1/* Copyright (C) 2007-2008 Jean-Marc Valin
2 * Copyright (C) 2008 Thorvald Natvig
3 */
4/**
5   @file resample_sse.h
6   @brief Resampler functions (SSE version)
7*/
8/*
9   Redistribution and use in source and binary forms, with or without
10   modification, are permitted provided that the following conditions
11   are met:
12
13   - Redistributions of source code must retain the above copyright
14   notice, this list of conditions and the following disclaimer.
15
16   - Redistributions in binary form must reproduce the above copyright
17   notice, this list of conditions and the following disclaimer in the
18   documentation and/or other materials provided with the distribution.
19
20   - Neither the name of the Xiph.org Foundation nor the names of its
21   contributors may be used to endorse or promote products derived from
22   this software without specific prior written permission.
23
24   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
25   ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
26   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
27   A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR
28   CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
29   EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
30   PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
31   PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
32   LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
33   NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
34   SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
35*/
36
37#include <xmmintrin.h>
38
39#define OVERRIDE_INNER_PRODUCT_SINGLE
40static inline float inner_product_single(const float *a, const float *b, unsigned int len)
41{
42   int i;
43   float ret;
44   __m128 sum = _mm_setzero_ps();
45   for (i=0;i<len;i+=8)
46   {
47      sum = _mm_add_ps(sum, _mm_mul_ps(_mm_loadu_ps(a+i), _mm_loadu_ps(b+i)));
48      sum = _mm_add_ps(sum, _mm_mul_ps(_mm_loadu_ps(a+i+4), _mm_loadu_ps(b+i+4)));
49   }
50   sum = _mm_add_ps(sum, _mm_movehl_ps(sum, sum));
51   sum = _mm_add_ss(sum, _mm_shuffle_ps(sum, sum, 0x55));
52   _mm_store_ss(&ret, sum);
53   return ret;
54}
55
56#define OVERRIDE_INTERPOLATE_PRODUCT_SINGLE
57static inline float interpolate_product_single(const float *a, const float *b, unsigned int len, const spx_uint32_t oversample, float *frac) {
58  int i;
59  float ret;
60  __m128 sum = _mm_setzero_ps();
61  __m128 f = _mm_loadu_ps(frac);
62  for(i=0;i<len;i+=2)
63  {
64    sum = _mm_add_ps(sum, _mm_mul_ps(_mm_load1_ps(a+i), _mm_loadu_ps(b+i*oversample)));
65    sum = _mm_add_ps(sum, _mm_mul_ps(_mm_load1_ps(a+i+1), _mm_loadu_ps(b+(i+1)*oversample)));
66  }
67   sum = _mm_mul_ps(f, sum);
68   sum = _mm_add_ps(sum, _mm_movehl_ps(sum, sum));
69   sum = _mm_add_ss(sum, _mm_shuffle_ps(sum, sum, 0x55));
70   _mm_store_ss(&ret, sum);
71   return ret;
72}
73
74#ifdef _USE_SSE2
75#include <emmintrin.h>
76#define OVERRIDE_INNER_PRODUCT_DOUBLE
77
78static inline double inner_product_double(const float *a, const float *b, unsigned int len)
79{
80   int i;
81   double ret;
82   __m128d sum = _mm_setzero_pd();
83   __m128 t;
84   for (i=0;i<len;i+=8)
85   {
86      t = _mm_mul_ps(_mm_loadu_ps(a+i), _mm_loadu_ps(b+i));
87      sum = _mm_add_pd(sum, _mm_cvtps_pd(t));
88      sum = _mm_add_pd(sum, _mm_cvtps_pd(_mm_movehl_ps(t, t)));
89
90      t = _mm_mul_ps(_mm_loadu_ps(a+i+4), _mm_loadu_ps(b+i+4));
91      sum = _mm_add_pd(sum, _mm_cvtps_pd(t));
92      sum = _mm_add_pd(sum, _mm_cvtps_pd(_mm_movehl_ps(t, t)));
93   }
94   sum = _mm_add_sd(sum, (__m128d) _mm_movehl_ps((__m128) sum, (__m128) sum));
95   _mm_store_sd(&ret, sum);
96   return ret;
97}
98
99#define OVERRIDE_INTERPOLATE_PRODUCT_DOUBLE
100static inline double interpolate_product_double(const float *a, const float *b, unsigned int len, const spx_uint32_t oversample, float *frac) {
101  int i;
102  double ret;
103  __m128d sum;
104  __m128d sum1 = _mm_setzero_pd();
105  __m128d sum2 = _mm_setzero_pd();
106  __m128 f = _mm_loadu_ps(frac);
107  __m128d f1 = _mm_cvtps_pd(f);
108  __m128d f2 = _mm_cvtps_pd(_mm_movehl_ps(f,f));
109  __m128 t;
110  for(i=0;i<len;i+=2)
111  {
112    t = _mm_mul_ps(_mm_load1_ps(a+i), _mm_loadu_ps(b+i*oversample));
113    sum1 = _mm_add_pd(sum1, _mm_cvtps_pd(t));
114    sum2 = _mm_add_pd(sum2, _mm_cvtps_pd(_mm_movehl_ps(t, t)));
115
116    t = _mm_mul_ps(_mm_load1_ps(a+i+1), _mm_loadu_ps(b+(i+1)*oversample));
117    sum1 = _mm_add_pd(sum1, _mm_cvtps_pd(t));
118    sum2 = _mm_add_pd(sum2, _mm_cvtps_pd(_mm_movehl_ps(t, t)));
119  }
120  sum1 = _mm_mul_pd(f1, sum1);
121  sum2 = _mm_mul_pd(f2, sum2);
122  sum = _mm_add_pd(sum1, sum2);
123  sum = _mm_add_sd(sum, (__m128d) _mm_movehl_ps((__m128) sum, (__m128) sum));
124  _mm_store_sd(&ret, sum);
125  return ret;
126}
127
128#endif
129