FFmpeg
 All Data Structures Namespaces Files Functions Variables Typedefs Enumerations Enumerator Macros Groups Pages
h264chroma_msa.c
Go to the documentation of this file.
1 /*
2  * Copyright (c) 2015 Shivraj Patil (Shivraj.Patil@imgtec.com)
3  *
4  * This file is part of FFmpeg.
5  *
6  * FFmpeg is free software; you can redistribute it and/or
7  * modify it under the terms of the GNU Lesser General Public
8  * License as published by the Free Software Foundation; either
9  * version 2.1 of the License, or (at your option) any later version.
10  *
11  * FFmpeg is distributed in the hope that it will be useful,
12  * but WITHOUT ANY WARRANTY; without even the implied warranty of
13  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14  * Lesser General Public License for more details.
15  *
16  * You should have received a copy of the GNU Lesser General Public
17  * License along with FFmpeg; if not, write to the Free Software
18  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
19  */
20 
21 #include "libavutil/mips/generic_macros_msa.h"
22 #include "h264chroma_mips.h"
23 
24 static const uint8_t chroma_mask_arr[16 * 5] = {
25  0, 1, 1, 2, 2, 3, 3, 4, 16, 17, 17, 18, 18, 19, 19, 20,
26  0, 2, 2, 4, 4, 6, 6, 8, 16, 18, 18, 20, 20, 22, 22, 24,
27  0, 1, 1, 2, 2, 3, 3, 4, 4, 5, 5, 6, 6, 7, 7, 8,
28  0, 1, 1, 2, 16, 17, 17, 18, 4, 5, 5, 6, 6, 7, 7, 8,
29  0, 1, 1, 2, 16, 17, 17, 18, 16, 17, 17, 18, 18, 19, 19, 20
30 };
31 
32 static void avc_chroma_hz_2x2_msa(uint8_t *src, int32_t src_stride,
33  uint8_t *dst, int32_t dst_stride,
34  uint32_t coeff0, uint32_t coeff1)
35 {
36  uint16_t out0, out1;
37  v16i8 src0, src1;
38  v8u16 res_r;
39  v8i16 res;
40  v16i8 mask;
41  v16i8 coeff_vec0 = __msa_fill_b(coeff0);
42  v16i8 coeff_vec1 = __msa_fill_b(coeff1);
43  v16u8 coeff_vec = (v16u8) __msa_ilvr_b(coeff_vec0, coeff_vec1);
44 
45  mask = LD_SB(&chroma_mask_arr[0]);
46 
47  LD_SB2(src, src_stride, src0, src1);
48 
49  src0 = __msa_vshf_b(mask, src1, src0);
50  res_r = __msa_dotp_u_h((v16u8) src0, coeff_vec);
51  res_r <<= 3;
52  res_r = (v8u16) __msa_srari_h((v8i16) res_r, 6);
53  res_r = __msa_sat_u_h(res_r, 7);
54  res = (v8i16) __msa_pckev_b((v16i8) res_r, (v16i8) res_r);
55 
56  out0 = __msa_copy_u_h(res, 0);
57  out1 = __msa_copy_u_h(res, 2);
58 
59  SH(out0, dst);
60  dst += dst_stride;
61  SH(out1, dst);
62 }
63 
64 static void avc_chroma_hz_2x4_msa(uint8_t *src, int32_t src_stride,
65  uint8_t *dst, int32_t dst_stride,
66  uint32_t coeff0, uint32_t coeff1)
67 {
68  v16u8 src0, src1, src2, src3;
69  v8u16 res_r;
70  v8i16 res;
71  v16i8 mask;
72  v16i8 coeff_vec0 = __msa_fill_b(coeff0);
73  v16i8 coeff_vec1 = __msa_fill_b(coeff1);
74  v16u8 coeff_vec = (v16u8) __msa_ilvr_b(coeff_vec0, coeff_vec1);
75 
76  mask = LD_SB(&chroma_mask_arr[64]);
77 
78  LD_UB4(src, src_stride, src0, src1, src2, src3);
79 
80  VSHF_B2_UB(src0, src1, src2, src3, mask, mask, src0, src2);
81 
82  src0 = (v16u8) __msa_ilvr_d((v2i64) src2, (v2i64) src0);
83 
84  res_r = __msa_dotp_u_h(src0, coeff_vec);
85  res_r <<= 3;
86  res_r = (v8u16) __msa_srari_h((v8i16) res_r, 6);
87  res_r = __msa_sat_u_h(res_r, 7);
88  res = (v8i16) __msa_pckev_b((v16i8) res_r, (v16i8) res_r);
89 
90  ST2x4_UB(res, 0, dst, dst_stride);
91 }
92 
93 static void avc_chroma_hz_2x8_msa(uint8_t *src, int32_t src_stride,
94  uint8_t *dst, int32_t dst_stride,
95  uint32_t coeff0, uint32_t coeff1)
96 {
97  v16u8 src0, src1, src2, src3, src4, src5, src6, src7;
98  v8u16 res_r;
99  v8i16 res;
100  v16i8 mask;
101  v16i8 coeff_vec0 = __msa_fill_b(coeff0);
102  v16i8 coeff_vec1 = __msa_fill_b(coeff1);
103  v16u8 coeff_vec = (v16u8) __msa_ilvr_b(coeff_vec0, coeff_vec1);
104 
105  mask = LD_SB(&chroma_mask_arr[64]);
106 
107  LD_UB8(src, src_stride, src0, src1, src2, src3, src4, src5, src6, src7);
108 
109  VSHF_B2_UB(src0, src1, src2, src3, mask, mask, src0, src2);
110  VSHF_B2_UB(src4, src5, src6, src7, mask, mask, src4, src6);
111 
112  ILVR_D2_UB(src2, src0, src6, src4, src0, src4);
113 
114  res_r = __msa_dotp_u_h(src0, coeff_vec);
115  res_r <<= 3;
116  res_r = (v8u16) __msa_srari_h((v8i16) res_r, 6);
117  res_r = __msa_sat_u_h(res_r, 7);
118  res = (v8i16) __msa_pckev_b((v16i8) res_r, (v16i8) res_r);
119 
120  ST2x4_UB(res, 0, dst, dst_stride);
121  dst += (4 * dst_stride);
122 
123  res_r = __msa_dotp_u_h(src4, coeff_vec);
124  res_r <<= 3;
125  res_r = (v8u16) __msa_srari_h((v8i16) res_r, 6);
126  res_r = __msa_sat_u_h(res_r, 7);
127  res = (v8i16) __msa_pckev_b((v16i8) res_r, (v16i8) res_r);
128 
129  ST2x4_UB(res, 0, dst, dst_stride);
130 }
131 
132 static void avc_chroma_hz_2w_msa(uint8_t *src, int32_t src_stride,
133  uint8_t *dst, int32_t dst_stride,
134  uint32_t coeff0, uint32_t coeff1,
135  int32_t height)
136 {
137  if (2 == height) {
138  avc_chroma_hz_2x2_msa(src, src_stride, dst, dst_stride, coeff0, coeff1);
139  } else if (4 == height) {
140  avc_chroma_hz_2x4_msa(src, src_stride, dst, dst_stride, coeff0, coeff1);
141  } else if (8 == height) {
142  avc_chroma_hz_2x8_msa(src, src_stride, dst, dst_stride, coeff0, coeff1);
143  }
144 }
145 
146 static void avc_chroma_hz_4x2_msa(uint8_t *src, int32_t src_stride,
147  uint8_t *dst, int32_t dst_stride,
148  uint32_t coeff0, uint32_t coeff1)
149 {
150  v16i8 src0, src1;
151  v8u16 res_r;
152  v4i32 res;
153  v16i8 mask;
154  v16i8 coeff_vec0 = __msa_fill_b(coeff0);
155  v16i8 coeff_vec1 = __msa_fill_b(coeff1);
156  v16u8 coeff_vec = (v16u8) __msa_ilvr_b(coeff_vec0, coeff_vec1);
157 
158  mask = LD_SB(&chroma_mask_arr[0]);
159 
160  LD_SB2(src, src_stride, src0, src1);
161 
162  src0 = __msa_vshf_b(mask, src1, src0);
163  res_r = __msa_dotp_u_h((v16u8) src0, coeff_vec);
164  res_r <<= 3;
165  res_r = (v8u16) __msa_srari_h((v8i16) res_r, 6);
166  res_r = __msa_sat_u_h(res_r, 7);
167  res = (v4i32) __msa_pckev_b((v16i8) res_r, (v16i8) res_r);
168 
169  ST4x2_UB(res, dst, dst_stride);
170 }
171 
172 static void avc_chroma_hz_4x4multiple_msa(uint8_t *src, int32_t src_stride,
173  uint8_t *dst, int32_t dst_stride,
174  uint32_t coeff0, uint32_t coeff1,
175  int32_t height)
176 {
177  uint32_t row;
178  v16u8 src0, src1, src2, src3;
179  v8u16 res0_r, res1_r;
180  v4i32 res0, res1;
181  v16i8 mask;
182  v16i8 coeff_vec0 = __msa_fill_b(coeff0);
183  v16i8 coeff_vec1 = __msa_fill_b(coeff1);
184  v16u8 coeff_vec = (v16u8) __msa_ilvr_b(coeff_vec0, coeff_vec1);
185 
186  mask = LD_SB(&chroma_mask_arr[0]);
187 
188  for (row = (height >> 2); row--;) {
189  LD_UB4(src, src_stride, src0, src1, src2, src3);
190  src += (4 * src_stride);
191 
192  VSHF_B2_UB(src0, src1, src2, src3, mask, mask, src0, src2);
193  DOTP_UB2_UH(src0, src2, coeff_vec, coeff_vec, res0_r, res1_r);
194 
195  res0_r <<= 3;
196  res1_r <<= 3;
197 
198  SRARI_H2_UH(res0_r, res1_r, 6);
199  SAT_UH2_UH(res0_r, res1_r, 7);
200  PCKEV_B2_SW(res0_r, res0_r, res1_r, res1_r, res0, res1);
201 
202  ST4x4_UB(res0, res1, 0, 1, 0, 1, dst, dst_stride);
203  dst += (4 * dst_stride);
204  }
205 }
206 
207 static void avc_chroma_hz_4w_msa(uint8_t *src, int32_t src_stride,
208  uint8_t *dst, int32_t dst_stride,
209  uint32_t coeff0, uint32_t coeff1,
210  int32_t height)
211 {
212  if (2 == height) {
213  avc_chroma_hz_4x2_msa(src, src_stride, dst, dst_stride, coeff0, coeff1);
214  } else {
215  avc_chroma_hz_4x4multiple_msa(src, src_stride, dst, dst_stride, coeff0,
216  coeff1, height);
217  }
218 }
219 
220 static void avc_chroma_hz_8w_msa(uint8_t *src, int32_t src_stride,
221  uint8_t *dst, int32_t dst_stride,
222  uint32_t coeff0, uint32_t coeff1,
223  int32_t height)
224 {
225  uint32_t row;
226  v16u8 src0, src1, src2, src3, out0, out1;
227  v8u16 res0, res1, res2, res3;
228  v16i8 mask;
229  v16i8 coeff_vec0 = __msa_fill_b(coeff0);
230  v16i8 coeff_vec1 = __msa_fill_b(coeff1);
231  v16u8 coeff_vec = (v16u8) __msa_ilvr_b(coeff_vec0, coeff_vec1);
232 
233  mask = LD_SB(&chroma_mask_arr[32]);
234 
235  for (row = height >> 2; row--;) {
236  LD_UB4(src, src_stride, src0, src1, src2, src3);
237  src += (4 * src_stride);
238 
239  VSHF_B2_UB(src0, src0, src1, src1, mask, mask, src0, src1);
240  VSHF_B2_UB(src2, src2, src3, src3, mask, mask, src2, src3);
241  DOTP_UB4_UH(src0, src1, src2, src3, coeff_vec, coeff_vec, coeff_vec,
242  coeff_vec, res0, res1, res2, res3);
243  SLLI_4V(res0, res1, res2, res3, 3);
244  SRARI_H4_UH(res0, res1, res2, res3, 6);
245  SAT_UH4_UH(res0, res1, res2, res3, 7);
246  PCKEV_B2_UB(res1, res0, res3, res2, out0, out1);
247  ST8x4_UB(out0, out1, dst, dst_stride);
248  dst += (4 * dst_stride);
249  }
250 
251  if (0 != (height % 4)) {
252  for (row = (height % 4); row--;) {
253  src0 = LD_UB(src);
254  src += src_stride;
255 
256  src0 = (v16u8) __msa_vshf_b(mask, (v16i8) src0, (v16i8) src0);
257 
258  res0 = __msa_dotp_u_h(src0, coeff_vec);
259  res0 <<= 3;
260  res0 = (v8u16) __msa_srari_h((v8i16) res0, 6);
261  res0 = __msa_sat_u_h(res0, 7);
262  res0 = (v8u16) __msa_pckev_b((v16i8) res0, (v16i8) res0);
263 
264  ST8x1_UB(res0, dst);
265  dst += dst_stride;
266  }
267  }
268 }
269 
270 static void avc_chroma_vt_2x2_msa(uint8_t *src, int32_t src_stride,
271  uint8_t *dst, int32_t dst_stride,
272  uint32_t coeff0, uint32_t coeff1)
273 {
274  uint16_t out0, out1;
275  v16i8 src0, src1, src2;
276  v16u8 tmp0, tmp1;
277  v8i16 res;
278  v8u16 res_r;
279  v16i8 coeff_vec0 = __msa_fill_b(coeff0);
280  v16i8 coeff_vec1 = __msa_fill_b(coeff1);
281  v16u8 coeff_vec = (v16u8) __msa_ilvr_b(coeff_vec0, coeff_vec1);
282 
283  LD_SB3(src, src_stride, src0, src1, src2);
284 
285  ILVR_B2_UB(src1, src0, src2, src1, tmp0, tmp1);
286 
287  tmp0 = (v16u8) __msa_ilvr_d((v2i64) tmp1, (v2i64) tmp0);
288 
289  res_r = __msa_dotp_u_h(tmp0, coeff_vec);
290  res_r <<= 3;
291  res_r = (v8u16) __msa_srari_h((v8i16) res_r, 6);
292  res_r = __msa_sat_u_h(res_r, 7);
293  res = (v8i16) __msa_pckev_b((v16i8) res_r, (v16i8) res_r);
294 
295  out0 = __msa_copy_u_h(res, 0);
296  out1 = __msa_copy_u_h(res, 2);
297 
298  SH(out0, dst);
299  dst += dst_stride;
300  SH(out1, dst);
301 }
302 
303 static void avc_chroma_vt_2x4_msa(uint8_t *src, int32_t src_stride,
304  uint8_t *dst, int32_t dst_stride,
305  uint32_t coeff0, uint32_t coeff1)
306 {
307  v16u8 src0, src1, src2, src3, src4;
308  v16u8 tmp0, tmp1, tmp2, tmp3;
309  v8i16 res;
310  v8u16 res_r;
311  v16i8 coeff_vec0 = __msa_fill_b(coeff0);
312  v16i8 coeff_vec1 = __msa_fill_b(coeff1);
313  v16u8 coeff_vec = (v16u8) __msa_ilvr_b(coeff_vec0, coeff_vec1);
314 
315  LD_UB5(src, src_stride, src0, src1, src2, src3, src4);
316  ILVR_B4_UB(src1, src0, src2, src1, src3, src2, src4, src3,
317  tmp0, tmp1, tmp2, tmp3);
318  ILVR_W2_UB(tmp1, tmp0, tmp3, tmp2, tmp0, tmp2);
319 
320  tmp0 = (v16u8) __msa_ilvr_d((v2i64) tmp2, (v2i64) tmp0);
321 
322  res_r = __msa_dotp_u_h(tmp0, coeff_vec);
323  res_r <<= 3;
324  res_r = (v8u16) __msa_srari_h((v8i16) res_r, 6);
325  res_r = __msa_sat_u_h(res_r, 7);
326 
327  res = (v8i16) __msa_pckev_b((v16i8) res_r, (v16i8) res_r);
328 
329  ST2x4_UB(res, 0, dst, dst_stride);
330 }
331 
332 static void avc_chroma_vt_2x8_msa(uint8_t *src, int32_t src_stride,
333  uint8_t *dst, int32_t dst_stride,
334  uint32_t coeff0, uint32_t coeff1)
335 {
336  v16u8 src0, src1, src2, src3, src4, src5, src6, src7, src8;
337  v16u8 tmp0, tmp1, tmp2, tmp3;
338  v8i16 res;
339  v8u16 res_r;
340  v16i8 coeff_vec0 = __msa_fill_b(coeff0);
341  v16i8 coeff_vec1 = __msa_fill_b(coeff1);
342  v16u8 coeff_vec = (v16u8) __msa_ilvr_b(coeff_vec0, coeff_vec1);
343 
344  LD_UB5(src, src_stride, src0, src1, src2, src3, src4);
345  src += (5 * src_stride);
346  LD_UB4(src, src_stride, src5, src6, src7, src8);
347 
348  ILVR_B4_UB(src1, src0, src2, src1, src3, src2, src4, src3,
349  tmp0, tmp1, tmp2, tmp3);
350  ILVR_W2_UB(tmp1, tmp0, tmp3, tmp2, tmp0, tmp2);
351 
352  tmp0 = (v16u8) __msa_ilvr_d((v2i64) tmp2, (v2i64) tmp0);
353 
354  res_r = __msa_dotp_u_h(tmp0, coeff_vec);
355  res_r <<= 3;
356  res_r = (v8u16) __msa_srari_h((v8i16) res_r, 6);
357  res_r = __msa_sat_u_h(res_r, 7);
358 
359  res = (v8i16) __msa_pckev_b((v16i8) res_r, (v16i8) res_r);
360 
361  ST2x4_UB(res, 0, dst, dst_stride);
362  dst += (4 * dst_stride);
363 
364  ILVR_B4_UB(src5, src4, src6, src5, src7, src6, src8, src7,
365  tmp0, tmp1, tmp2, tmp3);
366  ILVR_W2_UB(tmp1, tmp0, tmp3, tmp2, tmp0, tmp2);
367 
368  tmp0 = (v16u8) __msa_ilvr_d((v2i64) tmp2, (v2i64) tmp0);
369 
370  res_r = __msa_dotp_u_h(tmp0, coeff_vec);
371  res_r <<= 3;
372  res_r = (v8u16) __msa_srari_h((v8i16) res_r, 6);
373  res_r = __msa_sat_u_h(res_r, 7);
374 
375  res = (v8i16) __msa_pckev_b((v16i8) res_r, (v16i8) res_r);
376 
377  ST2x4_UB(res, 0, dst, dst_stride);
378  dst += (4 * dst_stride);
379 }
380 
381 static void avc_chroma_vt_2w_msa(uint8_t *src, int32_t src_stride,
382  uint8_t *dst, int32_t dst_stride,
383  uint32_t coeff0, uint32_t coeff1,
384  int32_t height)
385 {
386  if (2 == height) {
387  avc_chroma_vt_2x2_msa(src, src_stride, dst, dst_stride, coeff0, coeff1);
388  } else if (4 == height) {
389  avc_chroma_vt_2x4_msa(src, src_stride, dst, dst_stride, coeff0, coeff1);
390  } else if (8 == height) {
391  avc_chroma_vt_2x8_msa(src, src_stride, dst, dst_stride, coeff0, coeff1);
392  }
393 }
394 
395 static void avc_chroma_vt_4x2_msa(uint8_t *src, int32_t src_stride,
396  uint8_t *dst, int32_t dst_stride,
397  uint32_t coeff0, uint32_t coeff1)
398 {
399  v16u8 src0, src1, src2;
400  v16u8 tmp0, tmp1;
401  v4i32 res;
402  v8u16 res_r;
403  v16i8 coeff_vec0 = __msa_fill_b(coeff0);
404  v16i8 coeff_vec1 = __msa_fill_b(coeff1);
405  v16u8 coeff_vec = (v16u8) __msa_ilvr_b(coeff_vec0, coeff_vec1);
406 
407  LD_UB3(src, src_stride, src0, src1, src2);
408  ILVR_B2_UB(src1, src0, src2, src1, tmp0, tmp1);
409 
410  tmp0 = (v16u8) __msa_ilvr_d((v2i64) tmp1, (v2i64) tmp0);
411  res_r = __msa_dotp_u_h(tmp0, coeff_vec);
412  res_r <<= 3;
413  res_r = (v8u16) __msa_srari_h((v8i16) res_r, 6);
414  res_r = __msa_sat_u_h(res_r, 7);
415  res = (v4i32) __msa_pckev_b((v16i8) res_r, (v16i8) res_r);
416 
417  ST4x2_UB(res, dst, dst_stride);
418 }
419 
420 static void avc_chroma_vt_4x4multiple_msa(uint8_t *src, int32_t src_stride,
421  uint8_t *dst, int32_t dst_stride,
422  uint32_t coeff0, uint32_t coeff1,
423  int32_t height)
424 {
425  uint32_t row;
426  v16u8 src0, src1, src2, src3, src4;
427  v16u8 tmp0, tmp1, tmp2, tmp3;
428  v8u16 res0_r, res1_r;
429  v4i32 res0, res1;
430  v16i8 coeff_vec0 = __msa_fill_b(coeff0);
431  v16i8 coeff_vec1 = __msa_fill_b(coeff1);
432  v16u8 coeff_vec = (v16u8) __msa_ilvr_b(coeff_vec0, coeff_vec1);
433 
434  src0 = LD_UB(src);
435  src += src_stride;
436 
437  for (row = (height >> 2); row--;) {
438  LD_UB4(src, src_stride, src1, src2, src3, src4);
439  src += (4 * src_stride);
440 
441  ILVR_B4_UB(src1, src0, src2, src1, src3, src2, src4, src3,
442  tmp0, tmp1, tmp2, tmp3);
443  ILVR_D2_UB(tmp1, tmp0, tmp3, tmp2, tmp0, tmp2);
444  DOTP_UB2_UH(tmp0, tmp2, coeff_vec, coeff_vec, res0_r, res1_r);
445 
446  res0_r <<= 3;
447  res1_r <<= 3;
448 
449  SRARI_H2_UH(res0_r, res1_r, 6);
450  SAT_UH2_UH(res0_r, res1_r, 7);
451  PCKEV_B2_SW(res0_r, res0_r, res1_r, res1_r, res0, res1);
452 
453  ST4x4_UB(res0, res1, 0, 1, 0, 1, dst, dst_stride);
454  dst += (4 * dst_stride);
455  src0 = src4;
456  }
457 }
458 
459 static void avc_chroma_vt_4w_msa(uint8_t *src, int32_t src_stride,
460  uint8_t *dst, int32_t dst_stride,
461  uint32_t coeff0, uint32_t coeff1,
462  int32_t height)
463 {
464  if (2 == height) {
465  avc_chroma_vt_4x2_msa(src, src_stride, dst, dst_stride, coeff0, coeff1);
466  } else {
467  avc_chroma_vt_4x4multiple_msa(src, src_stride, dst, dst_stride, coeff0,
468  coeff1, height);
469  }
470 }
471 
472 static void avc_chroma_vt_8w_msa(uint8_t *src, int32_t src_stride,
473  uint8_t *dst, int32_t dst_stride,
474  uint32_t coeff0, uint32_t coeff1,
475  int32_t height)
476 {
477  uint32_t row;
478  v16u8 src0, src1, src2, src3, src4, out0, out1;
479  v8u16 res0, res1, res2, res3;
480  v16i8 coeff_vec0 = __msa_fill_b(coeff0);
481  v16i8 coeff_vec1 = __msa_fill_b(coeff1);
482  v16u8 coeff_vec = (v16u8) __msa_ilvr_b(coeff_vec0, coeff_vec1);
483 
484  src0 = LD_UB(src);
485  src += src_stride;
486 
487  for (row = height >> 2; row--;) {
488  LD_UB4(src, src_stride, src1, src2, src3, src4);
489  src += (4 * src_stride);
490 
491  ILVR_B4_UB(src1, src0, src2, src1, src3, src2, src4, src3,
492  src0, src1, src2, src3);
493  DOTP_UB4_UH(src0, src1, src2, src3, coeff_vec, coeff_vec, coeff_vec,
494  coeff_vec, res0, res1, res2, res3);
495  SLLI_4V(res0, res1, res2, res3, 3);
496  SRARI_H4_UH(res0, res1, res2, res3, 6);
497  SAT_UH4_UH(res0, res1, res2, res3, 7);
498  PCKEV_B2_UB(res1, res0, res3, res2, out0, out1);
499 
500  ST8x4_UB(out0, out1, dst, dst_stride);
501 
502  dst += (4 * dst_stride);
503  src0 = src4;
504  }
505 }
506 
507 static void avc_chroma_hv_2x2_msa(uint8_t *src, int32_t src_stride,
508  uint8_t *dst, int32_t dst_stride,
509  uint32_t coef_hor0, uint32_t coef_hor1,
510  uint32_t coef_ver0, uint32_t coef_ver1)
511 {
512  uint16_t out0, out1;
513  v16u8 src0, src1, src2;
514  v8u16 res_hz0, res_hz1, res_vt0, res_vt1;
515  v8i16 res_vert;
516  v16i8 mask;
517  v16i8 coeff_hz_vec0 = __msa_fill_b(coef_hor0);
518  v16i8 coeff_hz_vec1 = __msa_fill_b(coef_hor1);
519  v16u8 coeff_hz_vec = (v16u8) __msa_ilvr_b(coeff_hz_vec0, coeff_hz_vec1);
520  v8u16 coeff_vt_vec0 = (v8u16) __msa_fill_h(coef_ver0);
521  v8u16 coeff_vt_vec1 = (v8u16) __msa_fill_h(coef_ver1);
522 
523  mask = LD_SB(&chroma_mask_arr[48]);
524 
525  LD_UB3(src, src_stride, src0, src1, src2);
526  VSHF_B2_UB(src0, src1, src1, src2, mask, mask, src0, src1);
527  DOTP_UB2_UH(src0, src1, coeff_hz_vec, coeff_hz_vec, res_hz0, res_hz1);
528  MUL2(res_hz0, coeff_vt_vec1, res_hz1, coeff_vt_vec0, res_vt0, res_vt1);
529 
530  res_vt0 += res_vt1;
531  res_vt0 = (v8u16) __msa_srari_h((v8i16) res_vt0, 6);
532  res_vt0 = __msa_sat_u_h(res_vt0, 7);
533  res_vert = (v8i16) __msa_pckev_b((v16i8) res_vt0, (v16i8) res_vt0);
534 
535  out0 = __msa_copy_u_h(res_vert, 0);
536  out1 = __msa_copy_u_h(res_vert, 1);
537 
538  SH(out0, dst);
539  dst += dst_stride;
540  SH(out1, dst);
541 }
542 
543 static void avc_chroma_hv_2x4_msa(uint8_t *src, int32_t src_stride,
544  uint8_t *dst, int32_t dst_stride,
545  uint32_t coef_hor0, uint32_t coef_hor1,
546  uint32_t coef_ver0, uint32_t coef_ver1)
547 {
548  v16u8 src0, src1, src2, src3, src4;
549  v16u8 tmp0, tmp1, tmp2, tmp3;
550  v8u16 res_hz0, res_hz1, res_vt0, res_vt1;
551  v8i16 res;
552  v16i8 mask;
553  v16i8 coeff_hz_vec0 = __msa_fill_b(coef_hor0);
554  v16i8 coeff_hz_vec1 = __msa_fill_b(coef_hor1);
555  v16u8 coeff_hz_vec = (v16u8) __msa_ilvr_b(coeff_hz_vec0, coeff_hz_vec1);
556  v8u16 coeff_vt_vec0 = (v8u16) __msa_fill_h(coef_ver0);
557  v8u16 coeff_vt_vec1 = (v8u16) __msa_fill_h(coef_ver1);
558 
559  mask = LD_SB(&chroma_mask_arr[48]);
560 
561  LD_UB5(src, src_stride, src0, src1, src2, src3, src4);
562 
563  VSHF_B2_UB(src0, src1, src2, src3, mask, mask, tmp0, tmp1);
564  VSHF_B2_UB(src1, src2, src3, src4, mask, mask, tmp2, tmp3);
565  ILVR_D2_UB(tmp1, tmp0, tmp3, tmp2, src0, src1);
566  DOTP_UB2_UH(src0, src1, coeff_hz_vec, coeff_hz_vec, res_hz0, res_hz1);
567  MUL2(res_hz0, coeff_vt_vec1, res_hz1, coeff_vt_vec0, res_vt0, res_vt1);
568 
569  res_vt0 += res_vt1;
570  res_vt0 = (v8u16) __msa_srari_h((v8i16) res_vt0, 6);
571  res_vt0 = __msa_sat_u_h(res_vt0, 7);
572  res = (v8i16) __msa_pckev_b((v16i8) res_vt0, (v16i8) res_vt0);
573 
574  ST2x4_UB(res, 0, dst, dst_stride);
575 }
576 
577 static void avc_chroma_hv_2x8_msa(uint8_t *src, int32_t src_stride,
578  uint8_t *dst, int32_t dst_stride,
579  uint32_t coef_hor0, uint32_t coef_hor1,
580  uint32_t coef_ver0, uint32_t coef_ver1)
581 {
582  v16u8 src0, src1, src2, src3, src4, src5, src6, src7, src8;
583  v16u8 tmp0, tmp1, tmp2, tmp3;
584  v8u16 res_hz0, res_hz1, res_vt0, res_vt1;
585  v8i16 res;
586  v16i8 mask;
587  v16i8 coeff_hz_vec0 = __msa_fill_b(coef_hor0);
588  v16i8 coeff_hz_vec1 = __msa_fill_b(coef_hor1);
589  v16u8 coeff_hz_vec = (v16u8) __msa_ilvr_b(coeff_hz_vec0, coeff_hz_vec1);
590  v8u16 coeff_vt_vec0 = (v8u16) __msa_fill_h(coef_ver0);
591  v8u16 coeff_vt_vec1 = (v8u16) __msa_fill_h(coef_ver1);
592 
593  mask = LD_SB(&chroma_mask_arr[48]);
594 
595  LD_UB5(src, src_stride, src0, src1, src2, src3, src4);
596  src += (5 * src_stride);
597  LD_UB4(src, src_stride, src5, src6, src7, src8);
598 
599  VSHF_B2_UB(src0, src1, src2, src3, mask, mask, tmp0, tmp1);
600  VSHF_B2_UB(src1, src2, src3, src4, mask, mask, tmp2, tmp3);
601  ILVR_D2_UB(tmp1, tmp0, tmp3, tmp2, src0, src1);
602  VSHF_B2_UB(src4, src5, src6, src7, mask, mask, tmp0, tmp1);
603  VSHF_B2_UB(src5, src6, src7, src8, mask, mask, tmp2, tmp3);
604  ILVR_D2_UB(tmp1, tmp0, tmp3, tmp2, src4, src5);
605  DOTP_UB2_UH(src0, src1, coeff_hz_vec, coeff_hz_vec, res_hz0, res_hz1);
606  MUL2(res_hz0, coeff_vt_vec1, res_hz1, coeff_vt_vec0, res_vt0, res_vt1);
607 
608  res_vt0 += res_vt1;
609  res_vt0 = (v8u16) __msa_srari_h((v8i16) res_vt0, 6);
610  res_vt0 = __msa_sat_u_h(res_vt0, 7);
611 
612  res = (v8i16) __msa_pckev_b((v16i8) res_vt0, (v16i8) res_vt0);
613 
614  ST2x4_UB(res, 0, dst, dst_stride);
615  dst += (4 * dst_stride);
616 
617  DOTP_UB2_UH(src4, src5, coeff_hz_vec, coeff_hz_vec, res_hz0, res_hz1);
618  MUL2(res_hz0, coeff_vt_vec1, res_hz1, coeff_vt_vec0, res_vt0, res_vt1);
619 
620  res_vt0 += res_vt1;
621  res_vt0 = (v8u16) __msa_srari_h((v8i16) res_vt0, 6);
622  res_vt0 = __msa_sat_u_h(res_vt0, 7);
623 
624  res = (v8i16) __msa_pckev_b((v16i8) res_vt0, (v16i8) res_vt0);
625 
626  ST2x4_UB(res, 0, dst, dst_stride);
627 }
628 
629 static void avc_chroma_hv_2w_msa(uint8_t *src, int32_t src_stride,
630  uint8_t *dst, int32_t dst_stride,
631  uint32_t coef_hor0, uint32_t coef_hor1,
632  uint32_t coef_ver0, uint32_t coef_ver1,
633  int32_t height)
634 {
635  if (2 == height) {
636  avc_chroma_hv_2x2_msa(src, src_stride, dst, dst_stride, coef_hor0,
637  coef_hor1, coef_ver0, coef_ver1);
638  } else if (4 == height) {
639  avc_chroma_hv_2x4_msa(src, src_stride, dst, dst_stride, coef_hor0,
640  coef_hor1, coef_ver0, coef_ver1);
641  } else if (8 == height) {
642  avc_chroma_hv_2x8_msa(src, src_stride, dst, dst_stride, coef_hor0,
643  coef_hor1, coef_ver0, coef_ver1);
644  }
645 }
646 
647 static void avc_chroma_hv_4x2_msa(uint8_t *src, int32_t src_stride,
648  uint8_t *dst, int32_t dst_stride,
649  uint32_t coef_hor0, uint32_t coef_hor1,
650  uint32_t coef_ver0, uint32_t coef_ver1)
651 {
652  v16u8 src0, src1, src2;
653  v8u16 res_hz0, res_hz1, res_vt0, res_vt1;
654  v16i8 mask;
655  v4i32 res;
656  v16i8 coeff_hz_vec0 = __msa_fill_b(coef_hor0);
657  v16i8 coeff_hz_vec1 = __msa_fill_b(coef_hor1);
658  v16u8 coeff_hz_vec = (v16u8) __msa_ilvr_b(coeff_hz_vec0, coeff_hz_vec1);
659  v8u16 coeff_vt_vec0 = (v8u16) __msa_fill_h(coef_ver0);
660  v8u16 coeff_vt_vec1 = (v8u16) __msa_fill_h(coef_ver1);
661 
662  mask = LD_SB(&chroma_mask_arr[0]);
663  LD_UB3(src, src_stride, src0, src1, src2);
664  VSHF_B2_UB(src0, src1, src1, src2, mask, mask, src0, src1);
665  DOTP_UB2_UH(src0, src1, coeff_hz_vec, coeff_hz_vec, res_hz0, res_hz1);
666  MUL2(res_hz0, coeff_vt_vec1, res_hz1, coeff_vt_vec0, res_vt0, res_vt1);
667 
668  res_vt0 += res_vt1;
669  res_vt0 = (v8u16) __msa_srari_h((v8i16) res_vt0, 6);
670  res_vt0 = __msa_sat_u_h(res_vt0, 7);
671  res = (v4i32) __msa_pckev_b((v16i8) res_vt0, (v16i8) res_vt0);
672 
673  ST4x2_UB(res, dst, dst_stride);
674 }
675 
676 static void avc_chroma_hv_4x4multiple_msa(uint8_t *src, int32_t src_stride,
677  uint8_t *dst, int32_t dst_stride,
678  uint32_t coef_hor0,
679  uint32_t coef_hor1,
680  uint32_t coef_ver0,
681  uint32_t coef_ver1,
682  int32_t height)
683 {
684  uint32_t row;
685  v16u8 src0, src1, src2, src3, src4;
686  v8u16 res_hz0, res_hz1, res_hz2, res_hz3;
687  v8u16 res_vt0, res_vt1, res_vt2, res_vt3;
688  v16i8 mask;
689  v16i8 coeff_hz_vec0 = __msa_fill_b(coef_hor0);
690  v16i8 coeff_hz_vec1 = __msa_fill_b(coef_hor1);
691  v16u8 coeff_hz_vec = (v16u8) __msa_ilvr_b(coeff_hz_vec0, coeff_hz_vec1);
692  v8u16 coeff_vt_vec0 = (v8u16) __msa_fill_h(coef_ver0);
693  v8u16 coeff_vt_vec1 = (v8u16) __msa_fill_h(coef_ver1);
694  v4i32 res0, res1;
695 
696  mask = LD_SB(&chroma_mask_arr[0]);
697 
698  src0 = LD_UB(src);
699  src += src_stride;
700 
701  for (row = (height >> 2); row--;) {
702  LD_UB4(src, src_stride, src1, src2, src3, src4);
703  src += (4 * src_stride);
704 
705  VSHF_B2_UB(src0, src1, src1, src2, mask, mask, src0, src1);
706  VSHF_B2_UB(src2, src3, src3, src4, mask, mask, src2, src3);
707  DOTP_UB4_UH(src0, src1, src2, src3, coeff_hz_vec, coeff_hz_vec,
708  coeff_hz_vec, coeff_hz_vec, res_hz0, res_hz1, res_hz2,
709  res_hz3);
710  MUL4(res_hz0, coeff_vt_vec1, res_hz1, coeff_vt_vec0, res_hz2,
711  coeff_vt_vec1, res_hz3, coeff_vt_vec0, res_vt0, res_vt1, res_vt2,
712  res_vt3);
713  ADD2(res_vt0, res_vt1, res_vt2, res_vt3, res_vt0, res_vt1);
714  SRARI_H2_UH(res_vt0, res_vt1, 6);
715  SAT_UH2_UH(res_vt0, res_vt1, 7);
716  PCKEV_B2_SW(res_vt0, res_vt0, res_vt1, res_vt1, res0, res1);
717 
718  ST4x4_UB(res0, res1, 0, 1, 0, 1, dst, dst_stride);
719  dst += (4 * dst_stride);
720  src0 = src4;
721  }
722 }
723 
724 static void avc_chroma_hv_4w_msa(uint8_t *src, int32_t src_stride,
725  uint8_t *dst, int32_t dst_stride,
726  uint32_t coef_hor0, uint32_t coef_hor1,
727  uint32_t coef_ver0, uint32_t coef_ver1,
728  int32_t height)
729 {
730  if (2 == height) {
731  avc_chroma_hv_4x2_msa(src, src_stride, dst, dst_stride, coef_hor0,
732  coef_hor1, coef_ver0, coef_ver1);
733  } else {
734  avc_chroma_hv_4x4multiple_msa(src, src_stride, dst, dst_stride,
735  coef_hor0, coef_hor1, coef_ver0,
736  coef_ver1, height);
737  }
738 }
739 
740 static void avc_chroma_hv_8w_msa(uint8_t *src, int32_t src_stride,
741  uint8_t *dst, int32_t dst_stride,
742  uint32_t coef_hor0, uint32_t coef_hor1,
743  uint32_t coef_ver0, uint32_t coef_ver1,
744  int32_t height)
745 {
746  uint32_t row;
747  v16u8 src0, src1, src2, src3, src4, out0, out1;
748  v8u16 res_hz0, res_hz1, res_hz2, res_hz3, res_hz4;
749  v8u16 res_vt0, res_vt1, res_vt2, res_vt3;
750  v16i8 mask;
751  v16i8 coeff_hz_vec0 = __msa_fill_b(coef_hor0);
752  v16i8 coeff_hz_vec1 = __msa_fill_b(coef_hor1);
753  v16u8 coeff_hz_vec = (v16u8) __msa_ilvr_b(coeff_hz_vec0, coeff_hz_vec1);
754  v8u16 coeff_vt_vec0 = (v8u16) __msa_fill_h(coef_ver0);
755  v8u16 coeff_vt_vec1 = (v8u16) __msa_fill_h(coef_ver1);
756 
757  mask = LD_SB(&chroma_mask_arr[32]);
758 
759  src0 = LD_UB(src);
760  src += src_stride;
761 
762  src0 = (v16u8) __msa_vshf_b(mask, (v16i8) src0, (v16i8) src0);
763  res_hz0 = __msa_dotp_u_h(src0, coeff_hz_vec);
764 
765  for (row = (height >> 2); row--;) {
766  LD_UB4(src, src_stride, src1, src2, src3, src4);
767  src += (4 * src_stride);
768 
769  VSHF_B2_UB(src1, src1, src2, src2, mask, mask, src1, src2);
770  VSHF_B2_UB(src3, src3, src4, src4, mask, mask, src3, src4);
771  DOTP_UB4_UH(src1, src2, src3, src4, coeff_hz_vec, coeff_hz_vec,
772  coeff_hz_vec, coeff_hz_vec, res_hz1, res_hz2, res_hz3,
773  res_hz4);
774  MUL4(res_hz1, coeff_vt_vec0, res_hz2, coeff_vt_vec0, res_hz3,
775  coeff_vt_vec0, res_hz4, coeff_vt_vec0, res_vt0, res_vt1, res_vt2,
776  res_vt3);
777 
778  res_vt0 += (res_hz0 * coeff_vt_vec1);
779  res_vt1 += (res_hz1 * coeff_vt_vec1);
780  res_vt2 += (res_hz2 * coeff_vt_vec1);
781  res_vt3 += (res_hz3 * coeff_vt_vec1);
782 
783  SRARI_H4_UH(res_vt0, res_vt1, res_vt2, res_vt3, 6);
784  SAT_UH4_UH(res_vt0, res_vt1, res_vt2, res_vt3, 7);
785  PCKEV_B2_UB(res_vt1, res_vt0, res_vt3, res_vt2, out0, out1);
786  ST8x4_UB(out0, out1, dst, dst_stride);
787 
788  dst += (4 * dst_stride);
789 
790  res_hz0 = res_hz4;
791  }
792 }
793 
794 static void avc_chroma_hz_and_aver_dst_2x2_msa(uint8_t *src, int32_t src_stride,
795  uint8_t *dst, int32_t dst_stride,
796  uint32_t coeff0, uint32_t coeff1)
797 {
798  uint16_t out0, out1;
799  uint32_t load0, load1;
800  v16i8 src0, src1;
801  v16u8 dst_data = { 0 };
802  v8u16 res_r;
803  v16u8 res;
804  v16i8 mask;
805  v16i8 coeff_vec0 = __msa_fill_b(coeff0);
806  v16i8 coeff_vec1 = __msa_fill_b(coeff1);
807  v16u8 coeff_vec = (v16u8) __msa_ilvr_b(coeff_vec0, coeff_vec1);
808 
809  mask = LD_SB(&chroma_mask_arr[0]);
810 
811  LD_SB2(src, src_stride, src0, src1);
812 
813  load0 = LW(dst);
814  load1 = LW(dst + dst_stride);
815 
816  INSERT_W2_UB(load0, load1, dst_data);
817 
818  src0 = __msa_vshf_b(mask, src1, src0);
819 
820  res_r = __msa_dotp_u_h((v16u8) src0, coeff_vec);
821  res_r <<= 3;
822  res_r = (v8u16) __msa_srari_h((v8i16) res_r, 6);
823  res_r = __msa_sat_u_h(res_r, 7);
824 
825  res = (v16u8) __msa_pckev_b((v16i8) res_r, (v16i8) res_r);
826  dst_data = __msa_aver_u_b(res, dst_data);
827 
828  out0 = __msa_copy_u_h((v8i16) dst_data, 0);
829  out1 = __msa_copy_u_h((v8i16) dst_data, 2);
830 
831  SH(out0, dst);
832  dst += dst_stride;
833  SH(out1, dst);
834 }
835 
836 static void avc_chroma_hz_and_aver_dst_2x4_msa(uint8_t *src, int32_t src_stride,
837  uint8_t *dst, int32_t dst_stride,
838  uint32_t coeff0, uint32_t coeff1)
839 {
840  v16u8 src0, src1, src2, src3;
841  v16u8 dst0, dst1, dst2, dst3;
842  v8u16 res_r;
843  v16i8 res, mask;
844  v16i8 coeff_vec0 = __msa_fill_b(coeff0);
845  v16i8 coeff_vec1 = __msa_fill_b(coeff1);
846  v16u8 coeff_vec = (v16u8) __msa_ilvr_b(coeff_vec0, coeff_vec1);
847 
848  mask = LD_SB(&chroma_mask_arr[64]);
849 
850  LD_UB4(src, src_stride, src0, src1, src2, src3);
851  LD_UB4(dst, dst_stride, dst0, dst1, dst2, dst3);
852 
853  dst0 = (v16u8) __msa_insve_h((v8i16) dst0, 1, (v8i16) dst1);
854  dst0 = (v16u8) __msa_insve_h((v8i16) dst0, 2, (v8i16) dst2);
855  dst0 = (v16u8) __msa_insve_h((v8i16) dst0, 3, (v8i16) dst3);
856 
857  VSHF_B2_UB(src0, src1, src2, src3, mask, mask, src0, src2);
858 
859  src0 = (v16u8) __msa_ilvr_d((v2i64) src2, (v2i64) src0);
860 
861  res_r = __msa_dotp_u_h(src0, coeff_vec);
862  res_r <<= 3;
863  res_r = (v8u16) __msa_srari_h((v8i16) res_r, 6);
864  res_r = __msa_sat_u_h(res_r, 7);
865 
866  res = __msa_pckev_b((v16i8) res_r, (v16i8) res_r);
867  dst0 = __msa_aver_u_b((v16u8) res, dst0);
868 
869  ST2x4_UB(dst0, 0, dst, dst_stride);
870 }
871 
872 static void avc_chroma_hz_and_aver_dst_2x8_msa(uint8_t *src, int32_t src_stride,
873  uint8_t *dst, int32_t dst_stride,
874  uint32_t coeff0, uint32_t coeff1)
875 {
876  v16u8 src0, src1, src2, src3, src4, src5, src6, src7;
877  v16u8 dst0, dst1, dst2, dst3, dst4, dst5, dst6, dst7;
878  v8u16 res0_r, res1_r;
879  v16u8 res0, res1, mask;
880  v16i8 coeff_vec0 = __msa_fill_b(coeff0);
881  v16i8 coeff_vec1 = __msa_fill_b(coeff1);
882  v16u8 coeff_vec = (v16u8) __msa_ilvr_b(coeff_vec0, coeff_vec1);
883 
884  mask = LD_UB(&chroma_mask_arr[64]);
885 
886  LD_UB8(src, src_stride, src0, src1, src2, src3, src4, src5, src6, src7);
887  LD_UB8(dst, dst_stride, dst0, dst1, dst2, dst3, dst4, dst5, dst6, dst7);
888 
889  dst0 = (v16u8) __msa_insve_h((v8i16) dst0, 1, (v8i16) dst1);
890  dst0 = (v16u8) __msa_insve_h((v8i16) dst0, 2, (v8i16) dst2);
891  dst0 = (v16u8) __msa_insve_h((v8i16) dst0, 3, (v8i16) dst3);
892 
893  dst4 = (v16u8) __msa_insve_h((v8i16) dst4, 1, (v8i16) dst5);
894  dst4 = (v16u8) __msa_insve_h((v8i16) dst4, 2, (v8i16) dst6);
895  dst4 = (v16u8) __msa_insve_h((v8i16) dst4, 3, (v8i16) dst7);
896 
897  VSHF_B2_UB(src0, src1, src2, src3, mask, mask, src0, src2);
898  VSHF_B2_UB(src4, src5, src6, src7, mask, mask, src4, src6);
899  ILVR_D2_UB(src2, src0, src6, src4, src0, src4);
900  DOTP_UB2_UH(src0, src4, coeff_vec, coeff_vec, res0_r, res1_r);
901 
902  res0_r <<= 3;
903  res1_r <<= 3;
904 
905  SRARI_H2_UH(res0_r, res1_r, 6);
906  SAT_UH2_UH(res0_r, res1_r, 7);
907  PCKEV_B2_UB(res0_r, res0_r, res1_r, res1_r, res0, res1);
908  AVER_UB2_UB(res0, dst0, res1, dst4, dst0, dst4);
909 
910  ST2x4_UB(dst0, 0, dst, dst_stride);
911  dst += (4 * dst_stride);
912  ST2x4_UB(dst4, 0, dst, dst_stride);
913 }
914 
915 static void avc_chroma_hz_and_aver_dst_2w_msa(uint8_t *src, int32_t src_stride,
916  uint8_t *dst, int32_t dst_stride,
917  uint32_t coeff0, uint32_t coeff1,
918  int32_t height)
919 {
920  if (2 == height) {
921  avc_chroma_hz_and_aver_dst_2x2_msa(src, src_stride, dst, dst_stride,
922  coeff0, coeff1);
923  } else if (4 == height) {
924  avc_chroma_hz_and_aver_dst_2x4_msa(src, src_stride, dst, dst_stride,
925  coeff0, coeff1);
926  } else if (8 == height) {
927  avc_chroma_hz_and_aver_dst_2x8_msa(src, src_stride, dst, dst_stride,
928  coeff0, coeff1);
929  }
930 }
931 
932 static void avc_chroma_hz_and_aver_dst_4x2_msa(uint8_t *src, int32_t src_stride,
933  uint8_t *dst, int32_t dst_stride,
934  uint32_t coeff0, uint32_t coeff1)
935 {
936  uint32_t load0, load1;
937  v16i8 src0, src1;
938  v16u8 dst_data = { 0 };
939  v8u16 res_r;
940  v16i8 res, mask;
941  v16i8 coeff_vec0 = __msa_fill_b(coeff0);
942  v16i8 coeff_vec1 = __msa_fill_b(coeff1);
943  v16u8 coeff_vec = (v16u8) __msa_ilvr_b(coeff_vec0, coeff_vec1);
944 
945  mask = LD_SB(&chroma_mask_arr[0]);
946 
947  LD_SB2(src, src_stride, src0, src1);
948 
949  load0 = LW(dst);
950  load1 = LW(dst + dst_stride);
951 
952  INSERT_W2_UB(load0, load1, dst_data);
953 
954  src0 = __msa_vshf_b(mask, src1, src0);
955 
956  res_r = __msa_dotp_u_h((v16u8) src0, coeff_vec);
957  res_r <<= 3;
958  res_r = (v8u16) __msa_srari_h((v8i16) res_r, 6);
959  res_r = __msa_sat_u_h(res_r, 7);
960  res = __msa_pckev_b((v16i8) res_r, (v16i8) res_r);
961  dst_data = __msa_aver_u_b((v16u8) res, dst_data);
962 
963  ST4x2_UB(dst_data, dst, dst_stride);
964 }
965 
966 static void avc_chroma_hz_and_aver_dst_4x4multiple_msa(uint8_t *src,
967  int32_t src_stride,
968  uint8_t *dst,
969  int32_t dst_stride,
970  uint32_t coeff0,
971  uint32_t coeff1,
972  int32_t height)
973 {
974  uint32_t load0, load1;
975  uint32_t row;
976  v16u8 src0, src1, src2, src3;
977  v16u8 dst0 = { 0 };
978  v16u8 dst1 = { 0 };
979  v8u16 res0_r, res1_r;
980  v16u8 res0, res1, mask;
981  v16i8 coeff_vec0 = __msa_fill_b(coeff0);
982  v16i8 coeff_vec1 = __msa_fill_b(coeff1);
983  v16u8 coeff_vec = (v16u8) __msa_ilvr_b(coeff_vec0, coeff_vec1);
984 
985  mask = LD_UB(&chroma_mask_arr[0]);
986 
987  for (row = (height >> 2); row--;) {
988  LD_UB4(src, src_stride, src0, src1, src2, src3);
989  src += (4 * src_stride);
990 
991  load0 = LW(dst);
992  load1 = LW(dst + dst_stride);
993 
994  INSERT_W2_UB(load0, load1, dst0);
995 
996  load0 = LW(dst + 2 * dst_stride);
997  load1 = LW(dst + 3 * dst_stride);
998 
999  INSERT_W2_UB(load0, load1, dst1);
1000 
1001  VSHF_B2_UB(src0, src1, src2, src3, mask, mask, src0, src2);
1002  DOTP_UB2_UH(src0, src2, coeff_vec, coeff_vec, res0_r, res1_r);
1003 
1004  res0_r <<= 3;
1005  res1_r <<= 3;
1006 
1007  SRARI_H2_UH(res0_r, res1_r, 6);
1008  SAT_UH2_UH(res0_r, res1_r, 7);
1009  PCKEV_B2_UB(res0_r, res0_r, res1_r, res1_r, res0, res1);
1010  AVER_UB2_UB(res0, dst0, res1, dst1, dst0, dst1);
1011 
1012  ST4x4_UB(dst0, dst1, 0, 1, 0, 1, dst, dst_stride);
1013  dst += (4 * dst_stride);
1014  }
1015 }
1016 
1017 static void avc_chroma_hz_and_aver_dst_4w_msa(uint8_t *src, int32_t src_stride,
1018  uint8_t *dst, int32_t dst_stride,
1019  uint32_t coeff0, uint32_t coeff1,
1020  int32_t height)
1021 {
1022  if (2 == height) {
1023  avc_chroma_hz_and_aver_dst_4x2_msa(src, src_stride, dst, dst_stride,
1024  coeff0, coeff1);
1025  } else {
1026  avc_chroma_hz_and_aver_dst_4x4multiple_msa(src, src_stride,
1027  dst, dst_stride,
1028  coeff0, coeff1, height);
1029  }
1030 }
1031 
1032 static void avc_chroma_hz_and_aver_dst_8w_msa(uint8_t *src, int32_t src_stride,
1033  uint8_t *dst, int32_t dst_stride,
1034  uint32_t coeff0, uint32_t coeff1,
1035  int32_t height)
1036 {
1037  uint32_t row;
1038  v16u8 src0, src1, src2, src3, out0, out1;
1039  v8u16 res0, res1, res2, res3;
1040  v16u8 dst0, dst1, dst2, dst3;
1041  v16i8 mask;
1042  v16i8 coeff_vec0 = __msa_fill_b(coeff0);
1043  v16i8 coeff_vec1 = __msa_fill_b(coeff1);
1044  v16u8 coeff_vec = (v16u8) __msa_ilvr_b(coeff_vec0, coeff_vec1);
1045 
1046  mask = LD_SB(&chroma_mask_arr[32]);
1047 
1048  for (row = height >> 2; row--;) {
1049  LD_UB4(src, src_stride, src0, src1, src2, src3);
1050  src += (4 * src_stride);
1051  LD_UB4(dst, dst_stride, dst0, dst1, dst2, dst3);
1052  VSHF_B2_UB(src0, src0, src1, src1, mask, mask, src0, src1);
1053  VSHF_B2_UB(src2, src2, src3, src3, mask, mask, src2, src3);
1054  DOTP_UB4_UH(src0, src1, src2, src3, coeff_vec, coeff_vec, coeff_vec,
1055  coeff_vec, res0, res1, res2, res3);
1056  SLLI_4V(res0, res1, res2, res3, 3);
1057  SRARI_H4_UH(res0, res1, res2, res3, 6);
1058  SAT_UH4_UH(res0, res1, res2, res3, 7);
1059  PCKEV_B2_UB(res1, res0, res3, res2, out0, out1);
1060  PCKEV_D2_UB(dst1, dst0, dst3, dst2, dst0, dst1);
1061  AVER_UB2_UB(out0, dst0, out1, dst1, out0, out1);
1062  ST8x4_UB(out0, out1, dst, dst_stride);
1063  dst += (4 * dst_stride);
1064  }
1065 }
1066 
1067 static void avc_chroma_vt_and_aver_dst_2x2_msa(uint8_t *src, int32_t src_stride,
1068  uint8_t *dst, int32_t dst_stride,
1069  uint32_t coeff0, uint32_t coeff1)
1070 {
1071  uint16_t out0, out1;
1072  uint32_t load0, load1;
1073  v16i8 src0, src1, src2, tmp0, tmp1, res;
1074  v16u8 dst_data = { 0 };
1075  v8u16 res_r;
1076  v16i8 coeff_vec0 = __msa_fill_b(coeff0);
1077  v16i8 coeff_vec1 = __msa_fill_b(coeff1);
1078  v16u8 coeff_vec = (v16u8) __msa_ilvr_b(coeff_vec0, coeff_vec1);
1079 
1080  LD_SB3(src, src_stride, src0, src1, src2);
1081  load0 = LW(dst);
1082  load1 = LW(dst + dst_stride);
1083 
1084  INSERT_W2_UB(load0, load1, dst_data);
1085 
1086  ILVR_B2_SB(src1, src0, src2, src1, tmp0, tmp1);
1087 
1088  tmp0 = (v16i8) __msa_ilvr_d((v2i64) tmp1, (v2i64) tmp0);
1089  res_r = __msa_dotp_u_h((v16u8) tmp0, coeff_vec);
1090  res_r <<= 3;
1091  res_r = (v8u16) __msa_srari_h((v8i16) res_r, 6);
1092  res_r = __msa_sat_u_h(res_r, 7);
1093  res = __msa_pckev_b((v16i8) res_r, (v16i8) res_r);
1094  dst_data = __msa_aver_u_b((v16u8) res, dst_data);
1095  out0 = __msa_copy_u_h((v8i16) dst_data, 0);
1096  out1 = __msa_copy_u_h((v8i16) dst_data, 2);
1097 
1098  SH(out0, dst);
1099  dst += dst_stride;
1100  SH(out1, dst);
1101 }
1102 
1103 static void avc_chroma_vt_and_aver_dst_2x4_msa(uint8_t *src, int32_t src_stride,
1104  uint8_t *dst, int32_t dst_stride,
1105  uint32_t coeff0, uint32_t coeff1)
1106 {
1107  uint32_t load0, load1;
1108  v16i8 src0, src1, src2, src3, src4;
1109  v16u8 tmp0, tmp1, tmp2, tmp3;
1110  v8u16 res_r;
1111  v8i16 res;
1112  v16i8 coeff_vec0 = __msa_fill_b(coeff0);
1113  v16i8 coeff_vec1 = __msa_fill_b(coeff1);
1114  v16u8 coeff_vec = (v16u8) __msa_ilvr_b(coeff_vec0, coeff_vec1);
1115  v16u8 dst_data = { 0 };
1116 
1117  LD_SB5(src, src_stride, src0, src1, src2, src3, src4);
1118 
1119  load0 = LW(dst);
1120  load1 = LW(dst + dst_stride);
1121 
1122  dst_data = (v16u8) __msa_insert_h((v8i16) dst_data, 0, load0);
1123  dst_data = (v16u8) __msa_insert_h((v8i16) dst_data, 1, load1);
1124 
1125  load0 = LW(dst + 2 * dst_stride);
1126  load1 = LW(dst + 3 * dst_stride);
1127 
1128  dst_data = (v16u8) __msa_insert_h((v8i16) dst_data, 2, load0);
1129  dst_data = (v16u8) __msa_insert_h((v8i16) dst_data, 3, load1);
1130 
1131  ILVR_B4_UB(src1, src0, src2, src1, src3, src2, src4, src3,
1132  tmp0, tmp1, tmp2, tmp3);
1133  ILVR_W2_UB(tmp1, tmp0, tmp3, tmp2, tmp0, tmp2);
1134 
1135  tmp0 = (v16u8) __msa_ilvr_d((v2i64) tmp2, (v2i64) tmp0);
1136 
1137  res_r = __msa_dotp_u_h(tmp0, coeff_vec);
1138  res_r <<= 3;
1139  res_r = (v8u16) __msa_srari_h((v8i16) res_r, 6);
1140  res_r = __msa_sat_u_h(res_r, 7);
1141 
1142  res = (v8i16) __msa_pckev_b((v16i8) res_r, (v16i8) res_r);
1143  res = (v8i16) __msa_aver_u_b((v16u8) res, dst_data);
1144 
1145  ST2x4_UB(res, 0, dst, dst_stride);
1146  dst += (4 * dst_stride);
1147 }
1148 
1149 static void avc_chroma_vt_and_aver_dst_2x8_msa(uint8_t *src, int32_t src_stride,
1150  uint8_t *dst, int32_t dst_stride,
1151  uint32_t coeff0, uint32_t coeff1)
1152 {
1153  uint32_t load0, load1, load2, load3;
1154  v16i8 src0, src1, src2, src3, src4, src5, src6, src7, src8;
1155  v16u8 tmp0, tmp1, tmp2, tmp3;
1156  v8i16 res;
1157  v8u16 res_r;
1158  v16i8 coeff_vec0 = __msa_fill_b(coeff0);
1159  v16i8 coeff_vec1 = __msa_fill_b(coeff1);
1160  v16u8 coeff_vec = (v16u8) __msa_ilvr_b(coeff_vec0, coeff_vec1);
1161  v16u8 dst_data0 = { 0 };
1162  v16u8 dst_data1 = { 0 };
1163 
1164  LD_SB5(src, src_stride, src0, src1, src2, src3, src4);
1165  src += (5 * src_stride);
1166  LD_SB4(src, src_stride, src5, src6, src7, src8);
1167 
1168  LW4(dst, dst_stride, load0, load1, load2, load3);
1169 
1170  dst_data0 = (v16u8) __msa_insert_h((v8i16) dst_data0, 0, load0);
1171  dst_data0 = (v16u8) __msa_insert_h((v8i16) dst_data0, 1, load1);
1172  dst_data0 = (v16u8) __msa_insert_h((v8i16) dst_data0, 2, load2);
1173  dst_data0 = (v16u8) __msa_insert_h((v8i16) dst_data0, 3, load3);
1174 
1175  LW4(dst + 4 * dst_stride, dst_stride, load0, load1, load2, load3);
1176 
1177  dst_data1 = (v16u8) __msa_insert_h((v8i16) dst_data1, 0, load0);
1178  dst_data1 = (v16u8) __msa_insert_h((v8i16) dst_data1, 1, load1);
1179  dst_data1 = (v16u8) __msa_insert_h((v8i16) dst_data1, 2, load2);
1180  dst_data1 = (v16u8) __msa_insert_h((v8i16) dst_data1, 3, load3);
1181 
1182  ILVR_B4_UB(src1, src0, src2, src1, src3, src2, src4, src3,
1183  tmp0, tmp1, tmp2, tmp3);
1184 
1185  ILVR_W2_UB(tmp1, tmp0, tmp3, tmp2, tmp0, tmp2);
1186 
1187  tmp0 = (v16u8) __msa_ilvr_d((v2i64) tmp2, (v2i64) tmp0);
1188 
1189  res_r = __msa_dotp_u_h(tmp0, coeff_vec);
1190  res_r <<= 3;
1191  res_r = (v8u16) __msa_srari_h((v8i16) res_r, 6);
1192  res_r = __msa_sat_u_h(res_r, 7);
1193 
1194  res = (v8i16) __msa_pckev_b((v16i8) res_r, (v16i8) res_r);
1195  res = (v8i16) __msa_aver_u_b((v16u8) res, dst_data0);
1196 
1197  ST2x4_UB(res, 0, dst, dst_stride);
1198  dst += (4 * dst_stride);
1199 
1200  ILVR_B4_UB(src5, src4, src6, src5, src7, src6, src8, src7,
1201  tmp0, tmp1, tmp2, tmp3);
1202 
1203  ILVR_W2_UB(tmp1, tmp0, tmp3, tmp2, tmp0, tmp2);
1204 
1205  tmp0 = (v16u8) __msa_ilvr_d((v2i64) tmp2, (v2i64) tmp0);
1206 
1207  res_r = __msa_dotp_u_h(tmp0, coeff_vec);
1208  res_r <<= 3;
1209  res_r = (v8u16) __msa_srari_h((v8i16) res_r, 6);
1210  res_r = __msa_sat_u_h(res_r, 7);
1211 
1212  res = (v8i16) __msa_pckev_b((v16i8) res_r, (v16i8) res_r);
1213  res = (v8i16) __msa_aver_u_b((v16u8) res, dst_data1);
1214 
1215  ST2x4_UB(res, 0, dst, dst_stride);
1216 }
1217 
1218 static void avc_chroma_vt_and_aver_dst_2w_msa(uint8_t *src, int32_t src_stride,
1219  uint8_t *dst, int32_t dst_stride,
1220  uint32_t coeff0, uint32_t coeff1,
1221  int32_t height)
1222 {
1223  if (2 == height) {
1224  avc_chroma_vt_and_aver_dst_2x2_msa(src, src_stride, dst, dst_stride,
1225  coeff0, coeff1);
1226  } else if (4 == height) {
1227  avc_chroma_vt_and_aver_dst_2x4_msa(src, src_stride, dst, dst_stride,
1228  coeff0, coeff1);
1229  } else if (8 == height) {
1230  avc_chroma_vt_and_aver_dst_2x8_msa(src, src_stride, dst, dst_stride,
1231  coeff0, coeff1);
1232  }
1233 }
1234 
1235 static void avc_chroma_vt_and_aver_dst_4x2_msa(uint8_t *src, int32_t src_stride,
1236  uint8_t *dst, int32_t dst_stride,
1237  uint32_t coeff0, uint32_t coeff1)
1238 {
1239  uint32_t load0, load1;
1240  v16i8 src0, src1, src2, tmp0, tmp1;
1241  v16u8 dst_data = { 0 };
1242  v8u16 res_r;
1243  v16u8 res;
1244  v16i8 coeff_vec0 = __msa_fill_b(coeff0);
1245  v16i8 coeff_vec1 = __msa_fill_b(coeff1);
1246  v16u8 coeff_vec = (v16u8) __msa_ilvr_b(coeff_vec0, coeff_vec1);
1247 
1248  LD_SB3(src, src_stride, src0, src1, src2);
1249 
1250  load0 = LW(dst);
1251  load1 = LW(dst + dst_stride);
1252 
1253  INSERT_W2_UB(load0, load1, dst_data);
1254  ILVR_B2_SB(src1, src0, src2, src1, tmp0, tmp1);
1255 
1256  tmp0 = (v16i8) __msa_ilvr_d((v2i64) tmp1, (v2i64) tmp0);
1257 
1258  res_r = __msa_dotp_u_h((v16u8) tmp0, coeff_vec);
1259  res_r <<= 3;
1260  res_r = (v8u16) __msa_srari_h((v8i16) res_r, 6);
1261  res_r = __msa_sat_u_h(res_r, 7);
1262  res = (v16u8) __msa_pckev_b((v16i8) res_r, (v16i8) res_r);
1263  res = __msa_aver_u_b(res, dst_data);
1264 
1265  ST4x2_UB(res, dst, dst_stride);
1266 }
1267 
1268 static void avc_chroma_vt_and_aver_dst_4x4mul_msa(uint8_t *src,
1269  int32_t src_stride,
1270  uint8_t *dst,
1271  int32_t dst_stride,
1272  uint32_t coeff0,
1273  uint32_t coeff1,
1274  int32_t height)
1275 {
1276  uint32_t load0, load1, row;
1277  v16i8 src0, src1, src2, src3, src4;
1278  v16u8 tmp0, tmp1, tmp2, tmp3;
1279  v16u8 dst0 = { 0 };
1280  v16u8 dst1 = { 0 };
1281  v8u16 res0_r, res1_r;
1282  v16u8 res0, res1;
1283  v16i8 coeff_vec0 = __msa_fill_b(coeff0);
1284  v16i8 coeff_vec1 = __msa_fill_b(coeff1);
1285  v16u8 coeff_vec = (v16u8) __msa_ilvr_b(coeff_vec0, coeff_vec1);
1286 
1287  src0 = LD_SB(src);
1288  src += src_stride;
1289 
1290  for (row = (height >> 2); row--;) {
1291  LD_SB4(src, src_stride, src1, src2, src3, src4);
1292  src += (4 * src_stride);
1293 
1294  load0 = LW(dst);
1295  load1 = LW(dst + dst_stride);
1296 
1297  INSERT_W2_UB(load0, load1, dst0);
1298  load0 = LW(dst + 2 * dst_stride);
1299  load1 = LW(dst + 3 * dst_stride);
1300  INSERT_W2_UB(load0, load1, dst1);
1301 
1302  ILVR_B4_UB(src1, src0, src2, src1, src3, src2, src4, src3,
1303  tmp0, tmp1, tmp2, tmp3);
1304  ILVR_D2_UB(tmp1, tmp0, tmp3, tmp2, tmp0, tmp2);
1305  DOTP_UB2_UH(tmp0, tmp2, coeff_vec, coeff_vec, res0_r, res1_r);
1306 
1307  res0_r <<= 3;
1308  res1_r <<= 3;
1309 
1310  SRARI_H2_UH(res0_r, res1_r, 6);
1311  SAT_UH2_UH(res0_r, res1_r, 7);
1312  PCKEV_B2_UB(res0_r, res0_r, res1_r, res1_r, res0, res1);
1313  AVER_UB2_UB(res0, dst0, res1, dst1, res0, res1);
1314 
1315  ST4x4_UB(res0, res1, 0, 1, 0, 1, dst, dst_stride);
1316  dst += (4 * dst_stride);
1317  src0 = src4;
1318  }
1319 }
1320 
1321 static void avc_chroma_vt_and_aver_dst_4w_msa(uint8_t *src, int32_t src_stride,
1322  uint8_t *dst, int32_t dst_stride,
1323  uint32_t coeff0, uint32_t coeff1,
1324  int32_t height)
1325 {
1326  if (2 == height) {
1327  avc_chroma_vt_and_aver_dst_4x2_msa(src, src_stride, dst, dst_stride,
1328  coeff0, coeff1);
1329  } else {
1330  avc_chroma_vt_and_aver_dst_4x4mul_msa(src, src_stride, dst, dst_stride,
1331  coeff0, coeff1, height);
1332  }
1333 }
1334 
1335 static void avc_chroma_vt_and_aver_dst_8w_msa(uint8_t *src, int32_t src_stride,
1336  uint8_t *dst, int32_t dst_stride,
1337  uint32_t coeff0, uint32_t coeff1,
1338  int32_t height)
1339 {
1340  uint32_t row;
1341  v16u8 src0, src1, src2, src3, src4;
1342  v16u8 out0, out1;
1343  v8u16 res0, res1, res2, res3;
1344  v16u8 dst0, dst1, dst2, dst3;
1345  v16i8 coeff_vec0 = __msa_fill_b(coeff0);
1346  v16i8 coeff_vec1 = __msa_fill_b(coeff1);
1347  v16u8 coeff_vec = (v16u8) __msa_ilvr_b(coeff_vec0, coeff_vec1);
1348 
1349  src0 = LD_UB(src);
1350  src += src_stride;
1351 
1352  for (row = height >> 2; row--;) {
1353  LD_UB4(src, src_stride, src1, src2, src3, src4);
1354  src += (4 * src_stride);
1355  LD_UB4(dst, dst_stride, dst0, dst1, dst2, dst3);
1356  ILVR_B4_UB(src1, src0, src2, src1, src3, src2, src4, src3,
1357  src0, src1, src2, src3);
1358  DOTP_UB4_UH(src0, src1, src2, src3, coeff_vec, coeff_vec, coeff_vec,
1359  coeff_vec, res0, res1, res2, res3);
1360  SLLI_4V(res0, res1, res2, res3, 3);
1361  SRARI_H4_UH(res0, res1, res2, res3, 6);
1362  SAT_UH4_UH(res0, res1, res2, res3, 7);
1363  PCKEV_B2_UB(res1, res0, res3, res2, out0, out1);
1364  PCKEV_D2_UB(dst1, dst0, dst3, dst2, dst0, dst1);
1365  AVER_UB2_UB(out0, dst0, out1, dst1, out0, out1);
1366  ST8x4_UB(out0, out1, dst, dst_stride);
1367 
1368  dst += (4 * dst_stride);
1369  src0 = src4;
1370  }
1371 }
1372 
1373 static void avc_chroma_hv_and_aver_dst_2x2_msa(uint8_t *src, int32_t src_stride,
1374  uint8_t *dst, int32_t dst_stride,
1375  uint32_t coef_hor0,
1376  uint32_t coef_hor1,
1377  uint32_t coef_ver0,
1378  uint32_t coef_ver1)
1379 {
1380  uint16_t out0, out1;
1381  v16u8 dst0, dst1;
1382  v16u8 src0, src1, src2;
1383  v8u16 res_hz0, res_hz1, res_vt0, res_vt1;
1384  v16i8 res, mask;
1385  v16i8 coeff_hz_vec0 = __msa_fill_b(coef_hor0);
1386  v16i8 coeff_hz_vec1 = __msa_fill_b(coef_hor1);
1387  v16u8 coeff_hz_vec = (v16u8) __msa_ilvr_b(coeff_hz_vec0, coeff_hz_vec1);
1388  v8u16 coeff_vt_vec0 = (v8u16) __msa_fill_h(coef_ver0);
1389  v8u16 coeff_vt_vec1 = (v8u16) __msa_fill_h(coef_ver1);
1390 
1391  mask = LD_SB(&chroma_mask_arr[48]);
1392 
1393  LD_UB3(src, src_stride, src0, src1, src2);
1394  LD_UB2(dst, dst_stride, dst0, dst1);
1395  VSHF_B2_UB(src0, src1, src1, src2, mask, mask, src0, src1);
1396  DOTP_UB2_UH(src0, src1, coeff_hz_vec, coeff_hz_vec, res_hz0, res_hz1);
1397  MUL2(res_hz0, coeff_vt_vec1, res_hz1, coeff_vt_vec0, res_vt0, res_vt1);
1398 
1399  res_vt0 += res_vt1;
1400  res_vt0 = (v8u16) __msa_srari_h((v8i16) res_vt0, 6);
1401  res_vt0 = __msa_sat_u_h(res_vt0, 7);
1402  res = __msa_pckev_b((v16i8) res_vt0, (v16i8) res_vt0);
1403  dst0 = (v16u8) __msa_insve_h((v8i16) dst0, 1, (v8i16) dst1);
1404  dst0 = __msa_aver_u_b((v16u8) res, dst0);
1405  out0 = __msa_copy_u_h((v8i16) dst0, 0);
1406  out1 = __msa_copy_u_h((v8i16) dst0, 1);
1407 
1408  SH(out0, dst);
1409  dst += dst_stride;
1410  SH(out1, dst);
1411 }
1412 
1413 static void avc_chroma_hv_and_aver_dst_2x4_msa(uint8_t *src, int32_t src_stride,
1414  uint8_t *dst, int32_t dst_stride,
1415  uint32_t coef_hor0,
1416  uint32_t coef_hor1,
1417  uint32_t coef_ver0,
1418  uint32_t coef_ver1)
1419 {
1420  v16u8 src0, src1, src2, src3, src4;
1421  v16u8 tmp0, tmp1, tmp2, tmp3;
1422  v16u8 dst0, dst1, dst2, dst3;
1423  v8u16 res_hz0, res_hz1, res_vt0, res_vt1;
1424  v16i8 res, mask;
1425  v16i8 coeff_hz_vec0 = __msa_fill_b(coef_hor0);
1426  v16i8 coeff_hz_vec1 = __msa_fill_b(coef_hor1);
1427  v16u8 coeff_hz_vec = (v16u8) __msa_ilvr_b(coeff_hz_vec0, coeff_hz_vec1);
1428  v8u16 coeff_vt_vec0 = (v8u16) __msa_fill_h(coef_ver0);
1429  v8u16 coeff_vt_vec1 = (v8u16) __msa_fill_h(coef_ver1);
1430 
1431  mask = LD_SB(&chroma_mask_arr[48]);
1432 
1433  LD_UB5(src, src_stride, src0, src1, src2, src3, src4);
1434  LD_UB4(dst, dst_stride, dst0, dst1, dst2, dst3);
1435  VSHF_B2_UB(src0, src1, src2, src3, mask, mask, tmp0, tmp1);
1436  VSHF_B2_UB(src1, src2, src3, src4, mask, mask, tmp2, tmp3);
1437  ILVR_D2_UB(tmp1, tmp0, tmp3, tmp2, src0, src1);
1438  DOTP_UB2_UH(src0, src1, coeff_hz_vec, coeff_hz_vec, res_hz0, res_hz1);
1439  MUL2(res_hz0, coeff_vt_vec1, res_hz1, coeff_vt_vec0, res_vt0, res_vt1);
1440 
1441  res_vt0 += res_vt1;
1442  res_vt0 = (v8u16) __msa_srari_h((v8i16) res_vt0, 6);
1443  res_vt0 = __msa_sat_u_h(res_vt0, 7);
1444  res = __msa_pckev_b((v16i8) res_vt0, (v16i8) res_vt0);
1445 
1446  dst0 = (v16u8) __msa_insve_h((v8i16) dst0, 1, (v8i16) dst1);
1447  dst0 = (v16u8) __msa_insve_h((v8i16) dst0, 2, (v8i16) dst2);
1448  dst0 = (v16u8) __msa_insve_h((v8i16) dst0, 3, (v8i16) dst3);
1449  dst0 = __msa_aver_u_b((v16u8) res, dst0);
1450 
1451  ST2x4_UB(dst0, 0, dst, dst_stride);
1452 }
1453 
1454 static void avc_chroma_hv_and_aver_dst_2x8_msa(uint8_t *src, int32_t src_stride,
1455  uint8_t *dst, int32_t dst_stride,
1456  uint32_t coef_hor0,
1457  uint32_t coef_hor1,
1458  uint32_t coef_ver0,
1459  uint32_t coef_ver1)
1460 {
1461  v16u8 src0, src1, src2, src3, src4, src5, src6, src7, src8;
1462  v16u8 dst0, dst1, dst2, dst3, dst4, dst5, dst6, dst7;
1463  v16u8 tmp0, tmp1, tmp2, tmp3;
1464  v8u16 res_hz0, res_hz1, res_vt0, res_vt1;
1465  v16i8 res, mask;
1466  v16i8 coeff_hz_vec0 = __msa_fill_b(coef_hor0);
1467  v16i8 coeff_hz_vec1 = __msa_fill_b(coef_hor1);
1468  v16u8 coeff_hz_vec = (v16u8) __msa_ilvr_b(coeff_hz_vec0, coeff_hz_vec1);
1469  v8u16 coeff_vt_vec0 = (v8u16) __msa_fill_h(coef_ver0);
1470  v8u16 coeff_vt_vec1 = (v8u16) __msa_fill_h(coef_ver1);
1471 
1472  mask = LD_SB(&chroma_mask_arr[48]);
1473 
1474  LD_UB5(src, src_stride, src0, src1, src2, src3, src4);
1475  src += (5 * src_stride);
1476  LD_UB4(src, src_stride, src5, src6, src7, src8);
1477 
1478  LD_UB8(dst, dst_stride, dst0, dst1, dst2, dst3, dst4, dst5, dst6, dst7);
1479 
1480  dst0 = (v16u8) __msa_insve_h((v8i16) dst0, 1, (v8i16) dst1);
1481  dst0 = (v16u8) __msa_insve_h((v8i16) dst0, 2, (v8i16) dst2);
1482  dst0 = (v16u8) __msa_insve_h((v8i16) dst0, 3, (v8i16) dst3);
1483 
1484  dst4 = (v16u8) __msa_insve_h((v8i16) dst4, 1, (v8i16) dst5);
1485  dst4 = (v16u8) __msa_insve_h((v8i16) dst4, 2, (v8i16) dst6);
1486  dst4 = (v16u8) __msa_insve_h((v8i16) dst4, 3, (v8i16) dst7);
1487 
1488  VSHF_B2_UB(src0, src1, src2, src3, mask, mask, tmp0, tmp1);
1489  VSHF_B2_UB(src1, src2, src3, src4, mask, mask, tmp2, tmp3);
1490  ILVR_D2_UB(tmp1, tmp0, tmp3, tmp2, src0, src1);
1491  VSHF_B2_UB(src4, src5, src6, src7, mask, mask, tmp0, tmp1);
1492  VSHF_B2_UB(src5, src6, src7, src8, mask, mask, tmp2, tmp3);
1493  ILVR_D2_UB(tmp1, tmp0, tmp3, tmp2, src4, src5);
1494  DOTP_UB2_UH(src0, src1, coeff_hz_vec, coeff_hz_vec, res_hz0, res_hz1);
1495  MUL2(res_hz0, coeff_vt_vec1, res_hz1, coeff_vt_vec0, res_vt0, res_vt1);
1496 
1497  res_vt0 += res_vt1;
1498  res_vt0 = (v8u16) __msa_srari_h((v8i16) res_vt0, 6);
1499  res_vt0 = __msa_sat_u_h(res_vt0, 7);
1500  res = __msa_pckev_b((v16i8) res_vt0, (v16i8) res_vt0);
1501  dst0 = __msa_aver_u_b((v16u8) res, dst0);
1502 
1503  ST2x4_UB(dst0, 0, dst, dst_stride);
1504  dst += (4 * dst_stride);
1505 
1506  DOTP_UB2_UH(src4, src5, coeff_hz_vec, coeff_hz_vec, res_hz0, res_hz1);
1507  MUL2(res_hz0, coeff_vt_vec1, res_hz1, coeff_vt_vec0, res_vt0, res_vt1);
1508 
1509  res_vt0 += res_vt1;
1510  res_vt0 = (v8u16) __msa_srari_h((v8i16) res_vt0, 6);
1511  res_vt0 = __msa_sat_u_h(res_vt0, 7);
1512  res = __msa_pckev_b((v16i8) res_vt0, (v16i8) res_vt0);
1513  dst4 = __msa_aver_u_b((v16u8) res, dst4);
1514 
1515  ST2x4_UB(dst4, 0, dst, dst_stride);
1516 }
1517 
1518 static void avc_chroma_hv_and_aver_dst_2w_msa(uint8_t *src, int32_t src_stride,
1519  uint8_t *dst, int32_t dst_stride,
1520  uint32_t coef_hor0,
1521  uint32_t coef_hor1,
1522  uint32_t coef_ver0,
1523  uint32_t coef_ver1,
1524  int32_t height)
1525 {
1526  if (2 == height) {
1527  avc_chroma_hv_and_aver_dst_2x2_msa(src, src_stride, dst, dst_stride,
1528  coef_hor0, coef_hor1,
1529  coef_ver0, coef_ver1);
1530  } else if (4 == height) {
1531  avc_chroma_hv_and_aver_dst_2x4_msa(src, src_stride, dst, dst_stride,
1532  coef_hor0, coef_hor1,
1533  coef_ver0, coef_ver1);
1534  } else if (8 == height) {
1535  avc_chroma_hv_and_aver_dst_2x8_msa(src, src_stride, dst, dst_stride,
1536  coef_hor0, coef_hor1,
1537  coef_ver0, coef_ver1);
1538  }
1539 }
1540 
1541 static void avc_chroma_hv_and_aver_dst_4x2_msa(uint8_t *src, int32_t src_stride,
1542  uint8_t *dst, int32_t dst_stride,
1543  uint32_t coef_hor0,
1544  uint32_t coef_hor1,
1545  uint32_t coef_ver0,
1546  uint32_t coef_ver1)
1547 {
1548  v16u8 src0, src1, src2;
1549  v16u8 dst0, dst1;
1550  v8u16 res_hz0, res_hz1, res_vt0, res_vt1;
1551  v16i8 res, mask;
1552  v16i8 coeff_hz_vec0 = __msa_fill_b(coef_hor0);
1553  v16i8 coeff_hz_vec1 = __msa_fill_b(coef_hor1);
1554  v16u8 coeff_hz_vec = (v16u8) __msa_ilvr_b(coeff_hz_vec0, coeff_hz_vec1);
1555  v8u16 coeff_vt_vec0 = (v8u16) __msa_fill_h(coef_ver0);
1556  v8u16 coeff_vt_vec1 = (v8u16) __msa_fill_h(coef_ver1);
1557 
1558  mask = LD_SB(&chroma_mask_arr[0]);
1559 
1560  LD_UB3(src, src_stride, src0, src1, src2);
1561  LD_UB2(dst, dst_stride, dst0, dst1);
1562  VSHF_B2_UB(src0, src1, src1, src2, mask, mask, src0, src1);
1563  DOTP_UB2_UH(src0, src1, coeff_hz_vec, coeff_hz_vec, res_hz0, res_hz1);
1564  MUL2(res_hz0, coeff_vt_vec1, res_hz1, coeff_vt_vec0, res_vt0, res_vt1);
1565 
1566  res_vt0 += res_vt1;
1567  res_vt0 = (v8u16) __msa_srari_h((v8i16) res_vt0, 6);
1568  res_vt0 = __msa_sat_u_h(res_vt0, 7);
1569  res = __msa_pckev_b((v16i8) res_vt0, (v16i8) res_vt0);
1570  dst0 = (v16u8) __msa_insve_w((v4i32) dst0, 1, (v4i32) dst1);
1571  dst0 = __msa_aver_u_b((v16u8) res, dst0);
1572 
1573  ST4x2_UB(dst0, dst, dst_stride);
1574 }
1575 
1576 static void avc_chroma_hv_and_aver_dst_4x4mul_msa(uint8_t *src,
1577  int32_t src_stride,
1578  uint8_t *dst,
1579  int32_t dst_stride,
1580  uint32_t coef_hor0,
1581  uint32_t coef_hor1,
1582  uint32_t coef_ver0,
1583  uint32_t coef_ver1,
1584  int32_t height)
1585 {
1586  uint32_t row;
1587  v16u8 src0, src1, src2, src3, src4;
1588  v16u8 dst0, dst1, dst2, dst3;
1589  v8u16 res_hz0, res_hz1, res_hz2, res_hz3;
1590  v8u16 res_vt0, res_vt1, res_vt2, res_vt3;
1591  v16i8 mask;
1592  v16i8 coeff_hz_vec0 = __msa_fill_b(coef_hor0);
1593  v16i8 coeff_hz_vec1 = __msa_fill_b(coef_hor1);
1594  v16u8 coeff_hz_vec = (v16u8) __msa_ilvr_b(coeff_hz_vec0, coeff_hz_vec1);
1595  v8u16 coeff_vt_vec0 = (v8u16) __msa_fill_h(coef_ver0);
1596  v8u16 coeff_vt_vec1 = (v8u16) __msa_fill_h(coef_ver1);
1597  v16u8 res0, res1;
1598 
1599  mask = LD_SB(&chroma_mask_arr[0]);
1600 
1601  src0 = LD_UB(src);
1602  src += src_stride;
1603 
1604  for (row = (height >> 2); row--;) {
1605  LD_UB4(src, src_stride, src1, src2, src3, src4);
1606  src += (4 * src_stride);
1607 
1608  LD_UB4(dst, dst_stride, dst0, dst1, dst2, dst3);
1609 
1610  VSHF_B2_UB(src0, src1, src1, src2, mask, mask, src0, src1);
1611  VSHF_B2_UB(src2, src3, src3, src4, mask, mask, src2, src3);
1612  DOTP_UB4_UH(src0, src1, src2, src3, coeff_hz_vec, coeff_hz_vec,
1613  coeff_hz_vec, coeff_hz_vec, res_hz0, res_hz1, res_hz2,
1614  res_hz3);
1615  MUL4(res_hz0, coeff_vt_vec1, res_hz1, coeff_vt_vec0, res_hz2,
1616  coeff_vt_vec1, res_hz3, coeff_vt_vec0, res_vt0, res_vt1, res_vt2,
1617  res_vt3);
1618  ADD2(res_vt0, res_vt1, res_vt2, res_vt3, res_vt0, res_vt1);
1619  SRARI_H2_UH(res_vt0, res_vt1, 6);
1620  SAT_UH2_UH(res_vt0, res_vt1, 7);
1621  PCKEV_B2_UB(res_vt0, res_vt0, res_vt1, res_vt1, res0, res1);
1622 
1623  dst0 = (v16u8) __msa_insve_w((v4i32) dst0, 1, (v4i32) dst1);
1624  dst1 = (v16u8) __msa_insve_w((v4i32) dst2, 1, (v4i32) dst3);
1625 
1626  AVER_UB2_UB(res0, dst0, res1, dst1, dst0, dst1);
1627 
1628  ST4x4_UB(dst0, dst1, 0, 1, 0, 1, dst, dst_stride);
1629  dst += (4 * dst_stride);
1630  src0 = src4;
1631  }
1632 }
1633 
1634 static void avc_chroma_hv_and_aver_dst_4w_msa(uint8_t *src, int32_t src_stride,
1635  uint8_t *dst, int32_t dst_stride,
1636  uint32_t coef_hor0,
1637  uint32_t coef_hor1,
1638  uint32_t coef_ver0,
1639  uint32_t coef_ver1,
1640  int32_t height)
1641 {
1642  if (2 == height) {
1643  avc_chroma_hv_and_aver_dst_4x2_msa(src, src_stride, dst, dst_stride,
1644  coef_hor0, coef_hor1,
1645  coef_ver0, coef_ver1);
1646  } else {
1647  avc_chroma_hv_and_aver_dst_4x4mul_msa(src, src_stride, dst, dst_stride,
1648  coef_hor0, coef_hor1,
1649  coef_ver0, coef_ver1, height);
1650  }
1651 }
1652 
1653 static void avc_chroma_hv_and_aver_dst_8w_msa(uint8_t *src, int32_t src_stride,
1654  uint8_t *dst, int32_t dst_stride,
1655  uint32_t coef_hor0,
1656  uint32_t coef_hor1,
1657  uint32_t coef_ver0,
1658  uint32_t coef_ver1,
1659  int32_t height)
1660 {
1661  uint32_t row;
1662  v16u8 src0, src1, src2, src3, src4, out0, out1;
1663  v8u16 res_hz0, res_hz1, res_hz2;
1664  v8u16 res_hz3, res_hz4;
1665  v8u16 res_vt0, res_vt1, res_vt2, res_vt3;
1666  v16u8 dst0, dst1, dst2, dst3;
1667  v16i8 mask;
1668  v16i8 coeff_hz_vec0 = __msa_fill_b(coef_hor0);
1669  v16i8 coeff_hz_vec1 = __msa_fill_b(coef_hor1);
1670  v16u8 coeff_hz_vec = (v16u8) __msa_ilvr_b(coeff_hz_vec0, coeff_hz_vec1);
1671  v8u16 coeff_vt_vec0 = (v8u16) __msa_fill_h(coef_ver0);
1672  v8u16 coeff_vt_vec1 = (v8u16) __msa_fill_h(coef_ver1);
1673 
1674  mask = LD_SB(&chroma_mask_arr[32]);
1675 
1676  src0 = LD_UB(src);
1677  src += src_stride;
1678 
1679  src0 = (v16u8) __msa_vshf_b(mask, (v16i8) src0, (v16i8) src0);
1680  res_hz0 = __msa_dotp_u_h(src0, coeff_hz_vec);
1681 
1682  for (row = (height >> 2); row--;) {
1683  LD_UB4(src, src_stride, src1, src2, src3, src4);
1684  src += (4 * src_stride);
1685 
1686  LD_UB4(dst, dst_stride, dst0, dst1, dst2, dst3);
1687  VSHF_B2_UB(src1, src1, src2, src2, mask, mask, src1, src2);
1688  VSHF_B2_UB(src3, src3, src4, src4, mask, mask, src3, src4);
1689  DOTP_UB4_UH(src1, src2, src3, src4, coeff_hz_vec, coeff_hz_vec,
1690  coeff_hz_vec, coeff_hz_vec, res_hz1, res_hz2, res_hz3,
1691  res_hz4);
1692  MUL4(res_hz1, coeff_vt_vec0, res_hz2, coeff_vt_vec0, res_hz3,
1693  coeff_vt_vec0, res_hz4, coeff_vt_vec0, res_vt0, res_vt1, res_vt2,
1694  res_vt3);
1695 
1696  res_vt0 += (res_hz0 * coeff_vt_vec1);
1697  res_vt1 += (res_hz1 * coeff_vt_vec1);
1698  res_vt2 += (res_hz2 * coeff_vt_vec1);
1699  res_vt3 += (res_hz3 * coeff_vt_vec1);
1700 
1701  SRARI_H4_UH(res_vt0, res_vt1, res_vt2, res_vt3, 6);
1702  SAT_UH4_UH(res_vt0, res_vt1, res_vt2, res_vt3, 7);
1703 
1704  PCKEV_B2_UB(res_vt1, res_vt0, res_vt3, res_vt2, out0, out1);
1705  PCKEV_D2_UB(dst1, dst0, dst3, dst2, dst0, dst1);
1706  AVER_UB2_UB(out0, dst0, out1, dst1, out0, out1);
1707  ST8x4_UB(out0, out1, dst, dst_stride);
1708  dst += (4 * dst_stride);
1709 
1710  res_hz0 = res_hz4;
1711  }
1712 }
1713 
1714 static void copy_width8_msa(uint8_t *src, int32_t src_stride,
1715  uint8_t *dst, int32_t dst_stride,
1716  int32_t height)
1717 {
1718  int32_t cnt;
1719  uint64_t out0, out1, out2, out3, out4, out5, out6, out7;
1720  v16u8 src0, src1, src2, src3, src4, src5, src6, src7;
1721 
1722  if (0 == height % 12) {
1723  for (cnt = (height / 12); cnt--;) {
1724  LD_UB8(src, src_stride,
1725  src0, src1, src2, src3, src4, src5, src6, src7);
1726  src += (8 * src_stride);
1727 
1728  out0 = __msa_copy_u_d((v2i64) src0, 0);
1729  out1 = __msa_copy_u_d((v2i64) src1, 0);
1730  out2 = __msa_copy_u_d((v2i64) src2, 0);
1731  out3 = __msa_copy_u_d((v2i64) src3, 0);
1732  out4 = __msa_copy_u_d((v2i64) src4, 0);
1733  out5 = __msa_copy_u_d((v2i64) src5, 0);
1734  out6 = __msa_copy_u_d((v2i64) src6, 0);
1735  out7 = __msa_copy_u_d((v2i64) src7, 0);
1736 
1737  SD4(out0, out1, out2, out3, dst, dst_stride);
1738  dst += (4 * dst_stride);
1739  SD4(out4, out5, out6, out7, dst, dst_stride);
1740  dst += (4 * dst_stride);
1741 
1742  LD_UB4(src, src_stride, src0, src1, src2, src3);
1743  src += (4 * src_stride);
1744 
1745  out0 = __msa_copy_u_d((v2i64) src0, 0);
1746  out1 = __msa_copy_u_d((v2i64) src1, 0);
1747  out2 = __msa_copy_u_d((v2i64) src2, 0);
1748  out3 = __msa_copy_u_d((v2i64) src3, 0);
1749 
1750  SD4(out0, out1, out2, out3, dst, dst_stride);
1751  dst += (4 * dst_stride);
1752  }
1753  } else if (0 == height % 8) {
1754  for (cnt = height >> 3; cnt--;) {
1755  LD_UB8(src, src_stride,
1756  src0, src1, src2, src3, src4, src5, src6, src7);
1757  src += (8 * src_stride);
1758 
1759  out0 = __msa_copy_u_d((v2i64) src0, 0);
1760  out1 = __msa_copy_u_d((v2i64) src1, 0);
1761  out2 = __msa_copy_u_d((v2i64) src2, 0);
1762  out3 = __msa_copy_u_d((v2i64) src3, 0);
1763  out4 = __msa_copy_u_d((v2i64) src4, 0);
1764  out5 = __msa_copy_u_d((v2i64) src5, 0);
1765  out6 = __msa_copy_u_d((v2i64) src6, 0);
1766  out7 = __msa_copy_u_d((v2i64) src7, 0);
1767 
1768  SD4(out0, out1, out2, out3, dst, dst_stride);
1769  dst += (4 * dst_stride);
1770  SD4(out4, out5, out6, out7, dst, dst_stride);
1771  dst += (4 * dst_stride);
1772  }
1773  } else if (0 == height % 4) {
1774  for (cnt = (height / 4); cnt--;) {
1775  LD_UB4(src, src_stride, src0, src1, src2, src3);
1776  src += (4 * src_stride);
1777  out0 = __msa_copy_u_d((v2i64) src0, 0);
1778  out1 = __msa_copy_u_d((v2i64) src1, 0);
1779  out2 = __msa_copy_u_d((v2i64) src2, 0);
1780  out3 = __msa_copy_u_d((v2i64) src3, 0);
1781 
1782  SD4(out0, out1, out2, out3, dst, dst_stride);
1783  dst += (4 * dst_stride);
1784  }
1785  } else if (0 == height % 2) {
1786  for (cnt = (height / 2); cnt--;) {
1787  LD_UB2(src, src_stride, src0, src1);
1788  src += (2 * src_stride);
1789  out0 = __msa_copy_u_d((v2i64) src0, 0);
1790  out1 = __msa_copy_u_d((v2i64) src1, 0);
1791 
1792  SD(out0, dst);
1793  dst += dst_stride;
1794  SD(out1, dst);
1795  dst += dst_stride;
1796  }
1797  }
1798 }
1799 
1800 static void avg_width4_msa(uint8_t *src, int32_t src_stride,
1801  uint8_t *dst, int32_t dst_stride,
1802  int32_t height)
1803 {
1804  int32_t cnt;
1805  uint32_t out0, out1, out2, out3;
1806  v16u8 src0, src1, src2, src3;
1807  v16u8 dst0, dst1, dst2, dst3;
1808 
1809  if (0 == (height % 4)) {
1810  for (cnt = (height / 4); cnt--;) {
1811  LD_UB4(src, src_stride, src0, src1, src2, src3);
1812  src += (4 * src_stride);
1813 
1814  LD_UB4(dst, dst_stride, dst0, dst1, dst2, dst3);
1815 
1816  AVER_UB4_UB(src0, dst0, src1, dst1, src2, dst2, src3, dst3,
1817  dst0, dst1, dst2, dst3);
1818 
1819  out0 = __msa_copy_u_w((v4i32) dst0, 0);
1820  out1 = __msa_copy_u_w((v4i32) dst1, 0);
1821  out2 = __msa_copy_u_w((v4i32) dst2, 0);
1822  out3 = __msa_copy_u_w((v4i32) dst3, 0);
1823  SW4(out0, out1, out2, out3, dst, dst_stride);
1824  dst += (4 * dst_stride);
1825  }
1826  } else if (0 == (height % 2)) {
1827  for (cnt = (height / 2); cnt--;) {
1828  LD_UB2(src, src_stride, src0, src1);
1829  src += (2 * src_stride);
1830 
1831  LD_UB2(dst, dst_stride, dst0, dst1);
1832 
1833  AVER_UB2_UB(src0, dst0, src1, dst1, dst0, dst1);
1834 
1835  out0 = __msa_copy_u_w((v4i32) dst0, 0);
1836  out1 = __msa_copy_u_w((v4i32) dst1, 0);
1837  SW(out0, dst);
1838  dst += dst_stride;
1839  SW(out1, dst);
1840  dst += dst_stride;
1841  }
1842  }
1843 }
1844 
1845 static void avg_width8_msa(uint8_t *src, int32_t src_stride,
1846  uint8_t *dst, int32_t dst_stride,
1847  int32_t height)
1848 {
1849  int32_t cnt;
1850  uint64_t out0, out1, out2, out3;
1851  v16u8 src0, src1, src2, src3;
1852  v16u8 dst0, dst1, dst2, dst3;
1853 
1854  for (cnt = (height / 4); cnt--;) {
1855  LD_UB4(src, src_stride, src0, src1, src2, src3);
1856  src += (4 * src_stride);
1857  LD_UB4(dst, dst_stride, dst0, dst1, dst2, dst3);
1858 
1859  AVER_UB4_UB(src0, dst0, src1, dst1, src2, dst2, src3, dst3,
1860  dst0, dst1, dst2, dst3);
1861 
1862  out0 = __msa_copy_u_d((v2i64) dst0, 0);
1863  out1 = __msa_copy_u_d((v2i64) dst1, 0);
1864  out2 = __msa_copy_u_d((v2i64) dst2, 0);
1865  out3 = __msa_copy_u_d((v2i64) dst3, 0);
1866  SD4(out0, out1, out2, out3, dst, dst_stride);
1867  dst += (4 * dst_stride);
1868  }
1869 }
1870 
1871 void ff_put_h264_chroma_mc8_msa(uint8_t *dst, uint8_t *src,
1872  int stride, int height, int x, int y)
1873 {
1874  av_assert2(x < 8 && y < 8 && x >= 0 && y >= 0);
1875 
1876  if (x && y) {
1877  avc_chroma_hv_8w_msa(src, stride, dst,
1878  stride, x, (8 - x), y, (8 - y), height);
1879  } else if (x) {
1880  avc_chroma_hz_8w_msa(src, stride, dst, stride, x, (8 - x), height);
1881  } else if (y) {
1882  avc_chroma_vt_8w_msa(src, stride, dst, stride, y, (8 - y), height);
1883  } else {
1884  copy_width8_msa(src, stride, dst, stride, height);
1885  }
1886 }
1887 
1888 void ff_put_h264_chroma_mc4_msa(uint8_t *dst, uint8_t *src,
1889  int stride, int height, int x, int y)
1890 {
1891  int32_t cnt;
1892 
1893  av_assert2(x < 8 && y < 8 && x >= 0 && y >= 0);
1894 
1895  if (x && y) {
1896  avc_chroma_hv_4w_msa(src, stride, dst,
1897  stride, x, (8 - x), y, (8 - y), height);
1898  } else if (x) {
1899  avc_chroma_hz_4w_msa(src, stride, dst, stride, x, (8 - x), height);
1900  } else if (y) {
1901  avc_chroma_vt_4w_msa(src, stride, dst, stride, y, (8 - y), height);
1902  } else {
1903  for (cnt = height; cnt--;) {
1904  *((uint32_t *) dst) = *((uint32_t *) src);
1905 
1906  src += stride;
1907  dst += stride;
1908  }
1909  }
1910 }
1911 
1912 void ff_put_h264_chroma_mc2_msa(uint8_t *dst, uint8_t *src,
1913  int stride, int height, int x, int y)
1914 {
1915  int32_t cnt;
1916 
1917  av_assert2(x < 8 && y < 8 && x >= 0 && y >= 0);
1918 
1919  if (x && y) {
1920  avc_chroma_hv_2w_msa(src, stride, dst,
1921  stride, x, (8 - x), y, (8 - y), height);
1922  } else if (x) {
1923  avc_chroma_hz_2w_msa(src, stride, dst, stride, x, (8 - x), height);
1924  } else if (y) {
1925  avc_chroma_vt_2w_msa(src, stride, dst, stride, y, (8 - y), height);
1926  } else {
1927  for (cnt = height; cnt--;) {
1928  *((uint16_t *) dst) = *((uint16_t *) src);
1929 
1930  src += stride;
1931  dst += stride;
1932  }
1933  }
1934 }
1935 
1936 void ff_avg_h264_chroma_mc8_msa(uint8_t *dst, uint8_t *src,
1937  int stride, int height, int x, int y)
1938 {
1939  av_assert2(x < 8 && y < 8 && x >= 0 && y >= 0);
1940 
1941 
1942  if (x && y) {
1943  avc_chroma_hv_and_aver_dst_8w_msa(src, stride, dst,
1944  stride, x, (8 - x), y,
1945  (8 - y), height);
1946  } else if (x) {
1947  avc_chroma_hz_and_aver_dst_8w_msa(src, stride, dst,
1948  stride, x, (8 - x), height);
1949  } else if (y) {
1950  avc_chroma_vt_and_aver_dst_8w_msa(src, stride, dst,
1951  stride, y, (8 - y), height);
1952  } else {
1953  avg_width8_msa(src, stride, dst, stride, height);
1954  }
1955 }
1956 
1957 void ff_avg_h264_chroma_mc4_msa(uint8_t *dst, uint8_t *src,
1958  int stride, int height, int x, int y)
1959 {
1960  av_assert2(x < 8 && y < 8 && x >= 0 && y >= 0);
1961 
1962  if (x && y) {
1963  avc_chroma_hv_and_aver_dst_4w_msa(src, stride, dst,
1964  stride, x, (8 - x), y,
1965  (8 - y), height);
1966  } else if (x) {
1967  avc_chroma_hz_and_aver_dst_4w_msa(src, stride, dst,
1968  stride, x, (8 - x), height);
1969  } else if (y) {
1970  avc_chroma_vt_and_aver_dst_4w_msa(src, stride, dst,
1971  stride, y, (8 - y), height);
1972  } else {
1973  avg_width4_msa(src, stride, dst, stride, height);
1974  }
1975 }
1976 
1977 void ff_avg_h264_chroma_mc2_msa(uint8_t *dst, uint8_t *src,
1978  int stride, int height, int x, int y)
1979 {
1980  int32_t cnt;
1981 
1982  av_assert2(x < 8 && y < 8 && x >= 0 && y >= 0);
1983 
1984  if (x && y) {
1985  avc_chroma_hv_and_aver_dst_2w_msa(src, stride, dst,
1986  stride, x, (8 - x), y,
1987  (8 - y), height);
1988  } else if (x) {
1989  avc_chroma_hz_and_aver_dst_2w_msa(src, stride, dst,
1990  stride, x, (8 - x), height);
1991  } else if (y) {
1992  avc_chroma_vt_and_aver_dst_2w_msa(src, stride, dst,
1993  stride, y, (8 - y), height);
1994  } else {
1995  for (cnt = height; cnt--;) {
1996  dst[0] = (dst[0] + src[0] + 1) >> 1;
1997  dst[1] = (dst[1] + src[1] + 1) >> 1;
1998 
1999  src += stride;
2000  dst += stride;
2001  }
2002  }
2003 }
avc_chroma_vt_and_aver_dst_2x8_msa
static void avc_chroma_vt_and_aver_dst_2x8_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coeff0, uint32_t coeff1)
Definition: h264chroma_msa.c:1149
avc_chroma_hz_2x8_msa
static void avc_chroma_hz_2x8_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coeff0, uint32_t coeff1)
Definition: h264chroma_msa.c:93
avc_chroma_hz_and_aver_dst_2x4_msa
static void avc_chroma_hz_and_aver_dst_2x4_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coeff0, uint32_t coeff1)
Definition: h264chroma_msa.c:836
avc_chroma_hz_4x4multiple_msa
static void avc_chroma_hz_4x4multiple_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coeff0, uint32_t coeff1, int32_t height)
Definition: h264chroma_msa.c:172
avc_chroma_hz_and_aver_dst_2x2_msa
static void avc_chroma_hz_and_aver_dst_2x2_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coeff0, uint32_t coeff1)
Definition: h264chroma_msa.c:794
avc_chroma_hv_and_aver_dst_4x2_msa
static void avc_chroma_hv_and_aver_dst_4x2_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coef_hor0, uint32_t coef_hor1, uint32_t coef_ver0, uint32_t coef_ver1)
Definition: h264chroma_msa.c:1541
avc_chroma_hv_and_aver_dst_2x2_msa
static void avc_chroma_hv_and_aver_dst_2x2_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coef_hor0, uint32_t coef_hor1, uint32_t coef_ver0, uint32_t coef_ver1)
Definition: h264chroma_msa.c:1373
LW
#define LW(psrc)
Definition: generic_macros_msa.h:142
avc_chroma_hz_4w_msa
static void avc_chroma_hz_4w_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coeff0, uint32_t coeff1, int32_t height)
Definition: h264chroma_msa.c:207
MUL2
#define MUL2(in0, in1, in2, in3, out0, out1)
Definition: generic_macros_msa.h:2236
LD_SB
#define LD_SB(...)
Definition: generic_macros_msa.h:32
ff_avg_h264_chroma_mc4_msa
void ff_avg_h264_chroma_mc4_msa(uint8_t *dst, uint8_t *src, int stride, int height, int x, int y)
Definition: h264chroma_msa.c:1957
ILVR_D2_UB
#define ILVR_D2_UB(...)
Definition: generic_macros_msa.h:1639
LD_UB4
#define LD_UB4(...)
Definition: generic_macros_msa.h:323
ILVR_B2_SB
#define ILVR_B2_SB(...)
Definition: generic_macros_msa.h:1536
avc_chroma_hv_and_aver_dst_2w_msa
static void avc_chroma_hv_and_aver_dst_2w_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coef_hor0, uint32_t coef_hor1, uint32_t coef_ver0, uint32_t coef_ver1, int32_t height)
Definition: h264chroma_msa.c:1518
LD_SB2
#define LD_SB2(...)
Definition: generic_macros_msa.h:308
ST4x4_UB
#define ST4x4_UB(in0, in1, idx0, idx1, idx2, idx3, pdst, stride)
Definition: generic_macros_msa.h:582
avc_chroma_vt_and_aver_dst_8w_msa
static void avc_chroma_vt_and_aver_dst_8w_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coeff0, uint32_t coeff1, int32_t height)
Definition: h264chroma_msa.c:1335
avc_chroma_vt_and_aver_dst_2x4_msa
static void avc_chroma_vt_and_aver_dst_2x4_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coeff0, uint32_t coeff1)
Definition: h264chroma_msa.c:1103
ff_avg_h264_chroma_mc2_msa
void ff_avg_h264_chroma_mc2_msa(uint8_t *dst, uint8_t *src, int stride, int height, int x, int y)
Definition: h264chroma_msa.c:1977
uint8_t
uint8_t
Definition: audio_convert.c:194
av_assert2
#define av_assert2(cond)
assert() equivalent, that does lie in speed critical code.
Definition: avassert.h:63
LD_UB2
#define LD_UB2(...)
Definition: generic_macros_msa.h:307
avc_chroma_vt_4x2_msa
static void avc_chroma_vt_4x2_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coeff0, uint32_t coeff1)
Definition: h264chroma_msa.c:395
avc_chroma_hv_and_aver_dst_8w_msa
static void avc_chroma_hv_and_aver_dst_8w_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coef_hor0, uint32_t coef_hor1, uint32_t coef_ver0, uint32_t coef_ver1, int32_t height)
Definition: h264chroma_msa.c:1653
LD_UB5
#define LD_UB5(...)
Definition: generic_macros_msa.h:331
avc_chroma_hv_and_aver_dst_4x4mul_msa
static void avc_chroma_hv_and_aver_dst_4x4mul_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coef_hor0, uint32_t coef_hor1, uint32_t coef_ver0, uint32_t coef_ver1, int32_t height)
Definition: h264chroma_msa.c:1576
avc_chroma_vt_2x4_msa
static void avc_chroma_vt_2x4_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coeff0, uint32_t coeff1)
Definition: h264chroma_msa.c:303
MUL4
#define MUL4(in0, in1, in2, in3, in4, in5, in6, in7, out0, out1, out2, out3)
Definition: generic_macros_msa.h:2241
avc_chroma_hv_4w_msa
static void avc_chroma_hv_4w_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coef_hor0, uint32_t coef_hor1, uint32_t coef_ver0, uint32_t coef_ver1, int32_t height)
Definition: h264chroma_msa.c:724
SD
#define SD
Definition: dvdsubdec.c:735
INSERT_W2_UB
#define INSERT_W2_UB(...)
Definition: generic_macros_msa.h:1343
avc_chroma_hz_4x2_msa
static void avc_chroma_hz_4x2_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coeff0, uint32_t coeff1)
Definition: h264chroma_msa.c:146
mask
static const uint16_t mask[17]
Definition: lzw.c:38
avc_chroma_hv_and_aver_dst_4w_msa
static void avc_chroma_hv_and_aver_dst_4w_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coef_hor0, uint32_t coef_hor1, uint32_t coef_ver0, uint32_t coef_ver1, int32_t height)
Definition: h264chroma_msa.c:1634
generic_macros_msa.h
avc_chroma_vt_and_aver_dst_2x2_msa
static void avc_chroma_vt_and_aver_dst_2x2_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coeff0, uint32_t coeff1)
Definition: h264chroma_msa.c:1067
SW4
#define SW4(in0, in1, in2, in3, pdst, stride)
Definition: generic_macros_msa.h:271
avc_chroma_hz_and_aver_dst_4x2_msa
static void avc_chroma_hz_and_aver_dst_4x2_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coeff0, uint32_t coeff1)
Definition: h264chroma_msa.c:932
ILVR_B4_UB
#define ILVR_B4_UB(...)
Definition: generic_macros_msa.h:1556
y
float y
Definition: avf_showspectrum.c:96
avc_chroma_hz_2x4_msa
static void avc_chroma_hz_2x4_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coeff0, uint32_t coeff1)
Definition: h264chroma_msa.c:64
LD_UB3
#define LD_UB3(...)
Definition: generic_macros_msa.h:315
avc_chroma_vt_and_aver_dst_4x2_msa
static void avc_chroma_vt_and_aver_dst_4x2_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coeff0, uint32_t coeff1)
Definition: h264chroma_msa.c:1235
LD_UB8
#define LD_UB8(...)
Definition: generic_macros_msa.h:357
avc_chroma_hz_and_aver_dst_4x4multiple_msa
static void avc_chroma_hz_and_aver_dst_4x4multiple_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coeff0, uint32_t coeff1, int32_t height)
Definition: h264chroma_msa.c:966
SRARI_H2_UH
#define SRARI_H2_UH(...)
Definition: generic_macros_msa.h:2193
int32_t
int32_t
Definition: audio_convert.c:194
AVER_UB2_UB
#define AVER_UB2_UB(...)
Definition: generic_macros_msa.h:793
LD_SB3
#define LD_SB3(...)
Definition: generic_macros_msa.h:316
LD_SB4
#define LD_SB4(...)
Definition: generic_macros_msa.h:324
ST2x4_UB
#define ST2x4_UB(in, stidx, pdst, stride)
Definition: generic_macros_msa.h:534
avc_chroma_vt_2x8_msa
static void avc_chroma_vt_2x8_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coeff0, uint32_t coeff1)
Definition: h264chroma_msa.c:332
avc_chroma_hz_and_aver_dst_2w_msa
static void avc_chroma_hz_and_aver_dst_2w_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coeff0, uint32_t coeff1, int32_t height)
Definition: h264chroma_msa.c:915
AVER_UB4_UB
#define AVER_UB4_UB(...)
Definition: generic_macros_msa.h:801
chroma_mask_arr
static const uint8_t chroma_mask_arr[16 *5]
Definition: h264chroma_msa.c:24
avc_chroma_hz_2x2_msa
static void avc_chroma_hz_2x2_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coeff0, uint32_t coeff1)
Definition: h264chroma_msa.c:32
src1
#define src1
Definition: h264pred.c:139
src
AVS_Value src
Definition: avisynth_c.h:482
copy_width8_msa
static void copy_width8_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, int32_t height)
Definition: h264chroma_msa.c:1714
avc_chroma_hz_2w_msa
static void avc_chroma_hz_2w_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coeff0, uint32_t coeff1, int32_t height)
Definition: h264chroma_msa.c:132
height
BYTE int const BYTE int int int height
Definition: avisynth_c.h:676
DOTP_UB2_UH
#define DOTP_UB2_UH(...)
Definition: generic_macros_msa.h:957
avc_chroma_hv_2x4_msa
static void avc_chroma_hv_2x4_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coef_hor0, uint32_t coef_hor1, uint32_t coef_ver0, uint32_t coef_ver1)
Definition: h264chroma_msa.c:543
avc_chroma_vt_4x4multiple_msa
static void avc_chroma_vt_4x4multiple_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coeff0, uint32_t coeff1, int32_t height)
Definition: h264chroma_msa.c:420
avc_chroma_vt_2w_msa
static void avc_chroma_vt_2w_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coeff0, uint32_t coeff1, int32_t height)
Definition: h264chroma_msa.c:381
avc_chroma_hz_and_aver_dst_2x8_msa
static void avc_chroma_hz_and_aver_dst_2x8_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coeff0, uint32_t coeff1)
Definition: h264chroma_msa.c:872
SRARI_H4_UH
#define SRARI_H4_UH(...)
Definition: generic_macros_msa.h:2201
ff_put_h264_chroma_mc4_msa
void ff_put_h264_chroma_mc4_msa(uint8_t *dst, uint8_t *src, int stride, int height, int x, int y)
Definition: h264chroma_msa.c:1888
h264chroma_mips.h
src0
#define src0
Definition: h264pred.c:138
avc_chroma_hz_8w_msa
static void avc_chroma_hz_8w_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coeff0, uint32_t coeff1, int32_t height)
Definition: h264chroma_msa.c:220
ADD2
#define ADD2(in0, in1, in2, in3, out0, out1)
Definition: generic_macros_msa.h:2253
avc_chroma_vt_and_aver_dst_2w_msa
static void avc_chroma_vt_and_aver_dst_2w_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coeff0, uint32_t coeff1, int32_t height)
Definition: h264chroma_msa.c:1218
avc_chroma_hv_and_aver_dst_2x8_msa
static void avc_chroma_hv_and_aver_dst_2x8_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coef_hor0, uint32_t coef_hor1, uint32_t coef_ver0, uint32_t coef_ver1)
Definition: h264chroma_msa.c:1454
SH
#define SH(val, pdst)
Definition: generic_macros_msa.h:190
SD4
#define SD4(in0, in1, in2, in3, pdst, stride)
Definition: generic_macros_msa.h:286
avc_chroma_hv_8w_msa
static void avc_chroma_hv_8w_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coef_hor0, uint32_t coef_hor1, uint32_t coef_ver0, uint32_t coef_ver1, int32_t height)
Definition: h264chroma_msa.c:740
SW
#define SW(val, pdst)
Definition: generic_macros_msa.h:203
SLLI_4V
#define SLLI_4V(in0, in1, in2, in3, shift)
Definition: generic_macros_msa.h:2074
ILVR_W2_UB
#define ILVR_W2_UB(...)
Definition: generic_macros_msa.h:1612
LD_SB5
#define LD_SB5(...)
Definition: generic_macros_msa.h:332
avc_chroma_vt_4w_msa
static void avc_chroma_vt_4w_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coeff0, uint32_t coeff1, int32_t height)
Definition: h264chroma_msa.c:459
stride
GLint GLenum GLboolean GLsizei stride
Definition: opengl_enc.c:105
LW4
#define LW4(psrc, stride, out0, out1, out2, out3)
Definition: generic_macros_msa.h:238
ST8x4_UB
#define ST8x4_UB(in0, in1, pdst, stride)
Definition: generic_macros_msa.h:687
avc_chroma_hv_and_aver_dst_2x4_msa
static void avc_chroma_hv_and_aver_dst_2x4_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coef_hor0, uint32_t coef_hor1, uint32_t coef_ver0, uint32_t coef_ver1)
Definition: h264chroma_msa.c:1413
ff_avg_h264_chroma_mc8_msa
void ff_avg_h264_chroma_mc8_msa(uint8_t *dst, uint8_t *src, int stride, int height, int x, int y)
Definition: h264chroma_msa.c:1936
SAT_UH2_UH
#define SAT_UH2_UH(...)
Definition: generic_macros_msa.h:1735
avc_chroma_hv_4x4multiple_msa
static void avc_chroma_hv_4x4multiple_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coef_hor0, uint32_t coef_hor1, uint32_t coef_ver0, uint32_t coef_ver1, int32_t height)
Definition: h264chroma_msa.c:676
PCKEV_B2_SW
#define PCKEV_B2_SW(...)
Definition: generic_macros_msa.h:1881
SAT_UH4_UH
#define SAT_UH4_UH(...)
Definition: generic_macros_msa.h:1743
avc_chroma_hv_2x2_msa
static void avc_chroma_hv_2x2_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coef_hor0, uint32_t coef_hor1, uint32_t coef_ver0, uint32_t coef_ver1)
Definition: h264chroma_msa.c:507
ff_put_h264_chroma_mc2_msa
void ff_put_h264_chroma_mc2_msa(uint8_t *dst, uint8_t *src, int stride, int height, int x, int y)
Definition: h264chroma_msa.c:1912
avc_chroma_vt_and_aver_dst_4w_msa
static void avc_chroma_vt_and_aver_dst_4w_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coeff0, uint32_t coeff1, int32_t height)
Definition: h264chroma_msa.c:1321
avc_chroma_hv_4x2_msa
static void avc_chroma_hv_4x2_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coef_hor0, uint32_t coef_hor1, uint32_t coef_ver0, uint32_t coef_ver1)
Definition: h264chroma_msa.c:647
LD_UB
#define LD_UB(...)
Definition: generic_macros_msa.h:31
DOTP_UB4_UH
#define DOTP_UB4_UH(...)
Definition: generic_macros_msa.h:966
avc_chroma_vt_2x2_msa
static void avc_chroma_vt_2x2_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coeff0, uint32_t coeff1)
Definition: h264chroma_msa.c:270
avg_width8_msa
static void avg_width8_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, int32_t height)
Definition: h264chroma_msa.c:1845
avg_width4_msa
static void avg_width4_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, int32_t height)
Definition: h264chroma_msa.c:1800
VSHF_B2_UB
#define VSHF_B2_UB(...)
Definition: generic_macros_msa.h:878
avc_chroma_hv_2w_msa
static void avc_chroma_hv_2w_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coef_hor0, uint32_t coef_hor1, uint32_t coef_ver0, uint32_t coef_ver1, int32_t height)
Definition: h264chroma_msa.c:629
avc_chroma_hz_and_aver_dst_8w_msa
static void avc_chroma_hz_and_aver_dst_8w_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coeff0, uint32_t coeff1, int32_t height)
Definition: h264chroma_msa.c:1032
PCKEV_D2_UB
#define PCKEV_D2_UB(...)
Definition: generic_macros_msa.h:1946
avc_chroma_hv_2x8_msa
static void avc_chroma_hv_2x8_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coef_hor0, uint32_t coef_hor1, uint32_t coef_ver0, uint32_t coef_ver1)
Definition: h264chroma_msa.c:577
stride
#define stride
ST8x1_UB
#define ST8x1_UB(in, pdst)
Definition: generic_macros_msa.h:649
ff_put_h264_chroma_mc8_msa
void ff_put_h264_chroma_mc8_msa(uint8_t *dst, uint8_t *src, int stride, int height, int x, int y)
Definition: h264chroma_msa.c:1871
ST4x2_UB
#define ST4x2_UB(in, pdst, stride)
Definition: generic_macros_msa.h:558
avc_chroma_vt_and_aver_dst_4x4mul_msa
static void avc_chroma_vt_and_aver_dst_4x4mul_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coeff0, uint32_t coeff1, int32_t height)
Definition: h264chroma_msa.c:1268
PCKEV_B2_UB
#define PCKEV_B2_UB(...)
Definition: generic_macros_msa.h:1879
avc_chroma_vt_8w_msa
static void avc_chroma_vt_8w_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coeff0, uint32_t coeff1, int32_t height)
Definition: h264chroma_msa.c:472
ILVR_B2_UB
#define ILVR_B2_UB(...)
Definition: generic_macros_msa.h:1535
avc_chroma_hz_and_aver_dst_4w_msa
static void avc_chroma_hz_and_aver_dst_4w_msa(uint8_t *src, int32_t src_stride, uint8_t *dst, int32_t dst_stride, uint32_t coeff0, uint32_t coeff1, int32_t height)
Definition: h264chroma_msa.c:1017
Generated on Thu Sep 10 2015 11:38:59 for FFmpeg by   doxygen 1.8.6