Parent Directory
| 
Revision Log
Revision 1192 - (view) (download)
| 1 : | edgomez | 1192 | ;/**************************************************************************** |
| 2 : | ; * | ||
| 3 : | ; * XVID MPEG-4 VIDEO CODEC | ||
| 4 : | ; * - MMX and XMM YV12->YV12 conversion - | ||
| 5 : | ; * | ||
| 6 : | ; * Copyright(C) 2001 Michael Militzer <isibaar@xvid.org> | ||
| 7 : | ; * | ||
| 8 : | ; * This program is free software; you can redistribute it and/or modify it | ||
| 9 : | ; * under the terms of the GNU General Public License as published by | ||
| 10 : | ; * the Free Software Foundation; either version 2 of the License, or | ||
| 11 : | ; * (at your option) any later version. | ||
| 12 : | ; * | ||
| 13 : | ; * This program is distributed in the hope that it will be useful, | ||
| 14 : | ; * but WITHOUT ANY WARRANTY; without even the implied warranty of | ||
| 15 : | ; * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | ||
| 16 : | ; * GNU General Public License for more details. | ||
| 17 : | ; * | ||
| 18 : | ; * You should have received a copy of the GNU General Public License | ||
| 19 : | ; * along with this program; if not, write to the Free Software | ||
| 20 : | ; * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA | ||
| 21 : | ; * | ||
| 22 : | ; * $Id: colorspace_yuv_mmx.asm,v 1.2.2.1 2003-10-28 22:23:03 edgomez Exp $ | ||
| 23 : | ; * | ||
| 24 : | ; ***************************************************************************/ | ||
| 25 : | edgomez | 851 | |
| 26 : | BITS 32 | ||
| 27 : | |||
| 28 : | edgomez | 1192 | %macro cglobal 1 |
| 29 : | edgomez | 851 | %ifdef PREFIX |
| 30 : | edgomez | 1192 | global _%1 |
| 31 : | edgomez | 851 | %define %1 _%1 |
| 32 : | %else | ||
| 33 : | global %1 | ||
| 34 : | %endif | ||
| 35 : | %endmacro | ||
| 36 : | |||
| 37 : | edgomez | 1192 | ;============================================================================= |
| 38 : | ; Helper macros | ||
| 39 : | ;============================================================================= | ||
| 40 : | edgomez | 851 | |
| 41 : | ;------------------------------------------------------------------------------ | ||
| 42 : | ; PLANE_COPY ( DST, DST_DIF, SRC, SRC_DIF, WIDTH, HEIGHT, OPT ) | ||
| 43 : | ; DST dst buffer | ||
| 44 : | ; DST_DIF dst stride difference (e.g. stride - width) | ||
| 45 : | ; SRC src destination buffer | ||
| 46 : | ; SRC_DIF src stride difference (e.g. stride - width) | ||
| 47 : | ; WIDTH width | ||
| 48 : | ; HEIGHT height | ||
| 49 : | ; OPT 0=plain mmx, 1=xmm | ||
| 50 : | ;------------------------------------------------------------------------------ | ||
| 51 : | edgomez | 1192 | |
| 52 : | edgomez | 851 | %macro PLANE_COPY 7 |
| 53 : | %define DST %1 | ||
| 54 : | %define DST_DIF %2 | ||
| 55 : | %define SRC %3 | ||
| 56 : | %define SRC_DIF %4 | ||
| 57 : | %define WIDTH %5 | ||
| 58 : | %define HEIGHT %6 | ||
| 59 : | %define OPT %7 | ||
| 60 : | |||
| 61 : | edgomez | 1192 | mov eax, WIDTH |
| 62 : | mov ebp, HEIGHT ; $ebp$ = height | ||
| 63 : | mov esi, SRC | ||
| 64 : | mov edi, DST | ||
| 65 : | edgomez | 851 | |
| 66 : | edgomez | 1192 | mov ebx, eax |
| 67 : | shr eax, 6 ; $eax$ = width / 64 | ||
| 68 : | and ebx, 63 ; remainder = width % 64 | ||
| 69 : | mov edx, ebx | ||
| 70 : | shr ebx, 4 ; $ebx$ = remainder / 16 | ||
| 71 : | and edx, 15 ; $edx$ = remainder % 16 | ||
| 72 : | edgomez | 851 | |
| 73 : | %%loop64_start | ||
| 74 : | edgomez | 1192 | or eax, eax |
| 75 : | jz %%loop16_start | ||
| 76 : | mov ecx, eax ; width64 | ||
| 77 : | edgomez | 851 | %%loop64: |
| 78 : | edgomez | 1192 | %if OPT == 1 ; xmm |
| 79 : | prefetchnta [esi + 64] ; non temporal prefetch | ||
| 80 : | prefetchnta [esi + 96] | ||
| 81 : | edgomez | 851 | %endif |
| 82 : | edgomez | 1192 | movq mm1, [esi] ; read from src |
| 83 : | movq mm2, [esi + 8] | ||
| 84 : | movq mm3, [esi + 16] | ||
| 85 : | movq mm4, [esi + 24] | ||
| 86 : | movq mm5, [esi + 32] | ||
| 87 : | movq mm6, [esi + 40] | ||
| 88 : | movq mm7, [esi + 48] | ||
| 89 : | movq mm0, [esi + 56] | ||
| 90 : | edgomez | 851 | |
| 91 : | edgomez | 1192 | %if OPT == 0 ; plain mmx |
| 92 : | movq [edi], mm1 ; write to y_out | ||
| 93 : | movq [edi + 8], mm2 | ||
| 94 : | movq [edi + 16], mm3 | ||
| 95 : | movq [edi + 24], mm4 | ||
| 96 : | movq [edi + 32], mm5 | ||
| 97 : | movq [edi + 40], mm6 | ||
| 98 : | movq [edi + 48], mm7 | ||
| 99 : | movq [edi + 56], mm0 | ||
| 100 : | edgomez | 851 | %else |
| 101 : | edgomez | 1192 | movntq [edi], mm1 ; write to y_out |
| 102 : | movntq [edi + 8], mm2 | ||
| 103 : | movntq [edi + 16], mm3 | ||
| 104 : | movntq [edi + 24], mm4 | ||
| 105 : | movntq [edi + 32], mm5 | ||
| 106 : | movntq [edi + 40], mm6 | ||
| 107 : | movntq [edi + 48], mm7 | ||
| 108 : | movntq [edi + 56], mm0 | ||
| 109 : | edgomez | 851 | %endif |
| 110 : | |||
| 111 : | edgomez | 1192 | add esi, 64 |
| 112 : | add edi, 64 | ||
| 113 : | dec ecx | ||
| 114 : | jnz %%loop64 | ||
| 115 : | edgomez | 851 | |
| 116 : | |||
| 117 : | %%loop16_start | ||
| 118 : | edgomez | 1192 | or ebx, ebx |
| 119 : | jz %%loop1_start | ||
| 120 : | mov ecx, ebx ; width16 | ||
| 121 : | edgomez | 851 | %%loop16: |
| 122 : | edgomez | 1192 | movq mm1, [esi] |
| 123 : | movq mm2, [esi + 8] | ||
| 124 : | %if OPT == 0 ; plain mmx | ||
| 125 : | movq [edi], mm1 | ||
| 126 : | movq [edi + 8], mm2 | ||
| 127 : | edgomez | 851 | %else |
| 128 : | edgomez | 1192 | movntq [edi], mm1 |
| 129 : | movntq [edi + 8], mm2 | ||
| 130 : | edgomez | 851 | %endif |
| 131 : | |||
| 132 : | edgomez | 1192 | add esi, 16 |
| 133 : | add edi, 16 | ||
| 134 : | dec ecx | ||
| 135 : | jnz %%loop16 | ||
| 136 : | edgomez | 851 | |
| 137 : | |||
| 138 : | %%loop1_start | ||
| 139 : | edgomez | 1192 | mov ecx, edx |
| 140 : | rep movsb | ||
| 141 : | edgomez | 851 | |
| 142 : | edgomez | 1192 | add esi, SRC_DIF |
| 143 : | add edi, DST_DIF | ||
| 144 : | dec ebp | ||
| 145 : | jnz near %%loop64_start | ||
| 146 : | edgomez | 851 | %endmacro |
| 147 : | |||
| 148 : | ;------------------------------------------------------------------------------ | ||
| 149 : | ; MAKE_YV12_TO_YV12( NAME, OPT ) | ||
| 150 : | ; NAME function name | ||
| 151 : | ; OPT 0=plain mmx, 1=xmm | ||
| 152 : | ; | ||
| 153 : | edgomez | 1192 | ; yv12_to_yv12_mmx(uint8_t * y_dst, uint8_t * u_dst, uint8_t * v_dst, |
| 154 : | edgomez | 851 | ; int y_dst_stride, int uv_dst_stride, |
| 155 : | edgomez | 1192 | ; uint8_t * y_src, uint8_t * u_src, uint8_t * v_src, |
| 156 : | edgomez | 851 | ; int y_src_stride, int uv_src_stride, |
| 157 : | ; int width, int height, int vflip) | ||
| 158 : | ;------------------------------------------------------------------------------ | ||
| 159 : | %macro MAKE_YV12_TO_YV12 2 | ||
| 160 : | %define NAME %1 | ||
| 161 : | %define OPT %2 | ||
| 162 : | edgomez | 1192 | ALIGN 16 |
| 163 : | edgomez | 851 | cglobal NAME |
| 164 : | edgomez | 1192 | NAME: |
| 165 : | edgomez | 851 | %define pushsize 16 |
| 166 : | %define localsize 24 | ||
| 167 : | |||
| 168 : | %define vflip esp + localsize + pushsize + 52 | ||
| 169 : | %define height esp + localsize + pushsize + 48 | ||
| 170 : | %define width esp + localsize + pushsize + 44 | ||
| 171 : | %define uv_src_stride esp + localsize + pushsize + 40 | ||
| 172 : | %define y_src_stride esp + localsize + pushsize + 36 | ||
| 173 : | %define v_src esp + localsize + pushsize + 32 | ||
| 174 : | %define u_src esp + localsize + pushsize + 28 | ||
| 175 : | %define y_src esp + localsize + pushsize + 24 | ||
| 176 : | %define uv_dst_stride esp + localsize + pushsize + 20 | ||
| 177 : | %define y_dst_stride esp + localsize + pushsize + 16 | ||
| 178 : | %define v_dst esp + localsize + pushsize + 12 | ||
| 179 : | %define u_dst esp + localsize + pushsize + 8 | ||
| 180 : | %define y_dst esp + localsize + pushsize + 4 | ||
| 181 : | %define _ip esp + localsize + pushsize + 0 | ||
| 182 : | |||
| 183 : | edgomez | 1192 | push ebx ; esp + localsize + 16 |
| 184 : | push esi ; esp + localsize + 8 | ||
| 185 : | push edi ; esp + localsize + 4 | ||
| 186 : | push ebp ; esp + localsize + 0 | ||
| 187 : | edgomez | 851 | |
| 188 : | %define width2 esp + localsize - 4 | ||
| 189 : | %define height2 esp + localsize - 8 | ||
| 190 : | %define y_src_dif esp + localsize - 12 | ||
| 191 : | %define y_dst_dif esp + localsize - 16 | ||
| 192 : | %define uv_src_dif esp + localsize - 20 | ||
| 193 : | %define uv_dst_dif esp + localsize - 24 | ||
| 194 : | |||
| 195 : | edgomez | 1192 | sub esp, localsize |
| 196 : | edgomez | 851 | |
| 197 : | edgomez | 1192 | mov eax, [width] |
| 198 : | mov ebx, [height] | ||
| 199 : | shr eax, 1 ; calculate widht/2, heigh/2 | ||
| 200 : | shr ebx, 1 | ||
| 201 : | mov [width2], eax | ||
| 202 : | mov [height2], ebx | ||
| 203 : | edgomez | 851 | |
| 204 : | edgomez | 1192 | mov ebp, [vflip] |
| 205 : | or ebp, ebp | ||
| 206 : | jz near .dont_flip | ||
| 207 : | edgomez | 851 | |
| 208 : | ; flipping support | ||
| 209 : | edgomez | 1192 | mov eax, [height] |
| 210 : | mov esi, [y_src] | ||
| 211 : | mov edx, [y_src_stride] | ||
| 212 : | push edx | ||
| 213 : | mul edx | ||
| 214 : | pop edx | ||
| 215 : | add esi, eax ; y_src += (height-1) * y_src_stride | ||
| 216 : | neg edx | ||
| 217 : | mov [y_src], esi | ||
| 218 : | mov [y_src_stride], edx ; y_src_stride = -y_src_stride | ||
| 219 : | edgomez | 851 | |
| 220 : | edgomez | 1192 | mov eax, [height2] |
| 221 : | mov esi, [u_src] | ||
| 222 : | mov edi, [v_src] | ||
| 223 : | mov edx, [uv_src_stride] | ||
| 224 : | sub eax, 1 ; ebp = height2 - 1 | ||
| 225 : | push edx | ||
| 226 : | mul edx | ||
| 227 : | pop edx | ||
| 228 : | add esi, eax ; u_src += (height2-1) * uv_src_stride | ||
| 229 : | add edi, eax ; v_src += (height2-1) * uv_src_stride | ||
| 230 : | neg edx | ||
| 231 : | mov [u_src], esi | ||
| 232 : | mov [v_src], edi | ||
| 233 : | mov [uv_src_stride], edx ; uv_src_stride = -uv_src_stride | ||
| 234 : | edgomez | 851 | |
| 235 : | .dont_flip | ||
| 236 : | |||
| 237 : | edgomez | 1192 | mov eax, [y_src_stride] |
| 238 : | mov ebx, [y_dst_stride] | ||
| 239 : | mov ecx, [uv_src_stride] | ||
| 240 : | mov edx, [uv_dst_stride] | ||
| 241 : | sub eax, [width] | ||
| 242 : | sub ebx, [width] | ||
| 243 : | sub ecx, [width2] | ||
| 244 : | sub edx, [width2] | ||
| 245 : | mov [y_src_dif], eax ; y_src_dif = y_src_stride - width | ||
| 246 : | mov [y_dst_dif], ebx ; y_dst_dif = y_dst_stride - width | ||
| 247 : | mov [uv_src_dif], ecx ; uv_src_dif = uv_src_stride - width2 | ||
| 248 : | mov [uv_dst_dif], edx ; uv_dst_dif = uv_dst_stride - width2 | ||
| 249 : | edgomez | 851 | |
| 250 : | edgomez | 1192 | PLANE_COPY [y_dst], [y_dst_dif], [y_src], [y_src_dif], [width], [height], OPT |
| 251 : | PLANE_COPY [u_dst], [uv_dst_dif], [u_src], [uv_src_dif], [width2], [height2], OPT | ||
| 252 : | PLANE_COPY [v_dst], [uv_dst_dif], [v_src], [uv_src_dif], [width2], [height2], OPT | ||
| 253 : | edgomez | 851 | |
| 254 : | edgomez | 1192 | add esp, localsize |
| 255 : | pop ebp | ||
| 256 : | pop edi | ||
| 257 : | pop esi | ||
| 258 : | pop ebx | ||
| 259 : | edgomez | 851 | |
| 260 : | edgomez | 1192 | ret |
| 261 : | edgomez | 851 | %endmacro |
| 262 : | |||
| 263 : | edgomez | 1192 | ;============================================================================= |
| 264 : | ; Code | ||
| 265 : | ;============================================================================= | ||
| 266 : | edgomez | 851 | |
| 267 : | edgomez | 1192 | SECTION .text |
| 268 : | |||
| 269 : | edgomez | 851 | MAKE_YV12_TO_YV12 yv12_to_yv12_mmx, 0 |
| 270 : | edgomez | 1192 | |
| 271 : | edgomez | 851 | MAKE_YV12_TO_YV12 yv12_to_yv12_xmm, 1 |
| No admin address has been configured | ViewVC Help |
| Powered by ViewVC 1.0.4 |