1 |
|
#ifndef _INTERPOLATE8X8_H_ |
2 |
|
#define _INTERPOLATE8X8_H_ |
3 |
|
|
4 |
#include "../utils/mem_transfer.h" |
#include "../utils/mem_transfer.h" |
5 |
|
|
6 |
typedef void (INTERPOLATE8X8) (uint8_t * const dst, |
typedef void (INTERPOLATE8X8) (uint8_t * const dst, |
9 |
const uint32_t rounding); |
const uint32_t rounding); |
10 |
typedef INTERPOLATE8X8 *INTERPOLATE8X8_PTR; |
typedef INTERPOLATE8X8 *INTERPOLATE8X8_PTR; |
11 |
|
|
12 |
|
typedef void (INTERPOLATE8X8_AVG2) (uint8_t *dst, |
13 |
|
const uint8_t *src1, |
14 |
|
const uint8_t *src2, |
15 |
|
const uint32_t stride, |
16 |
|
const uint32_t rounding, |
17 |
|
const uint32_t height); |
18 |
|
typedef INTERPOLATE8X8_AVG2 *INTERPOLATE8X8_AVG2_PTR; |
19 |
|
|
20 |
|
typedef void (INTERPOLATE8X8_AVG4) (uint8_t *dst, |
21 |
|
const uint8_t *src1, |
22 |
|
const uint8_t *src2, |
23 |
|
const uint8_t *src3, |
24 |
|
const uint8_t *src4, |
25 |
|
const uint32_t stride, |
26 |
|
const uint32_t rounding); |
27 |
|
typedef INTERPOLATE8X8_AVG4 *INTERPOLATE8X8_AVG4_PTR; |
28 |
|
|
29 |
|
typedef void (INTERPOLATE_LOWPASS) (uint8_t *dst, |
30 |
|
uint8_t *src, |
31 |
|
int32_t stride, |
32 |
|
int32_t rounding); |
33 |
|
|
34 |
|
typedef INTERPOLATE_LOWPASS *INTERPOLATE_LOWPASS_PTR; |
35 |
|
|
36 |
|
typedef void (INTERPOLATE_LOWPASS_HV) (uint8_t *dst1, |
37 |
|
uint8_t *dst2, |
38 |
|
uint8_t *src, |
39 |
|
int32_t stride, |
40 |
|
int32_t rounding); |
41 |
|
|
42 |
|
typedef INTERPOLATE_LOWPASS_HV *INTERPOLATE_LOWPASS_HV_PTR; |
43 |
|
|
44 |
|
typedef void (INTERPOLATE8X8_6TAP_LOWPASS) (uint8_t *dst, |
45 |
|
uint8_t *src, |
46 |
|
int32_t stride, |
47 |
|
int32_t rounding); |
48 |
|
|
49 |
|
typedef INTERPOLATE8X8_6TAP_LOWPASS *INTERPOLATE8X8_6TAP_LOWPASS_PTR; |
50 |
|
|
51 |
extern INTERPOLATE8X8_PTR interpolate8x8_halfpel_h; |
extern INTERPOLATE8X8_PTR interpolate8x8_halfpel_h; |
52 |
extern INTERPOLATE8X8_PTR interpolate8x8_halfpel_v; |
extern INTERPOLATE8X8_PTR interpolate8x8_halfpel_v; |
53 |
extern INTERPOLATE8X8_PTR interpolate8x8_halfpel_hv; |
extern INTERPOLATE8X8_PTR interpolate8x8_halfpel_hv; |
54 |
|
|
55 |
|
extern INTERPOLATE8X8_AVG2_PTR interpolate8x8_avg2; |
56 |
|
extern INTERPOLATE8X8_AVG4_PTR interpolate8x8_avg4; |
57 |
|
|
58 |
|
extern INTERPOLATE_LOWPASS_PTR interpolate8x8_lowpass_h; |
59 |
|
extern INTERPOLATE_LOWPASS_PTR interpolate8x8_lowpass_v; |
60 |
|
|
61 |
|
extern INTERPOLATE_LOWPASS_PTR interpolate16x16_lowpass_h; |
62 |
|
extern INTERPOLATE_LOWPASS_PTR interpolate16x16_lowpass_v; |
63 |
|
|
64 |
|
extern INTERPOLATE_LOWPASS_HV_PTR interpolate8x8_lowpass_hv; |
65 |
|
extern INTERPOLATE_LOWPASS_HV_PTR interpolate16x16_lowpass_hv; |
66 |
|
|
67 |
|
extern INTERPOLATE8X8_6TAP_LOWPASS_PTR interpolate8x8_6tap_lowpass_h; |
68 |
|
extern INTERPOLATE8X8_6TAP_LOWPASS_PTR interpolate8x8_6tap_lowpass_v; |
69 |
|
|
70 |
INTERPOLATE8X8 interpolate8x8_halfpel_h_c; |
INTERPOLATE8X8 interpolate8x8_halfpel_h_c; |
71 |
INTERPOLATE8X8 interpolate8x8_halfpel_v_c; |
INTERPOLATE8X8 interpolate8x8_halfpel_v_c; |
72 |
INTERPOLATE8X8 interpolate8x8_halfpel_hv_c; |
INTERPOLATE8X8 interpolate8x8_halfpel_hv_c; |
87 |
INTERPOLATE8X8 interpolate8x8_halfpel_v_ia64; |
INTERPOLATE8X8 interpolate8x8_halfpel_v_ia64; |
88 |
INTERPOLATE8X8 interpolate8x8_halfpel_hv_ia64; |
INTERPOLATE8X8 interpolate8x8_halfpel_hv_ia64; |
89 |
|
|
90 |
|
INTERPOLATE8X8_AVG2 interpolate8x8_avg2_c; |
91 |
|
INTERPOLATE8X8_AVG4 interpolate8x8_avg4_c; |
92 |
|
|
93 |
|
INTERPOLATE8X8_AVG2 interpolate8x8_avg2_mmx; |
94 |
|
INTERPOLATE8X8_AVG4 interpolate8x8_avg4_mmx; |
95 |
|
|
96 |
|
INTERPOLATE_LOWPASS interpolate8x8_lowpass_h_c; |
97 |
|
INTERPOLATE_LOWPASS interpolate8x8_lowpass_v_c; |
98 |
|
|
99 |
|
INTERPOLATE_LOWPASS interpolate16x16_lowpass_h_c; |
100 |
|
INTERPOLATE_LOWPASS interpolate16x16_lowpass_v_c; |
101 |
|
|
102 |
|
INTERPOLATE_LOWPASS_HV interpolate8x8_lowpass_hv_c; |
103 |
|
INTERPOLATE_LOWPASS_HV interpolate16x16_lowpass_hv_c; |
104 |
|
|
105 |
|
INTERPOLATE8X8_6TAP_LOWPASS interpolate8x8_6tap_lowpass_h_c; |
106 |
|
INTERPOLATE8X8_6TAP_LOWPASS interpolate8x8_6tap_lowpass_v_c; |
107 |
|
|
108 |
|
INTERPOLATE8X8_6TAP_LOWPASS interpolate8x8_6tap_lowpass_h_mmx; |
109 |
|
INTERPOLATE8X8_6TAP_LOWPASS interpolate8x8_6tap_lowpass_v_mmx; |
110 |
|
|
111 |
static __inline void |
static __inline void |
112 |
interpolate8x8_switch(uint8_t * const cur, |
interpolate8x8_switch(uint8_t * const cur, |
113 |
const uint8_t * const refn, |
const uint8_t * const refn, |
155 |
} |
} |
156 |
} |
} |
157 |
|
|
158 |
|
static __inline uint8_t * |
159 |
void interpolate8x8_c(uint8_t * const dst, |
interpolate8x8_switch2(uint8_t * const buffer, |
160 |
const uint8_t * const src, |
const uint8_t * const refn, |
161 |
const uint32_t x, |
const uint32_t x, |
162 |
const uint32_t y, |
const uint32_t y, |
163 |
const uint32_t stride); |
const int32_t dx, |
164 |
|
const int dy, |
165 |
|
const uint32_t stride, |
166 |
|
const uint32_t rounding) |
167 |
|
{ |
168 |
|
int32_t ddx, ddy; |
169 |
|
|
170 |
|
switch (((dx & 1) << 1) + (dy & 1)) // ((dx%2)?2:0)+((dy%2)?1:0) |
171 |
|
{ |
172 |
|
case 0: |
173 |
|
return (uint8_t *)refn + (int)((y + dy/2) * stride + x + dx/2); |
174 |
|
|
175 |
|
case 1: |
176 |
|
ddx = dx / 2; |
177 |
|
ddy = (dy - 1) / 2; |
178 |
|
interpolate8x8_halfpel_v(buffer, |
179 |
|
refn + (int)((y + ddy) * stride + x + ddx), stride, |
180 |
|
rounding); |
181 |
|
break; |
182 |
|
|
183 |
|
case 2: |
184 |
|
ddx = (dx - 1) / 2; |
185 |
|
ddy = dy / 2; |
186 |
|
interpolate8x8_halfpel_h(buffer, |
187 |
|
refn + (int)((y + ddy) * stride + x + ddx), stride, |
188 |
|
rounding); |
189 |
|
break; |
190 |
|
|
191 |
|
default: |
192 |
|
ddx = (dx - 1) / 2; |
193 |
|
ddy = (dy - 1) / 2; |
194 |
|
interpolate8x8_halfpel_hv(buffer, |
195 |
|
refn + (int)((y + ddy) * stride + x + ddx), stride, |
196 |
|
rounding); |
197 |
|
break; |
198 |
|
} |
199 |
|
return buffer; |
200 |
|
} |
201 |
|
|
202 |
|
static __inline void interpolate8x8_quarterpel(uint8_t * const cur, |
203 |
|
uint8_t * const refn, |
204 |
|
uint8_t * const refh, |
205 |
|
uint8_t * const refv, |
206 |
|
uint8_t * const refhv, |
207 |
|
const uint32_t x, const uint32_t y, |
208 |
|
const int32_t dx, const int dy, |
209 |
|
const uint32_t stride, |
210 |
|
const uint32_t rounding) |
211 |
|
{ |
212 |
|
const int32_t xRef = x*4 + dx; |
213 |
|
const int32_t yRef = y*4 + dy; |
214 |
|
|
215 |
|
uint8_t *src, *dst; |
216 |
|
uint8_t *halfpel_h, *halfpel_v, *halfpel_hv; |
217 |
|
int32_t x_int, y_int, x_frac, y_frac; |
218 |
|
|
219 |
|
x_int = xRef/4; |
220 |
|
if (xRef < 0 && xRef % 4) |
221 |
|
x_int--; |
222 |
|
|
223 |
|
x_frac = xRef - (4*x_int); |
224 |
|
|
225 |
|
y_int = yRef/4; |
226 |
|
if (yRef < 0 && yRef % 4) |
227 |
|
y_int--; |
228 |
|
|
229 |
|
y_frac = yRef - (4*y_int); |
230 |
|
|
231 |
|
src = refn + y_int * stride + x_int; |
232 |
|
halfpel_h = refh; |
233 |
|
halfpel_v = refv; |
234 |
|
halfpel_hv = refhv; |
235 |
|
|
236 |
|
dst = cur + y * stride + x; |
237 |
|
|
238 |
|
switch((y_frac << 2) | (x_frac)) { |
239 |
|
|
240 |
|
case 0: |
241 |
|
transfer8x8_copy(dst, src, stride); |
242 |
|
break; |
243 |
|
|
244 |
|
case 1: |
245 |
|
interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding); |
246 |
|
interpolate8x8_avg2(dst, src, halfpel_h, stride, rounding, 9); |
247 |
|
break; |
248 |
|
|
249 |
|
case 2: |
250 |
|
interpolate8x8_lowpass_h(dst, src, stride, rounding); |
251 |
|
break; |
252 |
|
|
253 |
|
case 3: |
254 |
|
interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding); |
255 |
|
interpolate8x8_avg2(dst, src + 1, halfpel_h, stride, rounding, 9); |
256 |
|
break; |
257 |
|
|
258 |
|
case 4: |
259 |
|
interpolate8x8_lowpass_v(halfpel_v, src, stride, rounding); |
260 |
|
interpolate8x8_avg2(dst, src, halfpel_v, stride, rounding, 9); |
261 |
|
break; |
262 |
|
|
263 |
|
case 5: |
264 |
|
interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding); |
265 |
|
interpolate8x8_avg2(halfpel_v, src, halfpel_h, stride, rounding, 9); |
266 |
|
interpolate8x8_lowpass_v(halfpel_hv, halfpel_v, stride, rounding); |
267 |
|
interpolate8x8_avg2(dst, halfpel_v, halfpel_hv, stride, rounding, 9); |
268 |
|
break; |
269 |
|
|
270 |
|
case 6: |
271 |
|
interpolate8x8_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
272 |
|
interpolate8x8_avg2(dst, halfpel_h, halfpel_hv, stride, rounding, 9); |
273 |
|
break; |
274 |
|
|
275 |
|
case 7: |
276 |
|
interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding); |
277 |
|
interpolate8x8_avg2(halfpel_v, src + 1, halfpel_h, stride, rounding, 9); |
278 |
|
interpolate8x8_lowpass_v(halfpel_hv, halfpel_v, stride, rounding); |
279 |
|
interpolate8x8_avg2(dst, halfpel_v, halfpel_hv, stride, rounding, 9); |
280 |
|
break; |
281 |
|
|
282 |
|
case 8: |
283 |
|
interpolate8x8_lowpass_v(dst, src, stride, rounding); |
284 |
|
break; |
285 |
|
|
286 |
|
case 9: |
287 |
|
interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding); |
288 |
|
interpolate8x8_avg2(halfpel_v, src, halfpel_h, stride, rounding, 9); |
289 |
|
interpolate8x8_lowpass_v(dst, halfpel_v, stride, rounding); |
290 |
|
break; |
291 |
|
|
292 |
|
case 10: |
293 |
|
interpolate8x8_lowpass_hv(dst, halfpel_h, src, stride, rounding); |
294 |
|
break; |
295 |
|
|
296 |
|
case 11: |
297 |
|
interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding); |
298 |
|
interpolate8x8_avg2(halfpel_v, src + 1, halfpel_h, stride, rounding, 9); |
299 |
|
interpolate8x8_lowpass_v(dst, halfpel_v, stride, rounding); |
300 |
|
break; |
301 |
|
|
302 |
|
case 12: |
303 |
|
interpolate8x8_lowpass_v(halfpel_v, src, stride, rounding); |
304 |
|
interpolate8x8_avg2(dst, src+stride, halfpel_v, stride, rounding, 9); |
305 |
|
break; |
306 |
|
|
307 |
|
case 13: |
308 |
|
interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding); |
309 |
|
interpolate8x8_avg2(halfpel_v, src, halfpel_h, stride, rounding, 9); |
310 |
|
interpolate8x8_lowpass_v(halfpel_hv, halfpel_v, stride, rounding); |
311 |
|
interpolate8x8_avg2(dst, halfpel_v+stride, halfpel_hv, stride, rounding, 9); |
312 |
|
break; |
313 |
|
|
314 |
|
case 14: |
315 |
|
interpolate8x8_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
316 |
|
interpolate8x8_avg2(dst, halfpel_h+stride, halfpel_hv, stride, rounding, 9); |
317 |
|
break; |
318 |
|
|
319 |
|
case 15: |
320 |
|
interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding); |
321 |
|
interpolate8x8_avg2(halfpel_v, src + 1, halfpel_h, stride, rounding, 9); |
322 |
|
interpolate8x8_lowpass_v(halfpel_hv, halfpel_v, stride, rounding); |
323 |
|
interpolate8x8_avg2(dst, halfpel_hv, halfpel_v + stride, stride, rounding, 9); |
324 |
|
break; |
325 |
|
} |
326 |
|
} |
327 |
|
|
328 |
|
static __inline void interpolate16x16_quarterpel(uint8_t * const cur, |
329 |
|
uint8_t * const refn, |
330 |
|
uint8_t * const refh, |
331 |
|
uint8_t * const refv, |
332 |
|
uint8_t * const refhv, |
333 |
|
const uint32_t x, const uint32_t y, |
334 |
|
const int32_t dx, const int dy, |
335 |
|
const uint32_t stride, |
336 |
|
const uint32_t rounding) |
337 |
|
{ |
338 |
|
const int32_t xRef = x*4 + dx; |
339 |
|
const int32_t yRef = y*4 + dy; |
340 |
|
|
341 |
|
uint8_t *src, *dst; |
342 |
|
uint8_t *halfpel_h, *halfpel_v, *halfpel_hv; |
343 |
|
int32_t x_int, y_int, x_frac, y_frac; |
344 |
|
|
345 |
|
x_int = xRef/4; |
346 |
|
if (xRef < 0 && xRef % 4) |
347 |
|
x_int--; |
348 |
|
|
349 |
|
x_frac = xRef - (4*x_int); |
350 |
|
|
351 |
|
y_int = yRef/4; |
352 |
|
if (yRef < 0 && yRef % 4) |
353 |
|
y_int--; |
354 |
|
|
355 |
|
y_frac = yRef - (4*y_int); |
356 |
|
|
357 |
|
src = refn + y_int * stride + x_int; |
358 |
|
halfpel_h = refh; |
359 |
|
halfpel_v = refv; |
360 |
|
halfpel_hv = refhv; |
361 |
|
|
362 |
|
dst = cur + y * stride + x; |
363 |
|
|
364 |
|
switch((y_frac << 2) | (x_frac)) { |
365 |
|
|
366 |
|
case 0: |
367 |
|
transfer8x8_copy(dst, src, stride); |
368 |
|
transfer8x8_copy(dst+8, src+8, stride); |
369 |
|
transfer8x8_copy(dst+8*stride, src+8*stride, stride); |
370 |
|
transfer8x8_copy(dst+8*stride+8, src+8*stride+8, stride); |
371 |
|
break; |
372 |
|
|
373 |
|
case 1: |
374 |
|
interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding); |
375 |
|
interpolate8x8_avg2(dst, src, halfpel_h, stride, rounding, 9); |
376 |
|
interpolate8x8_avg2(dst+8, src+8, halfpel_h+8, stride, rounding, 9); |
377 |
|
interpolate8x8_avg2(dst+8*stride, src+8*stride, halfpel_h+8*stride, stride, rounding, 9); |
378 |
|
interpolate8x8_avg2(dst+8*stride+8, src+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 9); |
379 |
|
break; |
380 |
|
|
381 |
|
case 2: |
382 |
|
interpolate16x16_lowpass_h(dst, src, stride, rounding); |
383 |
|
break; |
384 |
|
|
385 |
|
case 3: |
386 |
|
interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding); |
387 |
|
interpolate8x8_avg2(dst, src + 1, halfpel_h, stride, rounding, 9); |
388 |
|
interpolate8x8_avg2(dst+8, src + 8 + 1, halfpel_h+8, stride, rounding, 9); |
389 |
|
interpolate8x8_avg2(dst+8*stride, src + 8*stride + 1, halfpel_h+8*stride, stride, rounding, 9); |
390 |
|
interpolate8x8_avg2(dst+8*stride+8, src+8*stride+8 + 1, halfpel_h+8*stride+8, stride, rounding, 9); |
391 |
|
break; |
392 |
|
|
393 |
|
case 4: |
394 |
|
interpolate16x16_lowpass_v(halfpel_v, src, stride, rounding); |
395 |
|
interpolate8x8_avg2(dst, src, halfpel_v, stride, rounding, 9); |
396 |
|
interpolate8x8_avg2(dst+8, src+8, halfpel_v+8, stride, rounding, 9); |
397 |
|
interpolate8x8_avg2(dst+8*stride, src+8*stride, halfpel_v+8*stride, stride, rounding, 9); |
398 |
|
interpolate8x8_avg2(dst+8*stride+8, src+8*stride+8, halfpel_v+8*stride+8, stride, rounding, 9); |
399 |
|
break; |
400 |
|
|
401 |
|
case 5: |
402 |
|
interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding); |
403 |
|
interpolate8x8_avg2(halfpel_v, src, halfpel_h, stride, rounding, 9); |
404 |
|
interpolate8x8_avg2(halfpel_v+8, src + 8, halfpel_h+8, stride, rounding, 9); |
405 |
|
interpolate8x8_avg2(halfpel_v+8*stride, src + 8*stride, halfpel_h+8*stride, stride, rounding, 9); |
406 |
|
interpolate8x8_avg2(halfpel_v+8*stride+8, src+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 9); |
407 |
|
|
408 |
|
interpolate16x16_lowpass_v(halfpel_hv, halfpel_v, stride, rounding); |
409 |
|
interpolate8x8_avg2(dst, halfpel_hv, halfpel_v, stride, rounding, 9); |
410 |
|
interpolate8x8_avg2(dst+8, halfpel_hv+8, halfpel_v+8, stride, rounding, 9); |
411 |
|
interpolate8x8_avg2(dst+8*stride, halfpel_hv+8*stride, halfpel_v+8*stride, stride, rounding, 9); |
412 |
|
interpolate8x8_avg2(dst+8*stride+8, halfpel_hv+8*stride+8, halfpel_v+8*stride+8, stride, rounding, 9); |
413 |
|
break; |
414 |
|
|
415 |
|
case 6: |
416 |
|
interpolate16x16_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
417 |
|
interpolate8x8_avg2(dst, halfpel_h, halfpel_hv, stride, rounding, 9); |
418 |
|
interpolate8x8_avg2(dst+8, halfpel_h+8, halfpel_hv+8, stride, rounding, 9); |
419 |
|
interpolate8x8_avg2(dst+8*stride, halfpel_h+8*stride, halfpel_hv+8*stride, stride, rounding, 9); |
420 |
|
interpolate8x8_avg2(dst+8*stride+8, halfpel_h+8*stride+8, halfpel_hv+8*stride+8, stride, rounding, 9); |
421 |
|
break; |
422 |
|
|
423 |
|
case 7: |
424 |
|
interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding); |
425 |
|
interpolate8x8_avg2(halfpel_v, src+1, halfpel_h, stride, rounding, 9); |
426 |
|
interpolate8x8_avg2(halfpel_v+8, src+1 + 8, halfpel_h+8, stride, rounding, 9); |
427 |
|
interpolate8x8_avg2(halfpel_v+8*stride, src+1 + 8*stride, halfpel_h+8*stride, stride, rounding, 9); |
428 |
|
interpolate8x8_avg2(halfpel_v+8*stride+8, src+1+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 9); |
429 |
|
|
430 |
|
interpolate16x16_lowpass_v(halfpel_hv, halfpel_v, stride, rounding); |
431 |
|
interpolate8x8_avg2(dst, halfpel_hv, halfpel_v, stride, rounding, 9); |
432 |
|
interpolate8x8_avg2(dst+8, halfpel_hv+8, halfpel_v+8, stride, rounding, 9); |
433 |
|
interpolate8x8_avg2(dst+8*stride, halfpel_hv+8*stride, halfpel_v+8*stride, stride, rounding, 9); |
434 |
|
interpolate8x8_avg2(dst+8*stride+8, halfpel_hv+8*stride+8, halfpel_v+8*stride+8, stride, rounding, 9); |
435 |
|
break; |
436 |
|
|
437 |
|
case 8: |
438 |
|
interpolate16x16_lowpass_v(dst, src, stride, rounding); |
439 |
|
break; |
440 |
|
|
441 |
|
case 9: |
442 |
|
interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding); |
443 |
|
interpolate8x8_avg2(halfpel_v, src, halfpel_h, stride, rounding, 9); |
444 |
|
interpolate8x8_avg2(halfpel_v+8, src + 8, halfpel_h+8, stride, rounding, 9); |
445 |
|
interpolate8x8_avg2(halfpel_v+8*stride, src + 8*stride, halfpel_h+8*stride, stride, rounding, 9); |
446 |
|
interpolate8x8_avg2(halfpel_v+8*stride+8, src+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 9); |
447 |
|
interpolate16x16_lowpass_v(dst, halfpel_v, stride, rounding); |
448 |
|
break; |
449 |
|
|
450 |
|
case 10: |
451 |
|
interpolate16x16_lowpass_hv(dst, halfpel_h, src, stride, rounding); |
452 |
|
break; |
453 |
|
|
454 |
|
case 11: |
455 |
|
interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding); |
456 |
|
interpolate8x8_avg2(halfpel_v, src+1, halfpel_h, stride, rounding, 9); |
457 |
|
interpolate8x8_avg2(halfpel_v+8, src+1 + 8, halfpel_h+8, stride, rounding, 9); |
458 |
|
interpolate8x8_avg2(halfpel_v+8*stride, src+1 + 8*stride, halfpel_h+8*stride, stride, rounding, 9); |
459 |
|
interpolate8x8_avg2(halfpel_v+8*stride+8, src+1+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 9); |
460 |
|
interpolate16x16_lowpass_v(dst, halfpel_v, stride, rounding); |
461 |
|
break; |
462 |
|
|
463 |
|
case 12: |
464 |
|
interpolate16x16_lowpass_v(halfpel_v, src, stride, rounding); |
465 |
|
interpolate8x8_avg2(dst, src+stride, halfpel_v, stride, rounding, 9); |
466 |
|
interpolate8x8_avg2(dst+8, src+stride+8, halfpel_v+8, stride, rounding, 9); |
467 |
|
interpolate8x8_avg2(dst+8*stride, src+stride+8*stride, halfpel_v+8*stride, stride, rounding, 9); |
468 |
|
interpolate8x8_avg2(dst+8*stride+8, src+stride+8*stride+8, halfpel_v+8*stride+8, stride, rounding, 9); |
469 |
|
break; |
470 |
|
|
471 |
|
case 13: |
472 |
|
interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding); |
473 |
|
interpolate8x8_avg2(halfpel_v, src, halfpel_h, stride, rounding, 9); |
474 |
|
interpolate8x8_avg2(halfpel_v+8, src + 8, halfpel_h+8, stride, rounding, 9); |
475 |
|
interpolate8x8_avg2(halfpel_v+8*stride, src + 8*stride, halfpel_h+8*stride, stride, rounding, 9); |
476 |
|
interpolate8x8_avg2(halfpel_v+8*stride+8, src+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 9); |
477 |
|
|
478 |
|
interpolate16x16_lowpass_v(halfpel_hv, halfpel_v, stride, rounding); |
479 |
|
interpolate8x8_avg2(dst, halfpel_hv, halfpel_v+stride, stride, rounding, 9); |
480 |
|
interpolate8x8_avg2(dst+8, halfpel_hv+8, halfpel_v+stride+8, stride, rounding, 9); |
481 |
|
interpolate8x8_avg2(dst+8*stride, halfpel_hv+8*stride, halfpel_v+stride+8*stride, stride, rounding, 9); |
482 |
|
interpolate8x8_avg2(dst+8*stride+8, halfpel_hv+8*stride+8, halfpel_v+stride+8*stride+8, stride, rounding, 9); |
483 |
|
break; |
484 |
|
|
485 |
|
case 14: |
486 |
|
interpolate16x16_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
487 |
|
interpolate8x8_avg2(dst, halfpel_h+stride, halfpel_hv, stride, rounding, 9); |
488 |
|
interpolate8x8_avg2(dst+8, halfpel_h+stride+8, halfpel_hv+8, stride, rounding, 9); |
489 |
|
interpolate8x8_avg2(dst+8*stride, halfpel_h+stride+8*stride, halfpel_hv+8*stride, stride, rounding, 9); |
490 |
|
interpolate8x8_avg2(dst+8*stride+8, halfpel_h+stride+8*stride+8, halfpel_hv+8*stride+8, stride, rounding, 9); |
491 |
|
break; |
492 |
|
|
493 |
|
case 15: |
494 |
|
interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding); |
495 |
|
interpolate8x8_avg2(halfpel_v, src+1, halfpel_h, stride, rounding, 9); |
496 |
|
interpolate8x8_avg2(halfpel_v+8, src+1 + 8, halfpel_h+8, stride, rounding, 9); |
497 |
|
interpolate8x8_avg2(halfpel_v+8*stride, src+1 + 8*stride, halfpel_h+8*stride, stride, rounding, 9); |
498 |
|
interpolate8x8_avg2(halfpel_v+8*stride+8, src+1+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 9); |
499 |
|
|
500 |
|
interpolate16x16_lowpass_v(halfpel_hv, halfpel_v, stride, rounding); |
501 |
|
interpolate8x8_avg2(dst, halfpel_hv, halfpel_v+stride, stride, rounding, 9); |
502 |
|
interpolate8x8_avg2(dst+8, halfpel_hv+8, halfpel_v+stride+8, stride, rounding, 9); |
503 |
|
interpolate8x8_avg2(dst+8*stride, halfpel_hv+8*stride, halfpel_v+stride+8*stride, stride, rounding, 9); |
504 |
|
interpolate8x8_avg2(dst+8*stride+8, halfpel_hv+8*stride+8, halfpel_v+stride+8*stride+8, stride, rounding, 9); |
505 |
|
break; |
506 |
|
} |
507 |
|
} |
508 |
|
#endif |