2 * Copyright (c) 2015 Janne Grunau
4 * This file is part of FFmpeg.
6 * FFmpeg is free software; you can redistribute it and/or modify
7 * it under the terms of the GNU General Public License as published by
8 * the Free Software Foundation; either version 2 of the License, or
9 * (at your option) any later version.
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
14 * GNU General Public License for more details.
16 * You should have received a copy of the GNU General Public License along
17 * with FFmpeg; if not, write to the Free Software Foundation, Inc.,
18 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
26 #include "libavutil/internal.h"
27 #include "libavutil/intfloat.h"
28 #include "libavutil/mem_internal.h"
30 #include "libavcodec/dcadata.h"
31 #include "libavcodec/synth_filter.h"
37 #define randomize_input() \
40 for (i = 0; i < BUF_SIZE; i++) { \
41 float f = (float)rnd() / (UINT_MAX >> 5) - 16.0f; \
46 void checkasm_check_synth_filter(void)
49 SynthFilterContext synth;
51 ff_mdct_init(&imdct, 6, 1, 1.0);
52 ff_synth_filter_init(&synth);
54 if (check_func(synth.synth_filter_float, "synth_filter_float")) {
55 LOCAL_ALIGNED(32, float, out0, [BUF_SIZE]);
56 LOCAL_ALIGNED(32, float, out1, [BUF_SIZE]);
57 LOCAL_ALIGNED(32, float, out_b, [BUF_SIZE]);
58 LOCAL_ALIGNED(32, float, in, [BUF_SIZE]);
59 LOCAL_ALIGNED(32, float, buf2_0, [BUF_SIZE]);
60 LOCAL_ALIGNED(32, float, buf2_1, [BUF_SIZE]);
61 LOCAL_ALIGNED(32, float, buf2_b, [BUF_SIZE]);
62 LOCAL_ALIGNED(32, float, buf0, [512]);
63 LOCAL_ALIGNED(32, float, buf1, [512]);
64 LOCAL_ALIGNED(32, float, buf_b, [512]);
66 int i, offset0 = 0, offset1 = 0, offset_b = 0;
68 declare_func(void, FFTContext *, float *, int *, float[32], const float[512],
69 float[32], float[32], float);
71 memset(buf2_0, 0, sizeof(*buf2_0) * BUF_SIZE);
72 memset(buf2_1, 0, sizeof(*buf2_1) * BUF_SIZE);
73 memset(buf2_b, 0, sizeof(*buf2_b) * BUF_SIZE);
74 memset(buf0, 0, sizeof(*buf2_0) * 512);
75 memset(buf1, 0, sizeof(*buf2_1) * 512);
76 memset(buf_b, 0, sizeof(*buf2_b) * 512);
78 /* more than 1 synth_buf_offset wrap-around */
79 for (i = 0; i < 20; i++) {
81 const float * window = (i & 1) ? ff_dca_fir_32bands_perfect : ff_dca_fir_32bands_nonperfect;
83 memset(out0, 0, sizeof(*out0) * BUF_SIZE);
84 memset(out1, 0, sizeof(*out1) * BUF_SIZE);
85 memset(out_b, 0, sizeof(*out_b) * BUF_SIZE);
89 call_ref(&imdct, buf0, &offset0, buf2_0, window,
91 call_new(&imdct, buf1, &offset1, buf2_1, window,
94 if (offset0 != offset1) {
96 fprintf(stderr, "offsets do not match: %d, %d", offset0, offset1);
100 for (j = 0; j < BUF_SIZE; j++) {
101 if (!float_near_abs_eps_ulp(out0[j], out1[j], 7.0e-7, 16) ||
102 !float_near_abs_eps_ulp(buf2_0[j], buf2_1[j], 7.0e-7, 16)) {
103 union av_intfloat32 o0, o1, b0, b1;
106 o0.f = out0[j]; o1.f = out1[j];
107 b0.f = buf2_0[j], b1.f = buf2_1[j];
108 fprintf(stderr, "out: %11g (0x%08x); %11g (0x%08x); abs diff %11g\n",
109 o0.f, o0.i, o1.f, o1.i, fabsf(o0.f - o1.f));
110 fprintf(stderr, "buf2: %11g (0x%08x); %11g (0x%08x); abs diff %11g\n",
111 b0.f, b0.i, b1.f, b1.i, fabsf(b0.f - b1.f));
116 bench_new(&imdct, buf_b, &offset_b, buf2_b, window,
122 report("synth_filter");