fork/rust-sdks

git clone https://git.julia.blue/fork/rust-sdks

Théo Monnomsoxr-sys (#438)2e4d385

main
2.0 KiB89 linesraw
1/* SoX Resampler Library      Copyright (c) 2007-16 robs@users.sourceforge.net
2 * Licence for this file: LGPL v2.1                  See LICENCE for details. */
3
4#include <assert.h>
5#include <string.h>
6#include <stdlib.h>
7
8#include "soxr-config.h"
9
10#define SIMD_ALIGNMENT (sizeof(float) * (1 + (PFFFT_DOUBLE|AVCODEC_FOUND)) * 4)
11
12void * SIMD_ALIGNED_MALLOC(size_t size)
13{
14  char * p1 = 0, * p = malloc(size + SIMD_ALIGNMENT);
15  if (p) {
16    p1 = (char *)((size_t)(p + SIMD_ALIGNMENT) & ~(SIMD_ALIGNMENT - 1));
17    *((void * *)p1 - 1) = p;
18  }
19  return p1;
20}
21
22
23
24void * SIMD_ALIGNED_CALLOC(size_t nmemb, size_t size)
25{
26  void * p = SIMD_ALIGNED_MALLOC(nmemb * size);
27  if (p)
28    memset(p, 0, nmemb * size);
29  return p;
30}
31
32
33
34void SIMD_ALIGNED_FREE(void * p1)
35{
36  if (p1)
37    free(*((void * *)p1 - 1));
38}
39
40
41
42#define PFFT_MACROS_ONLY
43#include "pffft.c"
44
45
46
47void ORDERED_CONVOLVE_SIMD(int n, void * not_used, float * a, float const * b)
48{
49  int i;
50  float ab0, ab1;
51  v4sf       *   RESTRICT   va = (v4sf       *)a;
52  v4sf const *   RESTRICT   vb = (v4sf const *)b;
53  assert(VALIGNED(a) && VALIGNED(b));
54  ab0 = a[0] * b[0], ab1 = a[1] * b[1];
55  for (i = 0; i < n / 4; i += 2) {
56    v4sf a1r = va[i+0], a1i = va[i+1];
57    v4sf b1r = vb[i+0], b1i = vb[i+1];
58    UNINTERLEAVE2(a1r, a1i, a1r, a1i);
59    UNINTERLEAVE2(b1r, b1i, b1r, b1i);
60    VCPLXMUL(a1r, a1i, b1r, b1i);
61    INTERLEAVE2(a1r, a1i, a1r, a1i);
62    va[i+0] = a1r, va[i+1] = a1i;
63  }
64  a[0] = ab0, a[1] = ab1;
65  (void)not_used;
66}
67
68
69
70void ORDERED_PARTIAL_CONVOLVE_SIMD(int n, float * a, float const * b)
71{
72  int i;
73  float ab0;
74  v4sf       *   RESTRICT   va = (v4sf       *)a;
75  v4sf const *   RESTRICT   vb = (v4sf const *)b;
76  assert(VALIGNED(a) && VALIGNED(b));
77  ab0 = a[0] * b[0];
78  for (i = 0; i < n / 4; i += 2) {
79    v4sf a1r = va[i+0], a1i = va[i+1];
80    v4sf b1r = vb[i+0], b1i = vb[i+1];
81    UNINTERLEAVE2(a1r, a1i, a1r, a1i);
82    UNINTERLEAVE2(b1r, b1i, b1r, b1i);
83    VCPLXMUL(a1r, a1i, b1r, b1i);
84    INTERLEAVE2(a1r, a1i, a1r, a1i);
85    va[i+0] = a1r, va[i+1] = a1i;
86  }
87  a[0] = ab0;
88  a[1] = b[n] * a[n] - b[n+1] * a[n+1];
89}