19 |
* along with this program ; if not, write to the Free Software |
* along with this program ; if not, write to the Free Software |
20 |
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA |
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA |
21 |
* |
* |
22 |
* $Id: interpolate8x8.h,v 1.12 2004-04-05 20:36:36 edgomez Exp $ |
* $Id: interpolate8x8.h,v 1.15 2005-01-05 23:02:15 edgomez Exp $ |
23 |
* |
* |
24 |
****************************************************************************/ |
****************************************************************************/ |
25 |
|
|
73 |
|
|
74 |
typedef INTERPOLATE8X8_6TAP_LOWPASS *INTERPOLATE8X8_6TAP_LOWPASS_PTR; |
typedef INTERPOLATE8X8_6TAP_LOWPASS *INTERPOLATE8X8_6TAP_LOWPASS_PTR; |
75 |
|
|
76 |
|
/* These function do: dst = interpolate(src) */ |
77 |
extern INTERPOLATE8X8_PTR interpolate8x8_halfpel_h; |
extern INTERPOLATE8X8_PTR interpolate8x8_halfpel_h; |
78 |
extern INTERPOLATE8X8_PTR interpolate8x8_halfpel_v; |
extern INTERPOLATE8X8_PTR interpolate8x8_halfpel_v; |
79 |
extern INTERPOLATE8X8_PTR interpolate8x8_halfpel_hv; |
extern INTERPOLATE8X8_PTR interpolate8x8_halfpel_hv; |
80 |
|
|
81 |
|
/* These functions do: dst = (dst+interpolate(src) + 1)/2 |
82 |
|
* Suitable for direct/interpolated bvop prediction block |
83 |
|
* building w/o the need for intermediate interpolated result |
84 |
|
* storing/reading |
85 |
|
* NB: the rounding applies to the interpolation, but not |
86 |
|
* the averaging step which will always use rounding=0 */ |
87 |
|
extern INTERPOLATE8X8_PTR interpolate8x8_halfpel_add; |
88 |
|
extern INTERPOLATE8X8_PTR interpolate8x8_halfpel_h_add; |
89 |
|
extern INTERPOLATE8X8_PTR interpolate8x8_halfpel_v_add; |
90 |
|
extern INTERPOLATE8X8_PTR interpolate8x8_halfpel_hv_add; |
91 |
|
|
92 |
extern INTERPOLATE8X8_AVG2_PTR interpolate8x8_avg2; |
extern INTERPOLATE8X8_AVG2_PTR interpolate8x8_avg2; |
93 |
extern INTERPOLATE8X8_AVG4_PTR interpolate8x8_avg4; |
extern INTERPOLATE8X8_AVG4_PTR interpolate8x8_avg4; |
94 |
|
|
107 |
INTERPOLATE8X8 interpolate8x8_halfpel_h_c; |
INTERPOLATE8X8 interpolate8x8_halfpel_h_c; |
108 |
INTERPOLATE8X8 interpolate8x8_halfpel_v_c; |
INTERPOLATE8X8 interpolate8x8_halfpel_v_c; |
109 |
INTERPOLATE8X8 interpolate8x8_halfpel_hv_c; |
INTERPOLATE8X8 interpolate8x8_halfpel_hv_c; |
110 |
|
INTERPOLATE8X8 interpolate8x8_halfpel_add_c; |
111 |
|
INTERPOLATE8X8 interpolate8x8_halfpel_h_add_c; |
112 |
|
INTERPOLATE8X8 interpolate8x8_halfpel_v_add_c; |
113 |
|
INTERPOLATE8X8 interpolate8x8_halfpel_hv_add_c; |
114 |
|
|
115 |
#ifdef ARCH_IS_IA32 |
#ifdef ARCH_IS_IA32 |
116 |
INTERPOLATE8X8 interpolate8x8_halfpel_h_mmx; |
INTERPOLATE8X8 interpolate8x8_halfpel_h_mmx; |
117 |
INTERPOLATE8X8 interpolate8x8_halfpel_v_mmx; |
INTERPOLATE8X8 interpolate8x8_halfpel_v_mmx; |
118 |
INTERPOLATE8X8 interpolate8x8_halfpel_hv_mmx; |
INTERPOLATE8X8 interpolate8x8_halfpel_hv_mmx; |
119 |
|
|
120 |
|
INTERPOLATE8X8 interpolate8x8_halfpel_add_mmx; |
121 |
|
INTERPOLATE8X8 interpolate8x8_halfpel_h_add_mmx; |
122 |
|
INTERPOLATE8X8 interpolate8x8_halfpel_v_add_mmx; |
123 |
|
INTERPOLATE8X8 interpolate8x8_halfpel_hv_add_mmx; |
124 |
|
|
125 |
INTERPOLATE8X8 interpolate8x8_halfpel_h_xmm; |
INTERPOLATE8X8 interpolate8x8_halfpel_h_xmm; |
126 |
INTERPOLATE8X8 interpolate8x8_halfpel_v_xmm; |
INTERPOLATE8X8 interpolate8x8_halfpel_v_xmm; |
127 |
INTERPOLATE8X8 interpolate8x8_halfpel_hv_xmm; |
INTERPOLATE8X8 interpolate8x8_halfpel_hv_xmm; |
128 |
|
|
129 |
|
INTERPOLATE8X8 interpolate8x8_halfpel_add_xmm; |
130 |
|
INTERPOLATE8X8 interpolate8x8_halfpel_h_add_xmm; |
131 |
|
INTERPOLATE8X8 interpolate8x8_halfpel_v_add_xmm; |
132 |
|
INTERPOLATE8X8 interpolate8x8_halfpel_hv_add_xmm; |
133 |
|
|
134 |
INTERPOLATE8X8 interpolate8x8_halfpel_h_3dn; |
INTERPOLATE8X8 interpolate8x8_halfpel_h_3dn; |
135 |
INTERPOLATE8X8 interpolate8x8_halfpel_v_3dn; |
INTERPOLATE8X8 interpolate8x8_halfpel_v_3dn; |
136 |
INTERPOLATE8X8 interpolate8x8_halfpel_hv_3dn; |
INTERPOLATE8X8 interpolate8x8_halfpel_hv_3dn; |
150 |
INTERPOLATE8X8 interpolate8x8_halfpel_h_altivec_c; |
INTERPOLATE8X8 interpolate8x8_halfpel_h_altivec_c; |
151 |
INTERPOLATE8X8 interpolate8x8_halfpel_v_altivec_c; |
INTERPOLATE8X8 interpolate8x8_halfpel_v_altivec_c; |
152 |
INTERPOLATE8X8 interpolate8x8_halfpel_hv_altivec_c; |
INTERPOLATE8X8 interpolate8x8_halfpel_hv_altivec_c; |
153 |
|
|
154 |
|
INTERPOLATE8X8 interpolate8x8_halfpel_add_altivec_c; |
155 |
|
INTERPOLATE8X8 interpolate8x8_halfpel_h_add_altivec_c; |
156 |
|
INTERPOLATE8X8 interpolate8x8_halfpel_v_add_altivec_c; |
157 |
|
INTERPOLATE8X8 interpolate8x8_halfpel_hv_add_altivec_c; |
158 |
|
#endif |
159 |
|
|
160 |
|
#ifdef ARCH_IS_X86_64 |
161 |
|
INTERPOLATE8X8 interpolate8x8_halfpel_h_x86_64; |
162 |
|
INTERPOLATE8X8 interpolate8x8_halfpel_v_x86_64; |
163 |
|
INTERPOLATE8X8 interpolate8x8_halfpel_hv_x86_64; |
164 |
|
|
165 |
|
INTERPOLATE8X8 interpolate8x8_halfpel_add_x86_64; |
166 |
|
INTERPOLATE8X8 interpolate8x8_halfpel_h_add_x86_64; |
167 |
|
INTERPOLATE8X8 interpolate8x8_halfpel_v_add_x86_64; |
168 |
|
INTERPOLATE8X8 interpolate8x8_halfpel_hv_add_x86_64; |
169 |
#endif |
#endif |
170 |
|
|
171 |
INTERPOLATE8X8_AVG2 interpolate8x8_avg2_c; |
INTERPOLATE8X8_AVG2 interpolate8x8_avg2_c; |
181 |
INTERPOLATE8X8_AVG4 interpolate8x8_avg4_altivec_c; |
INTERPOLATE8X8_AVG4 interpolate8x8_avg4_altivec_c; |
182 |
#endif |
#endif |
183 |
|
|
184 |
|
#ifdef ARCH_IS_X86_64 |
185 |
|
INTERPOLATE8X8_AVG2 interpolate8x8_avg2_x86_64; |
186 |
|
INTERPOLATE8X8_AVG4 interpolate8x8_avg4_x86_64; |
187 |
|
#endif |
188 |
|
|
189 |
INTERPOLATE_LOWPASS interpolate8x8_lowpass_h_c; |
INTERPOLATE_LOWPASS interpolate8x8_lowpass_h_c; |
190 |
INTERPOLATE_LOWPASS interpolate8x8_lowpass_v_c; |
INTERPOLATE_LOWPASS interpolate8x8_lowpass_v_c; |
191 |
|
|
207 |
INTERPOLATE8X8_6TAP_LOWPASS interpolate8x8_6tap_lowpass_h_altivec_c; |
INTERPOLATE8X8_6TAP_LOWPASS interpolate8x8_6tap_lowpass_h_altivec_c; |
208 |
#endif |
#endif |
209 |
|
|
210 |
|
#ifdef ARCH_IS_X86_64 |
211 |
|
INTERPOLATE8X8_6TAP_LOWPASS interpolate8x8_6tap_lowpass_h_x86_64; |
212 |
|
INTERPOLATE8X8_6TAP_LOWPASS interpolate8x8_6tap_lowpass_v_x86_64; |
213 |
|
#endif |
214 |
|
|
215 |
static __inline void |
static __inline void |
216 |
interpolate8x8_switch(uint8_t * const cur, |
interpolate8x8_switch(uint8_t * const cur, |
217 |
const uint8_t * const refn, |
const uint8_t * const refn, |
242 |
} |
} |
243 |
} |
} |
244 |
|
|
245 |
|
static __inline void |
246 |
|
interpolate8x8_add_switch(uint8_t * const cur, |
247 |
|
const uint8_t * const refn, |
248 |
|
const uint32_t x, |
249 |
|
const uint32_t y, |
250 |
|
const int32_t dx, |
251 |
|
const int dy, |
252 |
|
const uint32_t stride, |
253 |
|
const uint32_t rounding) |
254 |
|
{ |
255 |
|
|
256 |
|
const uint8_t * const src = refn + (int)((y + (dy>>1)) * stride + x + (dx>>1)); |
257 |
|
uint8_t * const dst = cur + (int)(y * stride + x); |
258 |
|
|
259 |
|
switch (((dx & 1) << 1) + (dy & 1)) { /* ((dx%2)?2:0)+((dy%2)?1:0) */ |
260 |
|
case 0: |
261 |
|
interpolate8x8_halfpel_add(dst, src, stride, rounding); |
262 |
|
break; |
263 |
|
case 1: |
264 |
|
interpolate8x8_halfpel_v_add(dst, src, stride, rounding); |
265 |
|
break; |
266 |
|
case 2: |
267 |
|
interpolate8x8_halfpel_h_add(dst, src, stride, rounding); |
268 |
|
break; |
269 |
|
default: |
270 |
|
interpolate8x8_halfpel_hv_add(dst, src, stride, rounding); |
271 |
|
break; |
272 |
|
} |
273 |
|
} |
274 |
|
|
275 |
static __inline void |
static __inline void |
276 |
interpolate16x16_switch(uint8_t * const cur, |
interpolate16x16_switch(uint8_t * const cur, |
288 |
interpolate8x8_switch(cur, refn, x+8, y+8, dx, dy, stride, rounding); |
interpolate8x8_switch(cur, refn, x+8, y+8, dx, dy, stride, rounding); |
289 |
} |
} |
290 |
|
|
291 |
|
static __inline void |
292 |
|
interpolate16x16_add_switch(uint8_t * const cur, |
293 |
|
const uint8_t * const refn, |
294 |
|
const uint32_t x, |
295 |
|
const uint32_t y, |
296 |
|
const int32_t dx, |
297 |
|
const int dy, |
298 |
|
const uint32_t stride, |
299 |
|
const uint32_t rounding) |
300 |
|
{ |
301 |
|
interpolate8x8_add_switch(cur, refn, x, y, dx, dy, stride, rounding); |
302 |
|
interpolate8x8_add_switch(cur, refn, x+8, y, dx, dy, stride, rounding); |
303 |
|
interpolate8x8_add_switch(cur, refn, x, y+8, dx, dy, stride, rounding); |
304 |
|
interpolate8x8_add_switch(cur, refn, x+8, y+8, dx, dy, stride, rounding); |
305 |
|
} |
306 |
|
|
307 |
static __inline void |
static __inline void |
308 |
interpolate32x32_switch(uint8_t * const cur, |
interpolate32x32_switch(uint8_t * const cur, |
320 |
interpolate16x16_switch(cur, refn, x+16, y+16, dx, dy, stride, rounding); |
interpolate16x16_switch(cur, refn, x+16, y+16, dx, dy, stride, rounding); |
321 |
} |
} |
322 |
|
|
323 |
|
static __inline void |
324 |
|
interpolate32x32_add_switch(uint8_t * const cur, |
325 |
|
const uint8_t * const refn, |
326 |
|
const uint32_t x, |
327 |
|
const uint32_t y, |
328 |
|
const int32_t dx, |
329 |
|
const int dy, |
330 |
|
const uint32_t stride, |
331 |
|
const uint32_t rounding) |
332 |
|
{ |
333 |
|
interpolate16x16_add_switch(cur, refn, x, y, dx, dy, stride, rounding); |
334 |
|
interpolate16x16_add_switch(cur, refn, x+16, y, dx, dy, stride, rounding); |
335 |
|
interpolate16x16_add_switch(cur, refn, x, y+16, dx, dy, stride, rounding); |
336 |
|
interpolate16x16_add_switch(cur, refn, x+16, y+16, dx, dy, stride, rounding); |
337 |
|
} |
338 |
|
|
339 |
static __inline uint8_t * |
static __inline uint8_t * |
340 |
interpolate8x8_switch2(uint8_t * const buffer, |
interpolate8x8_switch2(uint8_t * const buffer, |
365 |
return buffer; |
return buffer; |
366 |
} |
} |
367 |
|
|
|
static __inline void interpolate8x8_quarterpel(uint8_t * const cur, |
|
|
uint8_t * const refn, |
|
|
uint8_t * const refh, |
|
|
uint8_t * const refv, |
|
|
uint8_t * const refhv, |
|
|
const uint32_t x, const uint32_t y, |
|
|
const int32_t dx, const int dy, |
|
|
const uint32_t stride, |
|
|
const uint32_t rounding) |
|
|
{ |
|
|
const int32_t xRef = x*4 + dx; |
|
|
const int32_t yRef = y*4 + dy; |
|
|
|
|
|
uint8_t *src, *dst; |
|
|
uint8_t *halfpel_h, *halfpel_v, *halfpel_hv; |
|
|
int32_t x_int, y_int, x_frac, y_frac; |
|
|
|
|
|
x_int = xRef/4; |
|
|
if (xRef < 0 && xRef % 4) |
|
|
x_int--; |
|
|
|
|
|
x_frac = xRef - (4*x_int); |
|
|
|
|
|
y_int = yRef/4; |
|
|
if (yRef < 0 && yRef % 4) |
|
|
y_int--; |
|
|
|
|
|
y_frac = yRef - (4*y_int); |
|
|
|
|
|
src = refn + y_int * stride + x_int; |
|
|
halfpel_h = refh; |
|
|
halfpel_v = refv; |
|
|
halfpel_hv = refhv; |
|
|
|
|
|
dst = cur + y * stride + x; |
|
|
|
|
|
switch((y_frac << 2) | (x_frac)) { |
|
|
|
|
|
case 0: |
|
|
transfer8x8_copy(dst, src, stride); |
|
|
break; |
|
|
|
|
|
case 1: |
|
|
interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding); |
|
|
interpolate8x8_avg2(dst, src, halfpel_h, stride, rounding, 8); |
|
|
break; |
|
|
|
|
|
case 2: |
|
|
interpolate8x8_lowpass_h(dst, src, stride, rounding); |
|
|
break; |
|
|
|
|
|
case 3: |
|
|
interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding); |
|
|
interpolate8x8_avg2(dst, src + 1, halfpel_h, stride, rounding, 8); |
|
|
break; |
|
|
|
|
|
case 4: |
|
|
interpolate8x8_lowpass_v(halfpel_v, src, stride, rounding); |
|
|
interpolate8x8_avg2(dst, src, halfpel_v, stride, rounding, 8); |
|
|
break; |
|
|
|
|
|
case 5: |
|
|
interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding); |
|
|
interpolate8x8_avg2(halfpel_v, src, halfpel_h, stride, rounding, 9); |
|
|
interpolate8x8_lowpass_v(halfpel_hv, halfpel_v, stride, rounding); |
|
|
interpolate8x8_avg2(dst, halfpel_v, halfpel_hv, stride, rounding, 8); |
|
|
break; |
|
|
|
|
|
case 6: |
|
|
interpolate8x8_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
|
|
interpolate8x8_avg2(dst, halfpel_h, halfpel_hv, stride, rounding, 8); |
|
|
break; |
|
|
|
|
|
case 7: |
|
|
interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding); |
|
|
interpolate8x8_avg2(halfpel_v, src + 1, halfpel_h, stride, rounding, 9); |
|
|
interpolate8x8_lowpass_v(halfpel_hv, halfpel_v, stride, rounding); |
|
|
interpolate8x8_avg2(dst, halfpel_v, halfpel_hv, stride, rounding, 8); |
|
|
break; |
|
|
|
|
|
case 8: |
|
|
interpolate8x8_lowpass_v(dst, src, stride, rounding); |
|
|
break; |
|
|
|
|
|
case 9: |
|
|
interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding); |
|
|
interpolate8x8_avg2(halfpel_v, src, halfpel_h, stride, rounding, 9); |
|
|
interpolate8x8_lowpass_v(dst, halfpel_v, stride, rounding); |
|
|
break; |
|
|
|
|
|
case 10: |
|
|
interpolate8x8_lowpass_hv(dst, halfpel_h, src, stride, rounding); |
|
|
break; |
|
|
|
|
|
case 11: |
|
|
interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding); |
|
|
interpolate8x8_avg2(halfpel_v, src + 1, halfpel_h, stride, rounding, 9); |
|
|
interpolate8x8_lowpass_v(dst, halfpel_v, stride, rounding); |
|
|
break; |
|
|
|
|
|
case 12: |
|
|
interpolate8x8_lowpass_v(halfpel_v, src, stride, rounding); |
|
|
interpolate8x8_avg2(dst, src+stride, halfpel_v, stride, rounding, 8); |
|
|
break; |
|
|
|
|
|
case 13: |
|
|
interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding); |
|
|
interpolate8x8_avg2(halfpel_v, src, halfpel_h, stride, rounding, 9); |
|
|
interpolate8x8_lowpass_v(halfpel_hv, halfpel_v, stride, rounding); |
|
|
interpolate8x8_avg2(dst, halfpel_v+stride, halfpel_hv, stride, rounding, 8); |
|
|
break; |
|
|
|
|
|
case 14: |
|
|
interpolate8x8_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
|
|
interpolate8x8_avg2(dst, halfpel_h+stride, halfpel_hv, stride, rounding, 8); |
|
|
break; |
|
|
|
|
|
case 15: |
|
|
interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding); |
|
|
interpolate8x8_avg2(halfpel_v, src + 1, halfpel_h, stride, rounding, 9); |
|
|
interpolate8x8_lowpass_v(halfpel_hv, halfpel_v, stride, rounding); |
|
|
interpolate8x8_avg2(dst, halfpel_hv, halfpel_v + stride, stride, rounding, 8); |
|
|
break; |
|
|
} |
|
|
} |
|
|
|
|
|
static __inline void interpolate16x16_quarterpel(uint8_t * const cur, |
|
|
uint8_t * const refn, |
|
|
uint8_t * const refh, |
|
|
uint8_t * const refv, |
|
|
uint8_t * const refhv, |
|
|
const uint32_t x, const uint32_t y, |
|
|
const int32_t dx, const int dy, |
|
|
const uint32_t stride, |
|
|
const uint32_t rounding) |
|
|
{ |
|
|
const int32_t xRef = x*4 + dx; |
|
|
const int32_t yRef = y*4 + dy; |
|
|
|
|
|
uint8_t *src, *dst; |
|
|
uint8_t *halfpel_h, *halfpel_v, *halfpel_hv; |
|
|
int32_t x_int, y_int, x_frac, y_frac; |
|
|
|
|
|
x_int = xRef/4; |
|
|
if (xRef < 0 && xRef % 4) |
|
|
x_int--; |
|
|
|
|
|
x_frac = xRef - (4*x_int); |
|
|
|
|
|
y_int = yRef/4; |
|
|
if (yRef < 0 && yRef % 4) |
|
|
y_int--; |
|
|
|
|
|
y_frac = yRef - (4*y_int); |
|
|
|
|
|
src = refn + y_int * stride + x_int; |
|
|
halfpel_h = refh; |
|
|
halfpel_v = refv; |
|
|
halfpel_hv = refhv; |
|
|
|
|
|
dst = cur + y * stride + x; |
|
|
|
|
|
switch((y_frac << 2) | (x_frac)) { |
|
|
|
|
|
case 0: |
|
|
transfer16x16_copy(dst, src, stride); |
|
|
break; |
|
|
|
|
|
case 1: |
|
|
interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding); |
|
|
interpolate8x8_avg2(dst, src, halfpel_h, stride, rounding, 8); |
|
|
interpolate8x8_avg2(dst+8, src+8, halfpel_h+8, stride, rounding, 8); |
|
|
interpolate8x8_avg2(dst+8*stride, src+8*stride, halfpel_h+8*stride, stride, rounding, 8); |
|
|
interpolate8x8_avg2(dst+8*stride+8, src+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 8); |
|
|
break; |
|
|
|
|
|
case 2: |
|
|
interpolate16x16_lowpass_h(dst, src, stride, rounding); |
|
|
break; |
|
|
|
|
|
case 3: |
|
|
interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding); |
|
|
interpolate8x8_avg2(dst, src + 1, halfpel_h, stride, rounding, 8); |
|
|
interpolate8x8_avg2(dst+8, src + 8 + 1, halfpel_h+8, stride, rounding, 8); |
|
|
interpolate8x8_avg2(dst+8*stride, src + 8*stride + 1, halfpel_h+8*stride, stride, rounding, 8); |
|
|
interpolate8x8_avg2(dst+8*stride+8, src+8*stride+8 + 1, halfpel_h+8*stride+8, stride, rounding, 8); |
|
|
break; |
|
|
|
|
|
case 4: |
|
|
interpolate16x16_lowpass_v(halfpel_v, src, stride, rounding); |
|
|
interpolate8x8_avg2(dst, src, halfpel_v, stride, rounding, 8); |
|
|
interpolate8x8_avg2(dst+8, src+8, halfpel_v+8, stride, rounding, 8); |
|
|
interpolate8x8_avg2(dst+8*stride, src+8*stride, halfpel_v+8*stride, stride, rounding, 8); |
|
|
interpolate8x8_avg2(dst+8*stride+8, src+8*stride+8, halfpel_v+8*stride+8, stride, rounding, 8); |
|
|
break; |
|
|
|
|
|
case 5: |
|
|
interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding); |
|
|
interpolate8x8_avg2(halfpel_v, src, halfpel_h, stride, rounding, 9); |
|
|
interpolate8x8_avg2(halfpel_v+8, src + 8, halfpel_h+8, stride, rounding, 9); |
|
|
interpolate8x8_avg2(halfpel_v+8*stride, src + 8*stride, halfpel_h+8*stride, stride, rounding, 9); |
|
|
interpolate8x8_avg2(halfpel_v+8*stride+8, src+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 9); |
|
|
|
|
|
interpolate16x16_lowpass_v(halfpel_hv, halfpel_v, stride, rounding); |
|
|
interpolate8x8_avg2(dst, halfpel_hv, halfpel_v, stride, rounding, 8); |
|
|
interpolate8x8_avg2(dst+8, halfpel_hv+8, halfpel_v+8, stride, rounding, 8); |
|
|
interpolate8x8_avg2(dst+8*stride, halfpel_hv+8*stride, halfpel_v+8*stride, stride, rounding, 8); |
|
|
interpolate8x8_avg2(dst+8*stride+8, halfpel_hv+8*stride+8, halfpel_v+8*stride+8, stride, rounding, 8); |
|
|
break; |
|
|
|
|
|
case 6: |
|
|
interpolate16x16_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
|
|
interpolate8x8_avg2(dst, halfpel_h, halfpel_hv, stride, rounding, 8); |
|
|
interpolate8x8_avg2(dst+8, halfpel_h+8, halfpel_hv+8, stride, rounding, 8); |
|
|
interpolate8x8_avg2(dst+8*stride, halfpel_h+8*stride, halfpel_hv+8*stride, stride, rounding, 8); |
|
|
interpolate8x8_avg2(dst+8*stride+8, halfpel_h+8*stride+8, halfpel_hv+8*stride+8, stride, rounding, 8); |
|
|
break; |
|
|
|
|
|
case 7: |
|
|
interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding); |
|
|
interpolate8x8_avg2(halfpel_v, src+1, halfpel_h, stride, rounding, 9); |
|
|
interpolate8x8_avg2(halfpel_v+8, src+1 + 8, halfpel_h+8, stride, rounding, 9); |
|
|
interpolate8x8_avg2(halfpel_v+8*stride, src+1 + 8*stride, halfpel_h+8*stride, stride, rounding, 9); |
|
|
interpolate8x8_avg2(halfpel_v+8*stride+8, src+1+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 9); |
|
|
|
|
|
interpolate16x16_lowpass_v(halfpel_hv, halfpel_v, stride, rounding); |
|
|
interpolate8x8_avg2(dst, halfpel_hv, halfpel_v, stride, rounding, 8); |
|
|
interpolate8x8_avg2(dst+8, halfpel_hv+8, halfpel_v+8, stride, rounding, 8); |
|
|
interpolate8x8_avg2(dst+8*stride, halfpel_hv+8*stride, halfpel_v+8*stride, stride, rounding, 8); |
|
|
interpolate8x8_avg2(dst+8*stride+8, halfpel_hv+8*stride+8, halfpel_v+8*stride+8, stride, rounding, 8); |
|
|
break; |
|
|
|
|
|
case 8: |
|
|
interpolate16x16_lowpass_v(dst, src, stride, rounding); |
|
|
break; |
|
|
|
|
|
case 9: |
|
|
interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding); |
|
|
interpolate8x8_avg2(halfpel_v, src, halfpel_h, stride, rounding, 9); |
|
|
interpolate8x8_avg2(halfpel_v+8, src + 8, halfpel_h+8, stride, rounding, 9); |
|
|
interpolate8x8_avg2(halfpel_v+8*stride, src + 8*stride, halfpel_h+8*stride, stride, rounding, 9); |
|
|
interpolate8x8_avg2(halfpel_v+8*stride+8, src+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 9); |
|
|
interpolate16x16_lowpass_v(dst, halfpel_v, stride, rounding); |
|
|
break; |
|
|
|
|
|
case 10: |
|
|
interpolate16x16_lowpass_hv(dst, halfpel_h, src, stride, rounding); |
|
|
break; |
|
|
|
|
|
case 11: |
|
|
interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding); |
|
|
interpolate8x8_avg2(halfpel_v, src+1, halfpel_h, stride, rounding, 9); |
|
|
interpolate8x8_avg2(halfpel_v+8, src+1 + 8, halfpel_h+8, stride, rounding, 9); |
|
|
interpolate8x8_avg2(halfpel_v+8*stride, src+1 + 8*stride, halfpel_h+8*stride, stride, rounding, 9); |
|
|
interpolate8x8_avg2(halfpel_v+8*stride+8, src+1+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 9); |
|
|
interpolate16x16_lowpass_v(dst, halfpel_v, stride, rounding); |
|
|
break; |
|
|
|
|
|
case 12: |
|
|
interpolate16x16_lowpass_v(halfpel_v, src, stride, rounding); |
|
|
interpolate8x8_avg2(dst, src+stride, halfpel_v, stride, rounding, 8); |
|
|
interpolate8x8_avg2(dst+8, src+stride+8, halfpel_v+8, stride, rounding, 8); |
|
|
interpolate8x8_avg2(dst+8*stride, src+stride+8*stride, halfpel_v+8*stride, stride, rounding, 8); |
|
|
interpolate8x8_avg2(dst+8*stride+8, src+stride+8*stride+8, halfpel_v+8*stride+8, stride, rounding, 8); |
|
|
break; |
|
|
|
|
|
case 13: |
|
|
interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding); |
|
|
interpolate8x8_avg2(halfpel_v, src, halfpel_h, stride, rounding, 9); |
|
|
interpolate8x8_avg2(halfpel_v+8, src + 8, halfpel_h+8, stride, rounding, 9); |
|
|
interpolate8x8_avg2(halfpel_v+8*stride, src + 8*stride, halfpel_h+8*stride, stride, rounding, 9); |
|
|
interpolate8x8_avg2(halfpel_v+8*stride+8, src+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 9); |
|
|
|
|
|
interpolate16x16_lowpass_v(halfpel_hv, halfpel_v, stride, rounding); |
|
|
interpolate8x8_avg2(dst, halfpel_hv, halfpel_v+stride, stride, rounding, 8); |
|
|
interpolate8x8_avg2(dst+8, halfpel_hv+8, halfpel_v+stride+8, stride, rounding, 8); |
|
|
interpolate8x8_avg2(dst+8*stride, halfpel_hv+8*stride, halfpel_v+stride+8*stride, stride, rounding, 8); |
|
|
interpolate8x8_avg2(dst+8*stride+8, halfpel_hv+8*stride+8, halfpel_v+stride+8*stride+8, stride, rounding, 8); |
|
|
break; |
|
|
|
|
|
case 14: |
|
|
interpolate16x16_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
|
|
interpolate8x8_avg2(dst, halfpel_h+stride, halfpel_hv, stride, rounding, 8); |
|
|
interpolate8x8_avg2(dst+8, halfpel_h+stride+8, halfpel_hv+8, stride, rounding, 8); |
|
|
interpolate8x8_avg2(dst+8*stride, halfpel_h+stride+8*stride, halfpel_hv+8*stride, stride, rounding, 8); |
|
|
interpolate8x8_avg2(dst+8*stride+8, halfpel_h+stride+8*stride+8, halfpel_hv+8*stride+8, stride, rounding, 8); |
|
|
break; |
|
|
|
|
|
case 15: |
|
|
interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding); |
|
|
interpolate8x8_avg2(halfpel_v, src+1, halfpel_h, stride, rounding, 9); |
|
|
interpolate8x8_avg2(halfpel_v+8, src+1 + 8, halfpel_h+8, stride, rounding, 9); |
|
|
interpolate8x8_avg2(halfpel_v+8*stride, src+1 + 8*stride, halfpel_h+8*stride, stride, rounding, 9); |
|
|
interpolate8x8_avg2(halfpel_v+8*stride+8, src+1+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 9); |
|
|
|
|
|
interpolate16x16_lowpass_v(halfpel_hv, halfpel_v, stride, rounding); |
|
|
interpolate8x8_avg2(dst, halfpel_hv, halfpel_v+stride, stride, rounding, 8); |
|
|
interpolate8x8_avg2(dst+8, halfpel_hv+8, halfpel_v+stride+8, stride, rounding, 8); |
|
|
interpolate8x8_avg2(dst+8*stride, halfpel_hv+8*stride, halfpel_v+stride+8*stride, stride, rounding, 8); |
|
|
interpolate8x8_avg2(dst+8*stride+8, halfpel_hv+8*stride+8, halfpel_v+stride+8*stride+8, stride, rounding, 8); |
|
|
break; |
|
|
} |
|
|
} |
|
|
|
|
368 |
#endif |
#endif |