FFmpeg
 All Data Structures Namespaces Files Functions Variables Typedefs Enumerations Enumerator Macros Groups Pages
h264chroma_mmi.c
Go to the documentation of this file.
1 /*
2  * Loongson SIMD optimized h264chroma
3  *
4  * Copyright (c) 2015 Loongson Technology Corporation Limited
5  * Copyright (c) 2015 Zhou Xiaoyong <zhouxiaoyong@loongson.cn>
6  * Zhang Shuangshuang <zhangshuangshuang@ict.ac.cn>
7  *
8  * This file is part of FFmpeg.
9  *
10  * FFmpeg is free software; you can redistribute it and/or
11  * modify it under the terms of the GNU Lesser General Public
12  * License as published by the Free Software Foundation; either
13  * version 2.1 of the License, or (at your option) any later version.
14  *
15  * FFmpeg is distributed in the hope that it will be useful,
16  * but WITHOUT ANY WARRANTY; without even the implied warranty of
17  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
18  * Lesser General Public License for more details.
19  *
20  * You should have received a copy of the GNU Lesser General Public
21  * License along with FFmpeg; if not, write to the Free Software
22  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
23  */
24 
25 #include "h264chroma_mips.h"
26 #include "constants.h"
27 #include "libavutil/mips/asmdefs.h"
28 
29 void ff_put_h264_chroma_mc8_mmi(uint8_t *dst, uint8_t *src, int stride,
30  int h, int x, int y)
31 {
32  const int A = (8 - x) * (8 - y);
33  const int B = x * (8 - y);
34  const int C = (8 - x) * y;
35  const int D = x * y;
36  const int E = B + C;
37  double ftmp[10];
38  uint64_t tmp[1];
39  mips_reg addr[1];
40 
41  if (D) {
42  __asm__ volatile (
43  "xor %[ftmp0], %[ftmp0], %[ftmp0] \n\t"
44  "dli %[tmp0], 0x06 \n\t"
45  "pshufh %[A], %[A], %[ftmp0] \n\t"
46  "pshufh %[B], %[B], %[ftmp0] \n\t"
47  "mtc1 %[tmp0], %[ftmp9] \n\t"
48  "pshufh %[C], %[C], %[ftmp0] \n\t"
49  "pshufh %[D], %[D], %[ftmp0] \n\t"
50  "1: \n\t"
51  PTR_ADDU "%[addr0], %[src], %[stride] \n\t"
52  "gsldlc1 %[ftmp1], 0x07(%[src]) \n\t"
53  "gsldrc1 %[ftmp1], 0x00(%[src]) \n\t"
54  "gsldlc1 %[ftmp2], 0x08(%[src]) \n\t"
55  "gsldrc1 %[ftmp2], 0x01(%[src]) \n\t"
56  "gsldlc1 %[ftmp3], 0x07(%[addr0]) \n\t"
57  "gsldrc1 %[ftmp3], 0x00(%[addr0]) \n\t"
58  "gsldlc1 %[ftmp4], 0x08(%[addr0]) \n\t"
59  "gsldrc1 %[ftmp4], 0x01(%[addr0]) \n\t"
60 
61  "punpcklbh %[ftmp5], %[ftmp1], %[ftmp0] \n\t"
62  "punpckhbh %[ftmp6], %[ftmp1], %[ftmp0] \n\t"
63  "punpcklbh %[ftmp7], %[ftmp2], %[ftmp0] \n\t"
64  "punpckhbh %[ftmp8], %[ftmp2], %[ftmp0] \n\t"
65  "pmullh %[ftmp5], %[ftmp5], %[A] \n\t"
66  "pmullh %[ftmp7], %[ftmp7], %[B] \n\t"
67  "paddh %[ftmp1], %[ftmp5], %[ftmp7] \n\t"
68  "pmullh %[ftmp6], %[ftmp6], %[A] \n\t"
69  "pmullh %[ftmp8], %[ftmp8], %[B] \n\t"
70  "paddh %[ftmp2], %[ftmp6], %[ftmp8] \n\t"
71 
72  "punpcklbh %[ftmp5], %[ftmp3], %[ftmp0] \n\t"
73  "punpckhbh %[ftmp6], %[ftmp3], %[ftmp0] \n\t"
74  "punpcklbh %[ftmp7], %[ftmp4], %[ftmp0] \n\t"
75  "punpckhbh %[ftmp8], %[ftmp4], %[ftmp0] \n\t"
76  "pmullh %[ftmp5], %[ftmp5], %[C] \n\t"
77  "pmullh %[ftmp7], %[ftmp7], %[D] \n\t"
78  "paddh %[ftmp3], %[ftmp5], %[ftmp7] \n\t"
79  "pmullh %[ftmp6], %[ftmp6], %[C] \n\t"
80  "pmullh %[ftmp8], %[ftmp8], %[D] \n\t"
81  "paddh %[ftmp4], %[ftmp6], %[ftmp8] \n\t"
82 
83  "paddh %[ftmp1], %[ftmp1], %[ftmp3] \n\t"
84  "paddh %[ftmp2], %[ftmp2], %[ftmp4] \n\t"
85  "paddh %[ftmp1], %[ftmp1], %[ff_pw_32] \n\t"
86  "paddh %[ftmp2], %[ftmp2], %[ff_pw_32] \n\t"
87  "psrlh %[ftmp1], %[ftmp1], %[ftmp9] \n\t"
88  "psrlh %[ftmp2], %[ftmp2], %[ftmp9] \n\t"
89  "packushb %[ftmp1], %[ftmp1], %[ftmp2] \n\t"
90  "addi %[h], %[h], -0x01 \n\t"
91  "sdc1 %[ftmp1], 0x00(%[dst]) \n\t"
92  PTR_ADDU "%[src], %[src], %[stride] \n\t"
93  PTR_ADDU "%[dst], %[dst], %[stride] \n\t"
94  "bnez %[h], 1b \n\t"
95  : [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]),
96  [ftmp2]"=&f"(ftmp[2]), [ftmp3]"=&f"(ftmp[3]),
97  [ftmp4]"=&f"(ftmp[4]), [ftmp5]"=&f"(ftmp[5]),
98  [ftmp6]"=&f"(ftmp[6]), [ftmp7]"=&f"(ftmp[7]),
99  [ftmp8]"=&f"(ftmp[8]), [ftmp9]"=&f"(ftmp[9]),
100  [tmp0]"=&r"(tmp[0]),
101  [addr0]"=&r"(addr[0]),
102  [dst]"+&r"(dst), [src]"+&r"(src),
103  [h]"+&r"(h)
104  : [stride]"r"((mips_reg)stride),[ff_pw_32]"f"(ff_pw_32),
105  [A]"f"(A), [B]"f"(B),
106  [C]"f"(C), [D]"f"(D)
107  : "memory"
108  );
109  } else if (E) {
110  const int step = C ? stride : 1;
111 
112  __asm__ volatile (
113  "xor %[ftmp0], %[ftmp0], %[ftmp0] \n\t"
114  "dli %[tmp0], 0x06 \n\t"
115  "pshufh %[A], %[A], %[ftmp0] \n\t"
116  "pshufh %[E], %[E], %[ftmp0] \n\t"
117  "mtc1 %[tmp0], %[ftmp7] \n\t"
118  "1: \n\t"
119  PTR_ADDU "%[addr0], %[src], %[step] \n\t"
120  "gsldlc1 %[ftmp1], 0x07(%[src]) \n\t"
121  "gsldrc1 %[ftmp1], 0x00(%[src]) \n\t"
122  "gsldlc1 %[ftmp2], 0x07(%[addr0]) \n\t"
123  "gsldrc1 %[ftmp2], 0x00(%[addr0]) \n\t"
124 
125  "punpcklbh %[ftmp3], %[ftmp1], %[ftmp0] \n\t"
126  "punpckhbh %[ftmp4], %[ftmp1], %[ftmp0] \n\t"
127  "punpcklbh %[ftmp5], %[ftmp2], %[ftmp0] \n\t"
128  "punpckhbh %[ftmp6], %[ftmp2], %[ftmp0] \n\t"
129  "pmullh %[ftmp3], %[ftmp3], %[A] \n\t"
130  "pmullh %[ftmp5], %[ftmp5], %[E] \n\t"
131  "paddh %[ftmp1], %[ftmp3], %[ftmp5] \n\t"
132  "pmullh %[ftmp4], %[ftmp4], %[A] \n\t"
133  "pmullh %[ftmp6], %[ftmp6], %[E] \n\t"
134  "paddh %[ftmp2], %[ftmp4], %[ftmp6] \n\t"
135 
136  "paddh %[ftmp1], %[ftmp1], %[ff_pw_32] \n\t"
137  "paddh %[ftmp2], %[ftmp2], %[ff_pw_32] \n\t"
138  "psrlh %[ftmp1], %[ftmp1], %[ftmp7] \n\t"
139  "psrlh %[ftmp2], %[ftmp2], %[ftmp7] \n\t"
140  "packushb %[ftmp1], %[ftmp1], %[ftmp2] \n\t"
141  "addi %[h], %[h], -0x01 \n\t"
142  "sdc1 %[ftmp1], 0x00(%[dst]) \n\t"
143  PTR_ADDU "%[src], %[src], %[stride] \n\t"
144  PTR_ADDU "%[dst], %[dst], %[stride] \n\t"
145  "bnez %[h], 1b \n\t"
146  : [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]),
147  [ftmp2]"=&f"(ftmp[2]), [ftmp3]"=&f"(ftmp[3]),
148  [ftmp4]"=&f"(ftmp[4]), [ftmp5]"=&f"(ftmp[5]),
149  [ftmp6]"=&f"(ftmp[6]), [ftmp7]"=&f"(ftmp[7]),
150  [tmp0]"=&r"(tmp[0]),
151  [addr0]"=&r"(addr[0]),
152  [dst]"+&r"(dst), [src]"+&r"(src),
153  [h]"+&r"(h)
154  : [stride]"r"((mips_reg)stride),[step]"r"((mips_reg)step),
155  [ff_pw_32]"f"(ff_pw_32),
156  [A]"f"(A), [E]"f"(E)
157  : "memory"
158  );
159  } else {
160  __asm__ volatile (
161  "xor %[ftmp0], %[ftmp0], %[ftmp0] \n\t"
162  "dli %[tmp0], 0x06 \n\t"
163  "pshufh %[A], %[A], %[ftmp0] \n\t"
164  "mtc1 %[tmp0], %[ftmp4] \n\t"
165  "1: \n\t"
166  "gsldlc1 %[ftmp1], 0x07(%[src]) \n\t"
167  "gsldrc1 %[ftmp1], 0x00(%[src]) \n\t"
168  "punpcklbh %[ftmp2], %[ftmp1], %[ftmp0] \n\t"
169  "punpckhbh %[ftmp3], %[ftmp1], %[ftmp0] \n\t"
170  "pmullh %[ftmp1], %[ftmp2], %[A] \n\t"
171  "pmullh %[ftmp2], %[ftmp3], %[A] \n\t"
172  "paddh %[ftmp1], %[ftmp1], %[ff_pw_32] \n\t"
173  "paddh %[ftmp2], %[ftmp2], %[ff_pw_32] \n\t"
174  "psrlh %[ftmp1], %[ftmp1], %[ftmp4] \n\t"
175  "psrlh %[ftmp2], %[ftmp2], %[ftmp4] \n\t"
176  "packushb %[ftmp1], %[ftmp1], %[ftmp2] \n\t"
177  PTR_ADDU "%[src], %[src], %[stride] \n\t"
178  "sdc1 %[ftmp1], 0x00(%[dst]) \n\t"
179 
180  PTR_ADDU "%[dst], %[dst], %[stride] \n\t"
181  "gsldlc1 %[ftmp1], 0x07(%[src]) \n\t"
182  "gsldrc1 %[ftmp1], 0x00(%[src]) \n\t"
183  "punpcklbh %[ftmp2], %[ftmp1], %[ftmp0] \n\t"
184  "punpckhbh %[ftmp3], %[ftmp1], %[ftmp0] \n\t"
185  "pmullh %[ftmp1], %[ftmp2], %[A] \n\t"
186  "pmullh %[ftmp2], %[ftmp3], %[A] \n\t"
187  "paddh %[ftmp1], %[ftmp1], %[ff_pw_32] \n\t"
188  "paddh %[ftmp2], %[ftmp2], %[ff_pw_32] \n\t"
189  "psrlh %[ftmp1], %[ftmp1], %[ftmp4] \n\t"
190  "psrlh %[ftmp2], %[ftmp2], %[ftmp4] \n\t"
191  "packushb %[ftmp1], %[ftmp1], %[ftmp2] \n\t"
192  "addi %[h], %[h], -0x02 \n\t"
193  "sdc1 %[ftmp1], 0x00(%[dst]) \n\t"
194 
195  PTR_ADDU "%[src], %[src], %[stride] \n\t"
196  PTR_ADDU "%[dst], %[dst], %[stride] \n\t"
197  "bnez %[h], 1b \n\t"
198  : [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]),
199  [ftmp2]"=&f"(ftmp[2]), [ftmp3]"=&f"(ftmp[3]),
200  [ftmp4]"=&f"(ftmp[4]),
201  [tmp0]"=&r"(tmp[0]),
202  [dst]"+&r"(dst), [src]"+&r"(src),
203  [h]"+&r"(h)
204  : [stride]"r"((mips_reg)stride),[ff_pw_32]"f"(ff_pw_32),
205  [A]"f"(A)
206  : "memory"
207  );
208  }
209 }
210 
211 void ff_avg_h264_chroma_mc8_mmi(uint8_t *dst, uint8_t *src, int stride,
212  int h, int x, int y)
213 {
214  const int A = (8 - x) * (8 - y);
215  const int B = x * (8 - y);
216  const int C = (8 - x) * y;
217  const int D = x * y;
218  const int E = B + C;
219  double ftmp[10];
220  uint64_t tmp[1];
221  mips_reg addr[1];
222 
223  if (D) {
224  __asm__ volatile (
225  "xor %[ftmp0], %[ftmp0], %[ftmp0] \n\t"
226  "dli %[tmp0], 0x06 \n\t"
227  "pshufh %[A], %[A], %[ftmp0] \n\t"
228  "pshufh %[B], %[B], %[ftmp0] \n\t"
229  "mtc1 %[tmp0], %[ftmp9] \n\t"
230  "pshufh %[C], %[C], %[ftmp0] \n\t"
231  "pshufh %[D], %[D], %[ftmp0] \n\t"
232  "1: \n\t"
233  PTR_ADDU "%[addr0], %[src], %[stride] \n\t"
234  "gsldlc1 %[ftmp1], 0x07(%[src]) \n\t"
235  "gsldrc1 %[ftmp1], 0x00(%[src]) \n\t"
236  "gsldlc1 %[ftmp2], 0x08(%[src]) \n\t"
237  "gsldrc1 %[ftmp2], 0x01(%[src]) \n\t"
238  "gsldlc1 %[ftmp3], 0x07(%[addr0]) \n\t"
239  "gsldrc1 %[ftmp3], 0x00(%[addr0]) \n\t"
240  "gsldlc1 %[ftmp4], 0x08(%[addr0]) \n\t"
241  "gsldrc1 %[ftmp4], 0x01(%[addr0]) \n\t"
242 
243  "punpcklbh %[ftmp5], %[ftmp1], %[ftmp0] \n\t"
244  "punpckhbh %[ftmp6], %[ftmp1], %[ftmp0] \n\t"
245  "punpcklbh %[ftmp7], %[ftmp2], %[ftmp0] \n\t"
246  "punpckhbh %[ftmp8], %[ftmp2], %[ftmp0] \n\t"
247  "pmullh %[ftmp5], %[ftmp5], %[A] \n\t"
248  "pmullh %[ftmp7], %[ftmp7], %[B] \n\t"
249  "paddh %[ftmp1], %[ftmp5], %[ftmp7] \n\t"
250  "pmullh %[ftmp6], %[ftmp6], %[A] \n\t"
251  "pmullh %[ftmp8], %[ftmp8], %[B] \n\t"
252  "paddh %[ftmp2], %[ftmp6], %[ftmp8] \n\t"
253 
254  "punpcklbh %[ftmp5], %[ftmp3], %[ftmp0] \n\t"
255  "punpckhbh %[ftmp6], %[ftmp3], %[ftmp0] \n\t"
256  "punpcklbh %[ftmp7], %[ftmp4], %[ftmp0] \n\t"
257  "punpckhbh %[ftmp8], %[ftmp4], %[ftmp0] \n\t"
258  "pmullh %[ftmp5], %[ftmp5], %[C] \n\t"
259  "pmullh %[ftmp7], %[ftmp7], %[D] \n\t"
260  "paddh %[ftmp3], %[ftmp5], %[ftmp7] \n\t"
261  "pmullh %[ftmp6], %[ftmp6], %[C] \n\t"
262  "pmullh %[ftmp8], %[ftmp8], %[D] \n\t"
263  "paddh %[ftmp4], %[ftmp6], %[ftmp8] \n\t"
264 
265  "paddh %[ftmp1], %[ftmp1], %[ftmp3] \n\t"
266  "paddh %[ftmp2], %[ftmp2], %[ftmp4] \n\t"
267  "paddh %[ftmp1], %[ftmp1], %[ff_pw_32] \n\t"
268  "paddh %[ftmp2], %[ftmp2], %[ff_pw_32] \n\t"
269  "psrlh %[ftmp1], %[ftmp1], %[ftmp9] \n\t"
270  "psrlh %[ftmp2], %[ftmp2], %[ftmp9] \n\t"
271  "packushb %[ftmp1], %[ftmp1], %[ftmp2] \n\t"
272  "ldc1 %[ftmp2], 0x00(%[dst]) \n\t"
273  "pavgb %[ftmp1], %[ftmp1], %[ftmp2] \n\t"
274  "addi %[h], %[h], -0x01 \n\t"
275  "sdc1 %[ftmp1], 0x00(%[dst]) \n\t"
276  PTR_ADDU "%[dst], %[dst], %[stride] \n\t"
277  PTR_ADDU "%[src], %[src], %[stride] \n\t"
278  "bnez %[h], 1b \n\t"
279  : [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]),
280  [ftmp2]"=&f"(ftmp[2]), [ftmp3]"=&f"(ftmp[3]),
281  [ftmp4]"=&f"(ftmp[4]), [ftmp5]"=&f"(ftmp[5]),
282  [ftmp6]"=&f"(ftmp[6]), [ftmp7]"=&f"(ftmp[7]),
283  [ftmp8]"=&f"(ftmp[8]), [ftmp9]"=&f"(ftmp[9]),
284  [tmp0]"=&r"(tmp[0]),
285  [addr0]"=&r"(addr[0]),
286  [dst]"+&r"(dst), [src]"+&r"(src),
287  [h]"+&r"(h)
288  : [stride]"r"((mips_reg)stride),[ff_pw_32]"f"(ff_pw_32),
289  [A]"f"(A), [B]"f"(B),
290  [C]"f"(C), [D]"f"(D)
291  : "memory"
292  );
293  } else if (E) {
294  const int step = C ? stride : 1;
295 
296  __asm__ volatile (
297  "xor %[ftmp0], %[ftmp0], %[ftmp0] \n\t"
298  "dli %[tmp0], 0x06 \n\t"
299  "pshufh %[A], %[A], %[ftmp0] \n\t"
300  "pshufh %[E], %[E], %[ftmp0] \n\t"
301  "mtc1 %[tmp0], %[ftmp7] \n\t"
302  "1: \n\t"
303  PTR_ADDU "%[addr0], %[src], %[step] \n\t"
304  "gsldlc1 %[ftmp1], 0x07(%[src]) \n\t"
305  "gsldrc1 %[ftmp1], 0x00(%[src]) \n\t"
306  "gsldlc1 %[ftmp2], 0x07(%[addr0]) \n\t"
307  "gsldrc1 %[ftmp2], 0x00(%[addr0]) \n\t"
308 
309  "punpcklbh %[ftmp3], %[ftmp1], %[ftmp0] \n\t"
310  "punpckhbh %[ftmp4], %[ftmp1], %[ftmp0] \n\t"
311  "punpcklbh %[ftmp5], %[ftmp2], %[ftmp0] \n\t"
312  "punpckhbh %[ftmp6], %[ftmp2], %[ftmp0] \n\t"
313  "pmullh %[ftmp3], %[ftmp3], %[A] \n\t"
314  "pmullh %[ftmp5], %[ftmp5], %[E] \n\t"
315  "paddh %[ftmp1], %[ftmp3], %[ftmp5] \n\t"
316  "pmullh %[ftmp4], %[ftmp4], %[A] \n\t"
317  "pmullh %[ftmp6], %[ftmp6], %[E] \n\t"
318  "paddh %[ftmp2], %[ftmp4], %[ftmp6] \n\t"
319 
320  "paddh %[ftmp1], %[ftmp1], %[ff_pw_32] \n\t"
321  "paddh %[ftmp2], %[ftmp2], %[ff_pw_32] \n\t"
322  "psrlh %[ftmp1], %[ftmp1], %[ftmp7] \n\t"
323  "psrlh %[ftmp2], %[ftmp2], %[ftmp7] \n\t"
324  "packushb %[ftmp1], %[ftmp1], %[ftmp2] \n\t"
325  "ldc1 %[ftmp2], 0x00(%[dst]) \n\t"
326  "pavgb %[ftmp1], %[ftmp1], %[ftmp2] \n\t"
327  "addi %[h], %[h], -0x01 \n\t"
328  "sdc1 %[ftmp1], 0x00(%[dst]) \n\t"
329  PTR_ADDU "%[src], %[src], %[stride] \n\t"
330  PTR_ADDU "%[dst], %[dst], %[stride] \n\t"
331  "bnez %[h], 1b \n\t"
332  : [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]),
333  [ftmp2]"=&f"(ftmp[2]), [ftmp3]"=&f"(ftmp[3]),
334  [ftmp4]"=&f"(ftmp[4]), [ftmp5]"=&f"(ftmp[5]),
335  [ftmp6]"=&f"(ftmp[6]), [ftmp7]"=&f"(ftmp[7]),
336  [tmp0]"=&r"(tmp[0]),
337  [addr0]"=&r"(addr[0]),
338  [dst]"+&r"(dst), [src]"+&r"(src),
339  [h]"+&r"(h)
340  : [stride]"r"((mips_reg)stride),[step]"r"((mips_reg)step),
341  [ff_pw_32]"f"(ff_pw_32),
342  [A]"f"(A), [E]"f"(E)
343  : "memory"
344  );
345  } else {
346  __asm__ volatile (
347  "xor %[ftmp0], %[ftmp0], %[ftmp0] \n\t"
348  "dli %[tmp0], 0x06 \n\t"
349  "pshufh %[A], %[A], %[ftmp0] \n\t"
350  "mtc1 %[tmp0], %[ftmp4] \n\t"
351  "1: \n\t"
352  "gsldlc1 %[ftmp1], 0x07(%[src]) \n\t"
353  "gsldrc1 %[ftmp1], 0x00(%[src]) \n\t"
354  "punpcklbh %[ftmp2], %[ftmp1], %[ftmp0] \n\t"
355  "punpckhbh %[ftmp3], %[ftmp1], %[ftmp0] \n\t"
356  "pmullh %[ftmp1], %[ftmp2], %[A] \n\t"
357  "pmullh %[ftmp2], %[ftmp3], %[A] \n\t"
358  "paddh %[ftmp1], %[ftmp1], %[ff_pw_32] \n\t"
359  "paddh %[ftmp2], %[ftmp2], %[ff_pw_32] \n\t"
360  "psrlh %[ftmp1], %[ftmp1], %[ftmp4] \n\t"
361  "psrlh %[ftmp2], %[ftmp2], %[ftmp4] \n\t"
362  "packushb %[ftmp1], %[ftmp1], %[ftmp2] \n\t"
363  "ldc1 %[ftmp2], 0x00(%[dst]) \n\t"
364  "pavgb %[ftmp1], %[ftmp1], %[ftmp2] \n\t"
365  PTR_ADDU "%[src], %[src], %[stride] \n\t"
366  "sdc1 %[ftmp1], 0x00(%[dst]) \n\t"
367  PTR_ADDU "%[dst], %[dst], %[stride] \n\t"
368 
369  "gsldlc1 %[ftmp1], 0x07(%[src]) \n\t"
370  "gsldrc1 %[ftmp1], 0x00(%[src]) \n\t"
371  "punpcklbh %[ftmp2], %[ftmp1], %[ftmp0] \n\t"
372  "punpckhbh %[ftmp3], %[ftmp1], %[ftmp0] \n\t"
373  "pmullh %[ftmp1], %[ftmp2], %[A] \n\t"
374  "pmullh %[ftmp2], %[ftmp3], %[A] \n\t"
375  "paddh %[ftmp1], %[ftmp1], %[ff_pw_32] \n\t"
376  "paddh %[ftmp2], %[ftmp2], %[ff_pw_32] \n\t"
377  "psrlh %[ftmp1], %[ftmp1], %[ftmp4] \n\t"
378  "psrlh %[ftmp2], %[ftmp2], %[ftmp4] \n\t"
379  "packushb %[ftmp1], %[ftmp1], %[ftmp2] \n\t"
380  "ldc1 %[ftmp2], 0x00(%[dst]) \n\t"
381  "pavgb %[ftmp1], %[ftmp1], %[ftmp2] \n\t"
382  "addi %[h], %[h], -0x02 \n\t"
383  "sdc1 %[ftmp1], 0x00(%[dst]) \n\t"
384 
385  PTR_ADDU "%[src], %[src], %[stride] \n\t"
386  PTR_ADDU "%[dst], %[dst], %[stride] \n\t"
387  "bnez %[h], 1b \n\t"
388  : [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]),
389  [ftmp2]"=&f"(ftmp[2]), [ftmp3]"=&f"(ftmp[3]),
390  [ftmp4]"=&f"(ftmp[4]),
391  [tmp0]"=&r"(tmp[0]),
392  [dst]"+&r"(dst), [src]"+&r"(src),
393  [h]"+&r"(h)
394  : [stride]"r"((mips_reg)stride),[ff_pw_32]"f"(ff_pw_32),
395  [A]"f"(A)
396  : "memory"
397  );
398  }
399 }
400 
401 void ff_put_h264_chroma_mc4_mmi(uint8_t *dst, uint8_t *src, int stride,
402  int h, int x, int y)
403 {
404  const int A = (8 - x) * (8 - y);
405  const int B = x * (8 - y);
406  const int C = (8 - x) * y;
407  const int D = x * y;
408  const int E = B + C;
409  double ftmp[8];
410  uint64_t tmp[1];
411  mips_reg addr[1];
412  uint64_t low32;
413 
414  if (D) {
415  __asm__ volatile (
416  "xor %[ftmp0], %[ftmp0], %[ftmp0] \n\t"
417  "dli %[tmp0], 0x06 \n\t"
418  "pshufh %[A], %[A], %[ftmp0] \n\t"
419  "pshufh %[B], %[B], %[ftmp0] \n\t"
420  "mtc1 %[tmp0], %[ftmp7] \n\t"
421  "pshufh %[C], %[C], %[ftmp0] \n\t"
422  "pshufh %[D], %[D], %[ftmp0] \n\t"
423  "1: \n\t"
424  PTR_ADDU "%[addr0], %[src], %[stride] \n\t"
425  "uld %[low32], 0x00(%[src]) \n\t"
426  "mtc1 %[low32], %[ftmp1] \n\t"
427  "uld %[low32], 0x01(%[src]) \n\t"
428  "mtc1 %[low32], %[ftmp2] \n\t"
429  "uld %[low32], 0x00(%[addr0]) \n\t"
430  "mtc1 %[low32], %[ftmp3] \n\t"
431  "uld %[low32], 0x01(%[addr0]) \n\t"
432  "mtc1 %[low32], %[ftmp4] \n\t"
433 
434  "punpcklbh %[ftmp5], %[ftmp1], %[ftmp0] \n\t"
435  "punpcklbh %[ftmp6], %[ftmp2], %[ftmp0] \n\t"
436  "pmullh %[ftmp5], %[ftmp5], %[A] \n\t"
437  "pmullh %[ftmp6], %[ftmp6], %[B] \n\t"
438  "paddh %[ftmp1], %[ftmp5], %[ftmp6] \n\t"
439 
440  "punpcklbh %[ftmp5], %[ftmp3], %[ftmp0] \n\t"
441  "punpcklbh %[ftmp6], %[ftmp4], %[ftmp0] \n\t"
442  "pmullh %[ftmp5], %[ftmp5], %[C] \n\t"
443  "pmullh %[ftmp6], %[ftmp6], %[D] \n\t"
444  "paddh %[ftmp2], %[ftmp5], %[ftmp6] \n\t"
445 
446  "paddh %[ftmp1], %[ftmp1], %[ftmp2] \n\t"
447  "paddh %[ftmp1], %[ftmp1], %[ff_pw_32] \n\t"
448  "psrlh %[ftmp1], %[ftmp1], %[ftmp7] \n\t"
449  "packushb %[ftmp1], %[ftmp1], %[ftmp0] \n\t"
450  "addi %[h], %[h], -0x01 \n\t"
451  "swc1 %[ftmp1], 0x00(%[dst]) \n\t"
452  PTR_ADDU "%[src], %[src], %[stride] \n\t"
453  PTR_ADDU "%[dst], %[dst], %[stride] \n\t"
454  "bnez %[h], 1b \n\t"
455  : [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]),
456  [ftmp2]"=&f"(ftmp[2]), [ftmp3]"=&f"(ftmp[3]),
457  [ftmp4]"=&f"(ftmp[4]), [ftmp5]"=&f"(ftmp[5]),
458  [ftmp6]"=&f"(ftmp[6]), [ftmp7]"=&f"(ftmp[7]),
459  [tmp0]"=&r"(tmp[0]),
460  [addr0]"=&r"(addr[0]),
461  [dst]"+&r"(dst), [src]"+&r"(src),
462  [h]"+&r"(h),
463  [low32]"=&r"(low32)
464  : [stride]"r"((mips_reg)stride),[ff_pw_32]"f"(ff_pw_32),
465  [A]"f"(A), [B]"f"(B),
466  [C]"f"(C), [D]"f"(D)
467  : "memory"
468  );
469  } else if (E) {
470  const int step = C ? stride : 1;
471 
472  __asm__ volatile (
473  "xor %[ftmp0], %[ftmp0], %[ftmp0] \n\t"
474  "dli %[tmp0], 0x06 \n\t"
475  "pshufh %[A], %[A], %[ftmp0] \n\t"
476  "pshufh %[E], %[E], %[ftmp0] \n\t"
477  "mtc1 %[tmp0], %[ftmp5] \n\t"
478  "1: \n\t"
479  PTR_ADDU "%[addr0], %[src], %[step] \n\t"
480  "uld %[low32], 0x00(%[src]) \n\t"
481  "mtc1 %[low32], %[ftmp1] \n\t"
482  "uld %[low32], 0x00(%[addr0]) \n\t"
483  "mtc1 %[low32], %[ftmp2] \n\t"
484 
485  "punpcklbh %[ftmp3], %[ftmp1], %[ftmp0] \n\t"
486  "punpcklbh %[ftmp4], %[ftmp2], %[ftmp0] \n\t"
487  "pmullh %[ftmp3], %[ftmp3], %[A] \n\t"
488  "pmullh %[ftmp4], %[ftmp4], %[E] \n\t"
489  "paddh %[ftmp1], %[ftmp3], %[ftmp4] \n\t"
490 
491  "paddh %[ftmp1], %[ftmp1], %[ff_pw_32] \n\t"
492  "psrlh %[ftmp1], %[ftmp1], %[ftmp5] \n\t"
493  "packushb %[ftmp1], %[ftmp1], %[ftmp0] \n\t"
494  "addi %[h], %[h], -0x01 \n\t"
495  "swc1 %[ftmp1], 0x00(%[dst]) \n\t"
496  PTR_ADDU "%[src], %[src], %[stride] \n\t"
497  PTR_ADDU "%[dst], %[dst], %[stride] \n\t"
498  "bnez %[h], 1b \n\t"
499  : [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]),
500  [ftmp2]"=&f"(ftmp[2]), [ftmp3]"=&f"(ftmp[3]),
501  [ftmp4]"=&f"(ftmp[4]), [ftmp5]"=&f"(ftmp[5]),
502  [tmp0]"=&r"(tmp[0]),
503  [addr0]"=&r"(addr[0]),
504  [dst]"+&r"(dst), [src]"+&r"(src),
505  [h]"+&r"(h),
506  [low32]"=&r"(low32)
507  : [stride]"r"((mips_reg)stride),[step]"r"((mips_reg)step),
508  [ff_pw_32]"f"(ff_pw_32),
509  [A]"f"(A), [E]"f"(E)
510  : "memory"
511  );
512  } else {
513  __asm__ volatile (
514  "xor %[ftmp0], %[ftmp0], %[ftmp0] \n\t"
515  "dli %[tmp0], 0x06 \n\t"
516  "pshufh %[A], %[A], %[ftmp0] \n\t"
517  "mtc1 %[tmp0], %[ftmp3] \n\t"
518  "1: \n\t"
519  "uld %[low32], 0x00(%[src]) \n\t"
520  "mtc1 %[low32], %[ftmp1] \n\t"
521  "punpcklbh %[ftmp2], %[ftmp1], %[ftmp0] \n\t"
522  "pmullh %[ftmp1], %[ftmp2], %[A] \n\t"
523  "paddh %[ftmp1], %[ftmp1], %[ff_pw_32] \n\t"
524  "psrlh %[ftmp1], %[ftmp1], %[ftmp3] \n\t"
525  "packushb %[ftmp1], %[ftmp1], %[ftmp0] \n\t"
526  PTR_ADDU "%[src], %[src], %[stride] \n\t"
527  "swc1 %[ftmp1], 0x00(%[dst]) \n\t"
528  PTR_ADDU "%[dst], %[dst], %[stride] \n\t"
529 
530  "uld %[low32], 0x00(%[src]) \n\t"
531  "mtc1 %[low32], %[ftmp1] \n\t"
532  "punpcklbh %[ftmp2], %[ftmp1], %[ftmp0] \n\t"
533  "pmullh %[ftmp1], %[ftmp2], %[A] \n\t"
534  "paddh %[ftmp1], %[ftmp1], %[ff_pw_32] \n\t"
535  "psrlh %[ftmp1], %[ftmp1], %[ftmp3] \n\t"
536  "packushb %[ftmp1], %[ftmp1], %[ftmp0] \n\t"
537  "addi %[h], %[h], -0x02 \n\t"
538  "swc1 %[ftmp1], 0x00(%[dst]) \n\t"
539 
540  PTR_ADDU "%[src], %[src], %[stride] \n\t"
541  PTR_ADDU "%[dst], %[dst], %[stride] \n\t"
542  "bnez %[h], 1b \n\t"
543  : [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]),
544  [ftmp2]"=&f"(ftmp[2]), [ftmp3]"=&f"(ftmp[3]),
545  [tmp0]"=&r"(tmp[0]),
546  [dst]"+&r"(dst), [src]"+&r"(src),
547  [h]"+&r"(h),
548  [low32]"=&r"(low32)
549  : [stride]"r"((mips_reg)stride),[ff_pw_32]"f"(ff_pw_32),
550  [A]"f"(A)
551  : "memory"
552  );
553  }
554 }
555 
556 void ff_avg_h264_chroma_mc4_mmi(uint8_t *dst, uint8_t *src, int stride,
557  int h, int x, int y)
558 {
559  const int A = (8 - x) *(8 - y);
560  const int B = x * (8 - y);
561  const int C = (8 - x) * y;
562  const int D = x * y;
563  const int E = B + C;
564  double ftmp[8];
565  uint64_t tmp[1];
566  mips_reg addr[1];
567  uint64_t low32;
568 
569  if (D) {
570  __asm__ volatile (
571  "xor %[ftmp0], %[ftmp0], %[ftmp0] \n\t"
572  "dli %[tmp0], 0x06 \n\t"
573  "pshufh %[A], %[A], %[ftmp0] \n\t"
574  "pshufh %[B], %[B], %[ftmp0] \n\t"
575  "mtc1 %[tmp0], %[ftmp7] \n\t"
576  "pshufh %[C], %[C], %[ftmp0] \n\t"
577  "pshufh %[D], %[D], %[ftmp0] \n\t"
578  "1: \n\t"
579  PTR_ADDU "%[addr0], %[src], %[stride] \n\t"
580  "uld %[low32], 0x00(%[src]) \n\t"
581  "mtc1 %[low32], %[ftmp1] \n\t"
582  "uld %[low32], 0x01(%[src]) \n\t"
583  "mtc1 %[low32], %[ftmp2] \n\t"
584  "uld %[low32], 0x00(%[addr0]) \n\t"
585  "mtc1 %[low32], %[ftmp3] \n\t"
586  "uld %[low32], 0x01(%[addr0]) \n\t"
587  "mtc1 %[low32], %[ftmp4] \n\t"
588 
589  "punpcklbh %[ftmp5], %[ftmp1], %[ftmp0] \n\t"
590  "punpcklbh %[ftmp6], %[ftmp2], %[ftmp0] \n\t"
591  "pmullh %[ftmp5], %[ftmp5], %[A] \n\t"
592  "pmullh %[ftmp6], %[ftmp6], %[B] \n\t"
593  "paddh %[ftmp1], %[ftmp5], %[ftmp6] \n\t"
594 
595  "punpcklbh %[ftmp5], %[ftmp3], %[ftmp0] \n\t"
596  "punpcklbh %[ftmp6], %[ftmp4], %[ftmp0] \n\t"
597  "pmullh %[ftmp5], %[ftmp5], %[C] \n\t"
598  "pmullh %[ftmp6], %[ftmp6], %[D] \n\t"
599  "paddh %[ftmp2], %[ftmp5], %[ftmp6] \n\t"
600 
601  "paddh %[ftmp1], %[ftmp1], %[ftmp2] \n\t"
602  "paddh %[ftmp1], %[ftmp1], %[ff_pw_32] \n\t"
603  "psrlh %[ftmp1], %[ftmp1], %[ftmp7] \n\t"
604  "packushb %[ftmp1], %[ftmp1], %[ftmp0] \n\t"
605  "lwc1 %[ftmp2], 0x00(%[dst]) \n\t"
606  "pavgb %[ftmp1], %[ftmp1], %[ftmp2] \n\t"
607  "addi %[h], %[h], -0x01 \n\t"
608  "swc1 %[ftmp1], 0x00(%[dst]) \n\t"
609  PTR_ADDU "%[src], %[src], %[stride] \n\t"
610  PTR_ADDU "%[dst], %[dst], %[stride] \n\t"
611  "bnez %[h], 1b \n\t"
612  : [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]),
613  [ftmp2]"=&f"(ftmp[2]), [ftmp3]"=&f"(ftmp[3]),
614  [ftmp4]"=&f"(ftmp[4]), [ftmp5]"=&f"(ftmp[5]),
615  [ftmp6]"=&f"(ftmp[6]), [ftmp7]"=&f"(ftmp[7]),
616  [tmp0]"=&r"(tmp[0]),
617  [addr0]"=&r"(addr[0]),
618  [dst]"+&r"(dst), [src]"+&r"(src),
619  [h]"+&r"(h),
620  [low32]"=&r"(low32)
621  : [stride]"r"((mips_reg)stride),[ff_pw_32]"f"(ff_pw_32),
622  [A]"f"(A), [B]"f"(B),
623  [C]"f"(C), [D]"f"(D)
624  : "memory"
625  );
626  } else if (E) {
627  const int step = C ? stride : 1;
628 
629  __asm__ volatile (
630  "xor %[ftmp0], %[ftmp0], %[ftmp0] \n\t"
631  "dli %[tmp0], 0x06 \n\t"
632  "pshufh %[A], %[A], %[ftmp0] \n\t"
633  "pshufh %[E], %[E], %[ftmp0] \n\t"
634  "mtc1 %[tmp0], %[ftmp5] \n\t"
635  "1: \n\t"
636  PTR_ADDU "%[addr0], %[src], %[step] \n\t"
637  "uld %[low32], 0x00(%[src]) \n\t"
638  "mtc1 %[low32], %[ftmp1] \n\t"
639  "uld %[low32], 0x00(%[addr0]) \n\t"
640  "mtc1 %[low32], %[ftmp2] \n\t"
641 
642  "punpcklbh %[ftmp3], %[ftmp1], %[ftmp0] \n\t"
643  "punpcklbh %[ftmp4], %[ftmp2], %[ftmp0] \n\t"
644  "pmullh %[ftmp3], %[ftmp3], %[A] \n\t"
645  "pmullh %[ftmp4], %[ftmp4], %[E] \n\t"
646  "paddh %[ftmp1], %[ftmp3], %[ftmp4] \n\t"
647 
648  "paddh %[ftmp1], %[ftmp1], %[ff_pw_32] \n\t"
649  "psrlh %[ftmp1], %[ftmp1], %[ftmp5] \n\t"
650  "packushb %[ftmp1], %[ftmp1], %[ftmp0] \n\t"
651  "lwc1 %[ftmp2], 0x00(%[dst]) \n\t"
652  "pavgb %[ftmp1], %[ftmp1], %[ftmp2] \n\t"
653  "addi %[h], %[h], -0x01 \n\t"
654  "swc1 %[ftmp1], 0x00(%[dst]) \n\t"
655  PTR_ADDU "%[src], %[src], %[stride] \n\t"
656  PTR_ADDU "%[dst], %[dst], %[stride] \n\t"
657  "bnez %[h], 1b \n\t"
658  : [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]),
659  [ftmp2]"=&f"(ftmp[2]), [ftmp3]"=&f"(ftmp[3]),
660  [ftmp4]"=&f"(ftmp[4]), [ftmp5]"=&f"(ftmp[5]),
661  [tmp0]"=&r"(tmp[0]),
662  [addr0]"=&r"(addr[0]),
663  [dst]"+&r"(dst), [src]"+&r"(src),
664  [h]"+&r"(h),
665  [low32]"=&r"(low32)
666  : [stride]"r"((mips_reg)stride),[step]"r"((mips_reg)step),
667  [ff_pw_32]"f"(ff_pw_32),
668  [A]"f"(A), [E]"f"(E)
669  : "memory"
670  );
671  } else {
672  __asm__ volatile (
673  "xor %[ftmp0], %[ftmp0], %[ftmp0] \n\t"
674  "dli %[tmp0], 0x06 \n\t"
675  "pshufh %[A], %[A], %[ftmp0] \n\t"
676  "mtc1 %[tmp0], %[ftmp3] \n\t"
677  "1: \n\t"
678  "uld %[low32], 0x00(%[src]) \n\t"
679  "mtc1 %[low32], %[ftmp1] \n\t"
680  "punpcklbh %[ftmp2], %[ftmp1], %[ftmp0] \n\t"
681  "pmullh %[ftmp1], %[ftmp2], %[A] \n\t"
682  "paddh %[ftmp1], %[ftmp1], %[ff_pw_32] \n\t"
683  "psrlh %[ftmp1], %[ftmp1], %[ftmp3] \n\t"
684  "packushb %[ftmp1], %[ftmp1], %[ftmp0] \n\t"
685  "lwc1 %[ftmp2], 0x00(%[dst]) \n\t"
686  "pavgb %[ftmp1], %[ftmp1], %[ftmp2] \n\t"
687  PTR_ADDU "%[src], %[src], %[stride] \n\t"
688  "swc1 %[ftmp1], 0x00(%[dst]) \n\t"
689  PTR_ADDU "%[dst], %[dst], %[stride] \n\t"
690 
691  "uld %[low32], 0x00(%[src]) \n\t"
692  "mtc1 %[low32], %[ftmp1] \n\t"
693  "punpcklbh %[ftmp2], %[ftmp1], %[ftmp0] \n\t"
694  "pmullh %[ftmp1], %[ftmp2], %[A] \n\t"
695  "paddh %[ftmp1], %[ftmp1], %[ff_pw_32] \n\t"
696  "psrlh %[ftmp1], %[ftmp1], %[ftmp3] \n\t"
697  "packushb %[ftmp1], %[ftmp1], %[ftmp0] \n\t"
698  "lwc1 %[ftmp2], 0x00(%[dst]) \n\t"
699  "pavgb %[ftmp1], %[ftmp1], %[ftmp2] \n\t"
700  "addi %[h], %[h], -0x02 \n\t"
701  "swc1 %[ftmp1], 0x00(%[dst]) \n\t"
702 
703  PTR_ADDU "%[src], %[src], %[stride] \n\t"
704  PTR_ADDU "%[dst], %[dst], %[stride] \n\t"
705  "bnez %[h], 1b \n\t"
706  : [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]),
707  [ftmp2]"=&f"(ftmp[2]), [ftmp3]"=&f"(ftmp[3]),
708  [tmp0]"=&r"(tmp[0]),
709  [dst]"+&r"(dst), [src]"+&r"(src),
710  [h]"+&r"(h),
711  [low32]"=&r"(low32)
712  : [stride]"r"((mips_reg)stride),[ff_pw_32]"f"(ff_pw_32),
713  [A]"f"(A)
714  : "memory"
715  );
716  }
717 }
mips_reg
#define mips_reg
Definition: asmdefs.h:44
asmdefs.h
MIPS assembly defines from sys/asm.h but rewritten for use with C inline assembly (rather than from w...
h
h
Definition: vp9dsp_template.c:2093
C
#define C
ff_avg_h264_chroma_mc8_mmi
void ff_avg_h264_chroma_mc8_mmi(uint8_t *dst, uint8_t *src, int stride, int h, int x, int y)
Definition: h264chroma_mmi.c:211
uint8_t
uint8_t
Definition: audio_convert.c:194
B
Definition: vf_geq.c:46
ff_put_h264_chroma_mc4_mmi
void ff_put_h264_chroma_mc4_mmi(uint8_t *dst, uint8_t *src, int stride, int h, int x, int y)
Definition: h264chroma_mmi.c:401
ff_pw_32
const uint64_t ff_pw_32
Definition: constants.c:38
A
#define A(x)
Definition: vp56_arith.h:28
ff_avg_h264_chroma_mc4_mmi
void ff_avg_h264_chroma_mc4_mmi(uint8_t *dst, uint8_t *src, int stride, int h, int x, int y)
Definition: h264chroma_mmi.c:556
E
#define E
Definition: avdct.c:32
src
#define src
Definition: vp9dsp.c:530
h264chroma_mips.h
stride
GLint GLenum GLboolean GLsizei stride
Definition: opengl_enc.c:105
D
D(D(float, sse)
Definition: rematrix_init.c:28
ff_put_h264_chroma_mc8_mmi
void ff_put_h264_chroma_mc8_mmi(uint8_t *dst, uint8_t *src, int stride, int h, int x, int y)
Definition: h264chroma_mmi.c:29
tmp
static uint8_t tmp[8]
Definition: des.c:38
PTR_ADDU
#define PTR_ADDU
Definition: asmdefs.h:47
stride
#define stride
Generated on Mon Jun 27 2016 02:34:37 for FFmpeg by   doxygen 1.8.6