1/*
2 * Copyright (c) 2012
3 *      MIPS Technologies, Inc., California.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
8 * 1. Redistributions of source code must retain the above copyright
9 *    notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 *    notice, this list of conditions and the following disclaimer in the
12 *    documentation and/or other materials provided with the distribution.
13 * 3. Neither the name of the MIPS Technologies, Inc., nor the names of its
14 *    contributors may be used to endorse or promote products derived from
15 *    this software without specific prior written permission.
16 *
17 * THIS SOFTWARE IS PROVIDED BY THE MIPS TECHNOLOGIES, INC. ``AS IS'' AND
18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20 * ARE DISCLAIMED.  IN NO EVENT SHALL THE MIPS TECHNOLOGIES, INC. BE LIABLE
21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27 * SUCH DAMAGE.
28 *
29 * Author:  Branimir Vasic (bvasic@mips.com)
30 * Author:  Zoran Lukic (zoranl@mips.com)
31 *
32 * This file is part of FFmpeg.
33 *
34 * FFmpeg is free software; you can redistribute it and/or
35 * modify it under the terms of the GNU Lesser General Public
36 * License as published by the Free Software Foundation; either
37 * version 2.1 of the License, or (at your option) any later version.
38 *
39 * FFmpeg is distributed in the hope that it will be useful,
40 * but WITHOUT ANY WARRANTY; without even the implied warranty of
41 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
42 * Lesser General Public License for more details.
43 *
44 * You should have received a copy of the GNU Lesser General Public
45 * License along with FFmpeg; if not, write to the Free Software
46 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
47 */
48
49/**
50 * @file
51 * Reference: libavutil/float_dsp.c
52 */
53
54#include "config.h"
55#include "libavutil/float_dsp.h"
56#include "libavutil/mips/asmdefs.h"
57
58#if HAVE_INLINE_ASM && HAVE_MIPSFPU
59#if !HAVE_MIPS32R6 && !HAVE_MIPS64R6
60static void vector_fmul_mips(float *dst, const float *src0, const float *src1,
61                             int len)
62{
63    int i;
64
65    if (len & 3) {
66        for (i = 0; i < len; i++)
67            dst[i] = src0[i] * src1[i];
68    } else {
69        float *d     = (float *)dst;
70        float *d_end = d + len;
71        float *s0    = (float *)src0;
72        float *s1    = (float *)src1;
73
74        float src0_0, src0_1, src0_2, src0_3;
75        float src1_0, src1_1, src1_2, src1_3;
76
77        __asm__ volatile (
78            "1:                                         \n\t"
79            "lwc1   %[src0_0],  0(%[s0])                \n\t"
80            "lwc1   %[src1_0],  0(%[s1])                \n\t"
81            "lwc1   %[src0_1],  4(%[s0])                \n\t"
82            "lwc1   %[src1_1],  4(%[s1])                \n\t"
83            "lwc1   %[src0_2],  8(%[s0])                \n\t"
84            "lwc1   %[src1_2],  8(%[s1])                \n\t"
85            "lwc1   %[src0_3],  12(%[s0])               \n\t"
86            "lwc1   %[src1_3],  12(%[s1])               \n\t"
87            "mul.s  %[src0_0],  %[src0_0],  %[src1_0]   \n\t"
88            "mul.s  %[src0_1],  %[src0_1],  %[src1_1]   \n\t"
89            "mul.s  %[src0_2],  %[src0_2],  %[src1_2]   \n\t"
90            "mul.s  %[src0_3],  %[src0_3],  %[src1_3]   \n\t"
91            "swc1   %[src0_0],  0(%[d])                 \n\t"
92            "swc1   %[src0_1],  4(%[d])                 \n\t"
93            "swc1   %[src0_2],  8(%[d])                 \n\t"
94            "swc1   %[src0_3],  12(%[d])                \n\t"
95            PTR_ADDIU "%[s0],   %[s0],      16          \n\t"
96            PTR_ADDIU "%[s1],   %[s1],      16          \n\t"
97            PTR_ADDIU "%[d],    %[d],       16          \n\t"
98            "bne    %[d],       %[d_end],   1b          \n\t"
99
100            : [src0_0]"=&f"(src0_0), [src0_1]"=&f"(src0_1),
101              [src0_2]"=&f"(src0_2), [src0_3]"=&f"(src0_3),
102              [src1_0]"=&f"(src1_0), [src1_1]"=&f"(src1_1),
103              [src1_2]"=&f"(src1_2), [src1_3]"=&f"(src1_3),
104              [d]"+r"(d), [s0]"+r"(s0), [s1]"+r"(s1)
105            : [d_end]"r"(d_end)
106            : "memory"
107        );
108    }
109}
110
111static void vector_fmul_scalar_mips(float *dst, const float *src, float mul,
112                                 int len)
113{
114    float temp0, temp1, temp2, temp3;
115    float *local_src = (float*)src;
116    float *end = local_src + len;
117
118    /* loop unrolled 4 times */
119    __asm__ volatile(
120        ".set    push                             \n\t"
121        ".set    noreorder                        \n\t"
122    "1:                                           \n\t"
123        "lwc1    %[temp0],   0(%[src])            \n\t"
124        "lwc1    %[temp1],   4(%[src])            \n\t"
125        "lwc1    %[temp2],   8(%[src])            \n\t"
126        "lwc1    %[temp3],   12(%[src])           \n\t"
127        PTR_ADDIU "%[dst],   %[dst],     16       \n\t"
128        "mul.s   %[temp0],   %[temp0],   %[mul]   \n\t"
129        "mul.s   %[temp1],   %[temp1],   %[mul]   \n\t"
130        "mul.s   %[temp2],   %[temp2],   %[mul]   \n\t"
131        "mul.s   %[temp3],   %[temp3],   %[mul]   \n\t"
132        PTR_ADDIU "%[src],   %[src],     16       \n\t"
133        "swc1    %[temp0],   -16(%[dst])          \n\t"
134        "swc1    %[temp1],   -12(%[dst])          \n\t"
135        "swc1    %[temp2],   -8(%[dst])           \n\t"
136        "bne     %[src],     %[end],     1b       \n\t"
137        " swc1   %[temp3],   -4(%[dst])           \n\t"
138        ".set    pop                              \n\t"
139
140        : [temp0]"=&f"(temp0), [temp1]"=&f"(temp1),
141          [temp2]"=&f"(temp2), [temp3]"=&f"(temp3),
142          [dst]"+r"(dst), [src]"+r"(local_src)
143        : [end]"r"(end), [mul]"f"(mul)
144        : "memory"
145    );
146}
147
148static void vector_fmul_window_mips(float *dst, const float *src0,
149                                    const float *src1, const float *win, int len)
150{
151    float * dst_j, *win_j, *src0_i, *src1_j, *dst_i, *win_i;
152    float temp, temp1, temp2, temp3;
153    float s0, s01, s1, s11;
154    float wi, wi1, wi2, wi3;
155    float wj, wj1, wj2, wj3;
156    const float * lp_end = win + len;
157
158    win_i  = (float *)win;
159    win_j  = (float *)(win + 2 * len -1);
160    src1_j = (float *)(src1 + len - 1);
161    src0_i = (float *)src0;
162    dst_i  = (float *)dst;
163    dst_j  = (float *)(dst + 2 * len -1);
164
165    /* loop unrolled 4 times */
166    __asm__ volatile (
167        "1:"
168        "lwc1    %[s1],     0(%[src1_j])                \n\t"
169        "lwc1    %[wi],     0(%[win_i])                 \n\t"
170        "lwc1    %[wj],     0(%[win_j])                 \n\t"
171        "lwc1    %[s11],   -4(%[src1_j])                \n\t"
172        "lwc1    %[wi1],    4(%[win_i])                 \n\t"
173        "lwc1    %[wj1],   -4(%[win_j])                 \n\t"
174        "lwc1    %[s0],     0(%[src0_i])                \n\t"
175        "lwc1    %[s01],    4(%[src0_i])                \n\t"
176        "mul.s   %[temp],   %[s1],   %[wi]              \n\t"
177        "mul.s   %[temp1],  %[s1],   %[wj]              \n\t"
178        "mul.s   %[temp2],  %[s11],  %[wi1]             \n\t"
179        "mul.s   %[temp3],  %[s11],  %[wj1]             \n\t"
180        "lwc1    %[s1],    -8(%[src1_j])                \n\t"
181        "lwc1    %[wi2],    8(%[win_i])                 \n\t"
182        "lwc1    %[wj2],   -8(%[win_j])                 \n\t"
183        "lwc1    %[s11],   -12(%[src1_j])               \n\t"
184        "msub.s  %[temp],   %[temp],  %[s0],  %[wj]     \n\t"
185        "madd.s  %[temp1],  %[temp1], %[s0],  %[wi]     \n\t"
186        "msub.s  %[temp2],  %[temp2], %[s01], %[wj1]    \n\t"
187        "madd.s  %[temp3],  %[temp3], %[s01], %[wi1]    \n\t"
188        "lwc1    %[wi3],    12(%[win_i])                \n\t"
189        "lwc1    %[wj3],   -12(%[win_j])                \n\t"
190        "lwc1    %[s0],     8(%[src0_i])                \n\t"
191        "lwc1    %[s01],    12(%[src0_i])               \n\t"
192        PTR_ADDIU "%[src1_j],-16                        \n\t"
193        PTR_ADDIU "%[win_i],16                          \n\t"
194        PTR_ADDIU "%[win_j],-16                         \n\t"
195        PTR_ADDIU "%[src0_i],16                         \n\t"
196        "swc1    %[temp],   0(%[dst_i])                 \n\t" /* dst[i] = s0*wj - s1*wi; */
197        "swc1    %[temp1],  0(%[dst_j])                 \n\t" /* dst[j] = s0*wi + s1*wj; */
198        "swc1    %[temp2],  4(%[dst_i])                 \n\t" /* dst[i+1] = s01*wj1 - s11*wi1; */
199        "swc1    %[temp3], -4(%[dst_j])                 \n\t" /* dst[j-1] = s01*wi1 + s11*wj1; */
200        "mul.s   %[temp],   %[s1],    %[wi2]            \n\t"
201        "mul.s   %[temp1],  %[s1],    %[wj2]            \n\t"
202        "mul.s   %[temp2],  %[s11],   %[wi3]            \n\t"
203        "mul.s   %[temp3],  %[s11],   %[wj3]            \n\t"
204        "msub.s  %[temp],   %[temp],  %[s0],  %[wj2]    \n\t"
205        "madd.s  %[temp1],  %[temp1], %[s0],  %[wi2]    \n\t"
206        "msub.s  %[temp2],  %[temp2], %[s01], %[wj3]    \n\t"
207        "madd.s  %[temp3],  %[temp3], %[s01], %[wi3]    \n\t"
208        "swc1    %[temp],   8(%[dst_i])                 \n\t" /* dst[i+2] = s0*wj2 - s1*wi2; */
209        "swc1    %[temp1], -8(%[dst_j])                 \n\t" /* dst[j-2] = s0*wi2 + s1*wj2; */
210        "swc1    %[temp2],  12(%[dst_i])                \n\t" /* dst[i+2] = s01*wj3 - s11*wi3; */
211        "swc1    %[temp3], -12(%[dst_j])                \n\t" /* dst[j-3] = s01*wi3 + s11*wj3; */
212        PTR_ADDIU "%[dst_i],16                          \n\t"
213        PTR_ADDIU "%[dst_j],-16                         \n\t"
214        "bne     %[win_i], %[lp_end], 1b                \n\t"
215        : [temp]"=&f"(temp), [temp1]"=&f"(temp1), [temp2]"=&f"(temp2),
216          [temp3]"=&f"(temp3), [src0_i]"+r"(src0_i), [win_i]"+r"(win_i),
217          [src1_j]"+r"(src1_j), [win_j]"+r"(win_j), [dst_i]"+r"(dst_i),
218          [dst_j]"+r"(dst_j), [s0] "=&f"(s0), [s01]"=&f"(s01), [s1] "=&f"(s1),
219          [s11]"=&f"(s11), [wi] "=&f"(wi), [wj] "=&f"(wj), [wi2]"=&f"(wi2),
220          [wj2]"=&f"(wj2), [wi3]"=&f"(wi3), [wj3]"=&f"(wj3), [wi1]"=&f"(wi1),
221          [wj1]"=&f"(wj1)
222        : [lp_end]"r"(lp_end)
223        : "memory"
224    );
225}
226
227static void butterflies_float_mips(float *av_restrict v1, float *av_restrict v2,
228                                int len)
229{
230    float temp0, temp1, temp2, temp3, temp4;
231    float temp5, temp6, temp7, temp8, temp9;
232    float temp10, temp11, temp12, temp13, temp14, temp15;
233    int pom;
234    pom = (len >> 2)-1;
235
236    /* loop unrolled 4 times */
237    __asm__ volatile (
238        "lwc1     %[temp0],    0(%[v1])                 \n\t"
239        "lwc1     %[temp1],    4(%[v1])                 \n\t"
240        "lwc1     %[temp2],    8(%[v1])                 \n\t"
241        "lwc1     %[temp3],    12(%[v1])                \n\t"
242        "lwc1     %[temp4],    0(%[v2])                 \n\t"
243        "lwc1     %[temp5],    4(%[v2])                 \n\t"
244        "lwc1     %[temp6],    8(%[v2])                 \n\t"
245        "lwc1     %[temp7],    12(%[v2])                \n\t"
246        "beq      %[pom],      $zero,       2f          \n\t"
247    "1:                                                 \n\t"
248        "sub.s    %[temp8],    %[temp0],    %[temp4]    \n\t"
249        "add.s    %[temp9],    %[temp0],    %[temp4]    \n\t"
250        "sub.s    %[temp10],   %[temp1],    %[temp5]    \n\t"
251        "add.s    %[temp11],   %[temp1],    %[temp5]    \n\t"
252        "sub.s    %[temp12],   %[temp2],    %[temp6]    \n\t"
253        "add.s    %[temp13],   %[temp2],    %[temp6]    \n\t"
254        "sub.s    %[temp14],   %[temp3],    %[temp7]    \n\t"
255        "add.s    %[temp15],   %[temp3],    %[temp7]    \n\t"
256        PTR_ADDIU "%[v1],      %[v1],       16          \n\t"
257        PTR_ADDIU "%[v2],      %[v2],       16          \n\t"
258        "addiu    %[pom],      %[pom],      -1          \n\t"
259        "lwc1     %[temp0],    0(%[v1])                 \n\t"
260        "lwc1     %[temp1],    4(%[v1])                 \n\t"
261        "lwc1     %[temp2],    8(%[v1])                 \n\t"
262        "lwc1     %[temp3],    12(%[v1])                \n\t"
263        "lwc1     %[temp4],    0(%[v2])                 \n\t"
264        "lwc1     %[temp5],    4(%[v2])                 \n\t"
265        "lwc1     %[temp6],    8(%[v2])                 \n\t"
266        "lwc1     %[temp7],    12(%[v2])                \n\t"
267        "swc1     %[temp9],    -16(%[v1])               \n\t"
268        "swc1     %[temp8],    -16(%[v2])               \n\t"
269        "swc1     %[temp11],   -12(%[v1])               \n\t"
270        "swc1     %[temp10],   -12(%[v2])               \n\t"
271        "swc1     %[temp13],   -8(%[v1])                \n\t"
272        "swc1     %[temp12],   -8(%[v2])                \n\t"
273        "swc1     %[temp15],   -4(%[v1])                \n\t"
274        "swc1     %[temp14],   -4(%[v2])                \n\t"
275        "bgtz     %[pom],      1b                       \n\t"
276    "2:                                                 \n\t"
277        "sub.s    %[temp8],    %[temp0],    %[temp4]    \n\t"
278        "add.s    %[temp9],    %[temp0],    %[temp4]    \n\t"
279        "sub.s    %[temp10],   %[temp1],    %[temp5]    \n\t"
280        "add.s    %[temp11],   %[temp1],    %[temp5]    \n\t"
281        "sub.s    %[temp12],   %[temp2],    %[temp6]    \n\t"
282        "add.s    %[temp13],   %[temp2],    %[temp6]    \n\t"
283        "sub.s    %[temp14],   %[temp3],    %[temp7]    \n\t"
284        "add.s    %[temp15],   %[temp3],    %[temp7]    \n\t"
285        "swc1     %[temp9],    0(%[v1])                 \n\t"
286        "swc1     %[temp8],    0(%[v2])                 \n\t"
287        "swc1     %[temp11],   4(%[v1])                 \n\t"
288        "swc1     %[temp10],   4(%[v2])                 \n\t"
289        "swc1     %[temp13],   8(%[v1])                 \n\t"
290        "swc1     %[temp12],   8(%[v2])                 \n\t"
291        "swc1     %[temp15],   12(%[v1])                \n\t"
292        "swc1     %[temp14],   12(%[v2])                \n\t"
293
294        : [v1]"+r"(v1), [v2]"+r"(v2), [pom]"+r"(pom), [temp0] "=&f" (temp0),
295          [temp1]"=&f"(temp1), [temp2]"=&f"(temp2), [temp3]"=&f"(temp3),
296          [temp4]"=&f"(temp4), [temp5]"=&f"(temp5), [temp6]"=&f"(temp6),
297          [temp7]"=&f"(temp7), [temp8]"=&f"(temp8), [temp9]"=&f"(temp9),
298          [temp10]"=&f"(temp10), [temp11]"=&f"(temp11), [temp12]"=&f"(temp12),
299          [temp13]"=&f"(temp13), [temp14]"=&f"(temp14), [temp15]"=&f"(temp15)
300        :
301        : "memory"
302    );
303}
304
305static void vector_fmul_reverse_mips(float *dst, const float *src0, const float *src1, int len){
306    int i;
307    float temp0, temp1, temp2, temp3, temp4, temp5, temp6, temp7;
308    src1 += len-1;
309
310    for(i=0; i<(len>>2); i++)
311    {
312        /* loop unrolled 4 times */
313        __asm__ volatile(
314            "lwc1      %[temp0],     0(%[src0])                 \n\t"
315            "lwc1      %[temp1],     0(%[src1])                 \n\t"
316            "lwc1      %[temp2],     4(%[src0])                 \n\t"
317            "lwc1      %[temp3],     -4(%[src1])                \n\t"
318            "lwc1      %[temp4],     8(%[src0])                 \n\t"
319            "lwc1      %[temp5],     -8(%[src1])                \n\t"
320            "lwc1      %[temp6],     12(%[src0])                \n\t"
321            "lwc1      %[temp7],     -12(%[src1])               \n\t"
322            "mul.s     %[temp0],     %[temp1],     %[temp0]     \n\t"
323            "mul.s     %[temp2],     %[temp3],     %[temp2]     \n\t"
324            "mul.s     %[temp4],     %[temp5],     %[temp4]     \n\t"
325            "mul.s     %[temp6],     %[temp7],     %[temp6]     \n\t"
326            PTR_ADDIU "%[src0],      %[src0],      16           \n\t"
327            PTR_ADDIU "%[src1],      %[src1],      -16          \n\t"
328            PTR_ADDIU "%[dst],       %[dst],       16           \n\t"
329            "swc1      %[temp0],     -16(%[dst])                \n\t"
330            "swc1      %[temp2],     -12(%[dst])                \n\t"
331            "swc1      %[temp4],     -8(%[dst])                 \n\t"
332            "swc1      %[temp6],     -4(%[dst])                 \n\t"
333
334            : [dst]"+r"(dst), [src0]"+r"(src0), [src1]"+r"(src1),
335              [temp0]"=&f"(temp0), [temp1]"=&f"(temp1),[temp2]"=&f"(temp2),
336              [temp3]"=&f"(temp3), [temp4]"=&f"(temp4), [temp5]"=&f"(temp5),
337              [temp6]"=&f"(temp6), [temp7]"=&f"(temp7)
338            :
339            : "memory"
340        );
341    }
342}
343#endif /* !HAVE_MIPS32R6 && !HAVE_MIPS64R6 */
344#endif /* HAVE_INLINE_ASM && HAVE_MIPSFPU */
345
346void ff_float_dsp_init_mips(AVFloatDSPContext *fdsp) {
347#if HAVE_INLINE_ASM && HAVE_MIPSFPU
348#if !HAVE_MIPS32R6 && !HAVE_MIPS64R6
349    fdsp->vector_fmul = vector_fmul_mips;
350    fdsp->vector_fmul_scalar  = vector_fmul_scalar_mips;
351    fdsp->vector_fmul_window = vector_fmul_window_mips;
352    fdsp->butterflies_float = butterflies_float_mips;
353    fdsp->vector_fmul_reverse = vector_fmul_reverse_mips;
354#endif /* !HAVE_MIPS32R6 && !HAVE_MIPS64R6 */
355#endif /* HAVE_INLINE_ASM && HAVE_MIPSFPU */
356}
357