The Tor Browser / file comparison
comparison: media/libspeex_resampler/src/resample_sse.h
media/libspeex_resampler/src/resample_sse.h
- changeset 0
- 6474c204b198
equal
deleted
inserted
replaced
| -1:000000000000 | 0:31f792b4e83e |
|---|---|
| 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 | |
| 40 static inline float inner_product_single(const float *a, const float *b, unsigned int len) | |
| 41 { | |
| 42 int i; | |
| 43 float ret; | |
| 44 if (1) | |
| 45 { | |
| 46 __m128 sum = _mm_setzero_ps(); | |
| 47 for (i=0;i<len;i+=8) | |
| 48 { | |
| 49 sum = _mm_add_ps(sum, _mm_mul_ps(_mm_loadu_ps(a+i), _mm_loadu_ps(b+i))); | |
| 50 sum = _mm_add_ps(sum, _mm_mul_ps(_mm_loadu_ps(a+i+4), _mm_loadu_ps(b+i+4))); | |
| 51 } | |
| 52 sum = _mm_add_ps(sum, _mm_movehl_ps(sum, sum)); | |
| 53 sum = _mm_add_ss(sum, _mm_shuffle_ps(sum, sum, 0x55)); | |
| 54 _mm_store_ss(&ret, sum); | |
| 55 } | |
| 56 else | |
| 57 { | |
| 58 ret = 0; | |
| 59 for (i=0;i<len;i++) ret += a[i] * b[i]; | |
| 60 } | |
| 61 return ret; | |
| 62 } | |
| 63 | |
| 64 #define OVERRIDE_INTERPOLATE_PRODUCT_SINGLE | |
| 65 static inline float interpolate_product_single(const float *a, const float *b, unsigned int len, const spx_uint32_t oversample, float *frac) { | |
| 66 int i; | |
| 67 float ret; | |
| 68 if (1) | |
| 69 { | |
| 70 __m128 sum = _mm_setzero_ps(); | |
| 71 __m128 f = _mm_loadu_ps(frac); | |
| 72 for(i=0;i<len;i+=2) | |
| 73 { | |
| 74 sum = _mm_add_ps(sum, _mm_mul_ps(_mm_load1_ps(a+i), _mm_loadu_ps(b+i*oversample))); | |
| 75 sum = _mm_add_ps(sum, _mm_mul_ps(_mm_load1_ps(a+i+1), _mm_loadu_ps(b+(i+1)*oversample))); | |
| 76 } | |
| 77 sum = _mm_mul_ps(f, sum); | |
| 78 sum = _mm_add_ps(sum, _mm_movehl_ps(sum, sum)); | |
| 79 sum = _mm_add_ss(sum, _mm_shuffle_ps(sum, sum, 0x55)); | |
| 80 _mm_store_ss(&ret, sum); | |
| 81 } | |
| 82 else | |
| 83 { | |
| 84 float accum[4] = {0,0,0,0}; | |
| 85 for(i=0;i<len;i++) | |
| 86 { | |
| 87 const float curr_in=a[i]; | |
| 88 accum[0] += curr_in * b[i * oversample + 0]; | |
| 89 accum[1] += curr_in * b[i * oversample + 1]; | |
| 90 accum[2] += curr_in * b[i * oversample + 2]; | |
| 91 accum[3] += curr_in * b[i * oversample + 3]; | |
| 92 } | |
| 93 ret = accum[0] * frac[0] + accum[1] * frac[1] + accum[2] * frac[2] + accum[3] * frac[3]; | |
| 94 } | |
| 95 return ret; | |
| 96 } | |
| 97 | |
| 98 #ifdef __SSE2__ | |
| 99 #include <emmintrin.h> | |
| 100 #define OVERRIDE_INNER_PRODUCT_DOUBLE | |
| 101 | |
| 102 static inline double inner_product_double(const float *a, const float *b, unsigned int len) | |
| 103 { | |
| 104 int i; | |
| 105 double ret; | |
| 106 __m128d sum = _mm_setzero_pd(); | |
| 107 __m128 t; | |
| 108 for (i=0;i<len;i+=8) | |
| 109 { | |
| 110 t = _mm_mul_ps(_mm_loadu_ps(a+i), _mm_loadu_ps(b+i)); | |
| 111 sum = _mm_add_pd(sum, _mm_cvtps_pd(t)); | |
| 112 sum = _mm_add_pd(sum, _mm_cvtps_pd(_mm_movehl_ps(t, t))); | |
| 113 | |
| 114 t = _mm_mul_ps(_mm_loadu_ps(a+i+4), _mm_loadu_ps(b+i+4)); | |
| 115 sum = _mm_add_pd(sum, _mm_cvtps_pd(t)); | |
| 116 sum = _mm_add_pd(sum, _mm_cvtps_pd(_mm_movehl_ps(t, t))); | |
| 117 } | |
| 118 sum = _mm_add_sd(sum, (__m128d) _mm_movehl_ps((__m128) sum, (__m128) sum)); | |
| 119 _mm_store_sd(&ret, sum); | |
| 120 return ret; | |
| 121 } | |
| 122 | |
| 123 #define OVERRIDE_INTERPOLATE_PRODUCT_DOUBLE | |
| 124 static inline double interpolate_product_double(const float *a, const float *b, unsigned int len, const spx_uint32_t oversample, float *frac) { | |
| 125 int i; | |
| 126 double ret; | |
| 127 __m128d sum; | |
| 128 __m128d sum1 = _mm_setzero_pd(); | |
| 129 __m128d sum2 = _mm_setzero_pd(); | |
| 130 __m128 f = _mm_loadu_ps(frac); | |
| 131 __m128d f1 = _mm_cvtps_pd(f); | |
| 132 __m128d f2 = _mm_cvtps_pd(_mm_movehl_ps(f,f)); | |
| 133 __m128 t; | |
| 134 for(i=0;i<len;i+=2) | |
| 135 { | |
| 136 t = _mm_mul_ps(_mm_load1_ps(a+i), _mm_loadu_ps(b+i*oversample)); | |
| 137 sum1 = _mm_add_pd(sum1, _mm_cvtps_pd(t)); | |
| 138 sum2 = _mm_add_pd(sum2, _mm_cvtps_pd(_mm_movehl_ps(t, t))); | |
| 139 | |
| 140 t = _mm_mul_ps(_mm_load1_ps(a+i+1), _mm_loadu_ps(b+(i+1)*oversample)); | |
| 141 sum1 = _mm_add_pd(sum1, _mm_cvtps_pd(t)); | |
| 142 sum2 = _mm_add_pd(sum2, _mm_cvtps_pd(_mm_movehl_ps(t, t))); | |
| 143 } | |
| 144 sum1 = _mm_mul_pd(f1, sum1); | |
| 145 sum2 = _mm_mul_pd(f2, sum2); | |
| 146 sum = _mm_add_pd(sum1, sum2); | |
| 147 sum = _mm_add_sd(sum, (__m128d) _mm_movehl_ps((__m128) sum, (__m128) sum)); | |
| 148 _mm_store_sd(&ret, sum); | |
| 149 return ret; | |
| 150 } | |
| 151 | |
| 152 #endif |
