comparison vp8dsp.c @ 11950:56aba5a9761c libavcodec

Make VP8 DSP functions take two strides This isn't useful for the C functions, but will allow re-using H and V functions for HV functions without adding separate H and V wrappers.
author darkshikari
date Fri, 25 Jun 2010 18:14:07 +0000
parents f2007d7c3f1d
children 4aae15516d2c
comparison
equal deleted inserted replaced
11949:778bdafd5496 11950:56aba5a9761c
248 { 0, 6, 50, 93, 9, 0 }, 248 { 0, 6, 50, 93, 9, 0 },
249 { 1, 8, 36, 108, 11, 2 }, 249 { 1, 8, 36, 108, 11, 2 },
250 { 0, 1, 12, 123, 6, 0 }, 250 { 0, 1, 12, 123, 6, 0 },
251 }; 251 };
252 252
253 #define PUT_PIXELS(WIDTH) \
254 static void put_vp8_pixels ## WIDTH ##_c(uint8_t *dst, int dststride, uint8_t *src, int srcstride, int h, int x, int y) { \
255 for (int y = 0; y < h; y++, dst+= dststride, src+= srcstride) { \
256 memcpy(dst, src, WIDTH); \
257 } \
258 }
259
260 PUT_PIXELS(16)
261 PUT_PIXELS(8)
262 PUT_PIXELS(4)
253 263
254 #define FILTER_6TAP(src, F, stride) \ 264 #define FILTER_6TAP(src, F, stride) \
255 av_clip_uint8((F[2]*src[x+0*stride] - F[1]*src[x-1*stride] + F[0]*src[x-2*stride] + \ 265 av_clip_uint8((F[2]*src[x+0*stride] - F[1]*src[x-1*stride] + F[0]*src[x-2*stride] + \
256 F[3]*src[x+1*stride] - F[4]*src[x+2*stride] + F[5]*src[x+3*stride] + 64) >> 7) 266 F[3]*src[x+1*stride] - F[4]*src[x+2*stride] + F[5]*src[x+3*stride] + 64) >> 7)
257 267
258 #define FILTER_4TAP(src, F, stride) \ 268 #define FILTER_4TAP(src, F, stride) \
259 av_clip_uint8((F[2]*src[x+0*stride] - F[1]*src[x-1*stride] + \ 269 av_clip_uint8((F[2]*src[x+0*stride] - F[1]*src[x-1*stride] + \
260 F[3]*src[x+1*stride] - F[4]*src[x+2*stride] + 64) >> 7) 270 F[3]*src[x+1*stride] - F[4]*src[x+2*stride] + 64) >> 7)
261 271
262 #define VP8_EPEL_H(SIZE, FILTER, FILTERNAME) \ 272 #define VP8_EPEL_H(SIZE, FILTER, FILTERNAME) \
263 static void put_vp8_epel ## SIZE ## _ ## FILTERNAME ## _c(uint8_t *dst, uint8_t *src, int stride, int h, int mx, int my) \ 273 static void put_vp8_epel ## SIZE ## _ ## FILTERNAME ## _c(uint8_t *dst, int dststride, uint8_t *src, int srcstride, int h, int mx, int my) \
264 { \ 274 { \
265 const uint8_t *filter = subpel_filters[mx-1]; \ 275 const uint8_t *filter = subpel_filters[mx-1]; \
266 int x, y; \ 276 int x, y; \
267 \ 277 \
268 for (y = 0; y < h; y++) { \ 278 for (y = 0; y < h; y++) { \
269 for (x = 0; x < SIZE; x++) \ 279 for (x = 0; x < SIZE; x++) \
270 dst[x] = FILTER(src, filter, 1); \ 280 dst[x] = FILTER(src, filter, 1); \
271 dst += stride; \ 281 dst += dststride; \
272 src += stride; \ 282 src += srcstride; \
273 } \ 283 } \
274 } 284 }
275 #define VP8_EPEL_V(SIZE, FILTER, FILTERNAME) \ 285 #define VP8_EPEL_V(SIZE, FILTER, FILTERNAME) \
276 static void put_vp8_epel ## SIZE ## _ ## FILTERNAME ## _c(uint8_t *dst, uint8_t *src, int stride, int h, int mx, int my) \ 286 static void put_vp8_epel ## SIZE ## _ ## FILTERNAME ## _c(uint8_t *dst, int dststride, uint8_t *src, int srcstride, int h, int mx, int my) \
277 { \ 287 { \
278 const uint8_t *filter = subpel_filters[my-1]; \ 288 const uint8_t *filter = subpel_filters[my-1]; \
279 int x, y; \ 289 int x, y; \
280 \ 290 \
281 for (y = 0; y < h; y++) { \ 291 for (y = 0; y < h; y++) { \
282 for (x = 0; x < SIZE; x++) \ 292 for (x = 0; x < SIZE; x++) \
283 dst[x] = FILTER(src, filter, stride); \ 293 dst[x] = FILTER(src, filter, srcstride); \
284 dst += stride; \ 294 dst += dststride; \
285 src += stride; \ 295 src += srcstride; \
286 } \ 296 } \
287 } 297 }
288 #define VP8_EPEL_HV(SIZE, FILTERX, FILTERY, FILTERNAME) \ 298 #define VP8_EPEL_HV(SIZE, FILTERX, FILTERY, FILTERNAME) \
289 static void put_vp8_epel ## SIZE ## _ ## FILTERNAME ## _c(uint8_t *dst, uint8_t *src, int stride, int h, int mx, int my) \ 299 static void put_vp8_epel ## SIZE ## _ ## FILTERNAME ## _c(uint8_t *dst, int dststride, uint8_t *src, int srcstride, int h, int mx, int my) \
290 { \ 300 { \
291 const uint8_t *filter = subpel_filters[mx-1]; \ 301 const uint8_t *filter = subpel_filters[mx-1]; \
292 int x, y; \ 302 int x, y; \
293 uint8_t tmp_array[(2*SIZE+5)*SIZE]; \ 303 uint8_t tmp_array[(2*SIZE+5)*SIZE]; \
294 uint8_t *tmp = tmp_array; \ 304 uint8_t *tmp = tmp_array; \
295 src -= 2*stride; \ 305 src -= 2*srcstride; \
296 \ 306 \
297 for (y = 0; y < h+5; y++) { \ 307 for (y = 0; y < h+5; y++) { \
298 for (x = 0; x < SIZE; x++) \ 308 for (x = 0; x < SIZE; x++) \
299 tmp[x] = FILTERX(src, filter, 1); \ 309 tmp[x] = FILTERX(src, filter, 1); \
300 tmp += SIZE; \ 310 tmp += SIZE; \
301 src += stride; \ 311 src += srcstride; \
302 } \ 312 } \
303 \ 313 \
304 tmp = tmp_array + 2*SIZE; \ 314 tmp = tmp_array + 2*SIZE; \
305 filter = subpel_filters[my-1]; \ 315 filter = subpel_filters[my-1]; \
306 \ 316 \
307 for (y = 0; y < h; y++) { \ 317 for (y = 0; y < h; y++) { \
308 for (x = 0; x < SIZE; x++) \ 318 for (x = 0; x < SIZE; x++) \
309 dst[x] = FILTERY(tmp, filter, SIZE); \ 319 dst[x] = FILTERY(tmp, filter, SIZE); \
310 dst += stride; \ 320 dst += dststride; \
311 tmp += SIZE; \ 321 tmp += SIZE; \
312 } \ 322 } \
313 } 323 }
314 324
315 VP8_EPEL_H(16, FILTER_4TAP, h4) 325 VP8_EPEL_H(16, FILTER_4TAP, h4)
336 VP8_EPEL_HV(16, FILTER_6TAP, FILTER_6TAP, h6v6) 346 VP8_EPEL_HV(16, FILTER_6TAP, FILTER_6TAP, h6v6)
337 VP8_EPEL_HV(8, FILTER_6TAP, FILTER_6TAP, h6v6) 347 VP8_EPEL_HV(8, FILTER_6TAP, FILTER_6TAP, h6v6)
338 VP8_EPEL_HV(4, FILTER_6TAP, FILTER_6TAP, h6v6) 348 VP8_EPEL_HV(4, FILTER_6TAP, FILTER_6TAP, h6v6)
339 349
340 #define VP8_MC_FUNC(IDX, SIZE) \ 350 #define VP8_MC_FUNC(IDX, SIZE) \
341 dsp->put_vp8_epel_pixels_tab[IDX][0][0] = ff_put_vp8_pixels ## SIZE ## _c; \ 351 dsp->put_vp8_epel_pixels_tab[IDX][0][0] = put_vp8_pixels ## SIZE ## _c; \
342 dsp->put_vp8_epel_pixels_tab[IDX][0][1] = put_vp8_epel ## SIZE ## _h4_c; \ 352 dsp->put_vp8_epel_pixels_tab[IDX][0][1] = put_vp8_epel ## SIZE ## _h4_c; \
343 dsp->put_vp8_epel_pixels_tab[IDX][0][2] = put_vp8_epel ## SIZE ## _h6_c; \ 353 dsp->put_vp8_epel_pixels_tab[IDX][0][2] = put_vp8_epel ## SIZE ## _h6_c; \
344 dsp->put_vp8_epel_pixels_tab[IDX][1][0] = put_vp8_epel ## SIZE ## _v4_c; \ 354 dsp->put_vp8_epel_pixels_tab[IDX][1][0] = put_vp8_epel ## SIZE ## _v4_c; \
345 dsp->put_vp8_epel_pixels_tab[IDX][1][1] = put_vp8_epel ## SIZE ## _h4v4_c; \ 355 dsp->put_vp8_epel_pixels_tab[IDX][1][1] = put_vp8_epel ## SIZE ## _h4v4_c; \
346 dsp->put_vp8_epel_pixels_tab[IDX][1][2] = put_vp8_epel ## SIZE ## _h6v4_c; \ 356 dsp->put_vp8_epel_pixels_tab[IDX][1][2] = put_vp8_epel ## SIZE ## _h6v4_c; \