmmxt; 00171 c->put_pixels_tab[0][1] = put_pixels16_x2_iwmmxt; 00172 c->put_pixels_tab[0][2] = put_pixels16_y2_iwmmxt; 00173 c->put_pixels_tab[0][3] = put_pixels16_xy2_iwmmxt; 00174 c->put_no_rnd_pixels_tab[0][0] = put_pixels16_iwmmxt; 00175 c->put_no_rnd_pixels_tab[0][1] = put_no_rnd_pixels16_x2_iwmmxt; 00176 c->put_no_rnd_pixels_tab[0][2] = put_no_rnd_pixels16_y2_iwmmxt; 00177 c->put_no_rnd_pixels_tab[0][3] = put_no_rnd_pixels16_xy2_iwmmxt; 00178 00179 c->put_pixels_tab[1][0] = put_pixels8_iwmmxt; 00180 c->put_pixels_tab[1][1] = put_pixels8_x2_iwmmxt; 00181 c->put_pixels_tab[1][2] = put_pixels8_y2_iwmmxt; 00182 c->put_pixels_tab[1][3] = put_pixels8_xy2_iwmmxt; 00183 c->put_no_rnd_pixels_tab[1][0] = put_pixels8_iwmmxt; 00184 c->put_no_rnd_pixels_tab[1][1] = put_no_rnd_pixels8_x2_iwmmxt; 00185 c->put_no_rnd_pixels_tab[1][2] = put_no_rnd_pixels8_y2_iwmmxt; 00186 c->put_no_rnd_pixels_tab[1][3] = put_no_rnd_pixels8_xy2_iwmmxt; 00187 00188 c->avg_pixels_tab[0][0] = avg_pixels16_iwmmxt; 00189 c->avg_pixels_tab[0][1] = avg_pixels16_x2_iwmmxt; 00190 c->avg_pixels_tab[0][2] = avg_pixels16_y2_iwmmxt; 00191 c->avg_pixels_tab[0][3] = avg_pixels16_xy2_iwmmxt; 00192 c->avg_no_rnd_pixels_tab[0][0] = avg_pixels16_iwmmxt; 00193 c->avg_no_rnd_pixels_tab[0][1] = avg_no_rnd_pixels16_x2_iwmmxt; 00194 c->avg_no_rnd_pixels_tab[0][2] = avg_no_rnd_pixels16_y2_iwmmxt; 00195 c->avg_no_rnd_pixels_tab[0][3] = avg_no_rnd_pixels16_xy2_iwmmxt; 00196 00197 c->avg_pixels_tab[1][0] = avg_pixels8_iwmmxt; 00198 c->avg_pixels_tab[1][1] = avg_pixels8_x2_iwmmxt; 00199 c->avg_pixels_tab[1][2] = avg_pixels8_y2_iwmmxt; 00200 c->avg_pixels_tab[1][3] = avg_pixels8_xy2_iwmmxt; 00201 c->avg_no_rnd_pixels_tab[1][0] = avg_no_rnd_pixels8_iwmmxt; 00202 c->avg_no_rnd_pixels_tab[1][1] = avg_no_rnd_pixels8_x2_iwmmxt; 00203 c->avg_no_rnd_pixels_tab[1][2] = avg_no_rnd_pixels8_y2_iwmmxt; 00204 c->avg_no_rnd_pixels_tab[1][3] = avg_no_rnd_pixels8_xy2_iwmmxt; 00205 }