1 /* Copyright (c) 2013 Jean-Marc Valin and John Ridges 2 Copyright (c) 2014, Cisco Systems, INC MingXiang WeiZhou MinPeng YanWang*/ 3 /** 4 @file pitch_sse.h 5 @brief Pitch analysis 6 */ 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 THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS 21 ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT 22 LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR 23 A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER 24 OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, 25 EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, 26 PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR 27 PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF 28 LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING 29 NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS 30 SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 31 */ 32 33 #ifndef PITCH_SSE_H 34 #define PITCH_SSE_H 35 36 #if defined(HAVE_CONFIG_H) 37 #include "config.h" 38 #endif 39 40 #if defined(OPUS_X86_MAY_HAVE_SSE4_1) && defined(FIXED_POINT) 41 void xcorr_kernel_sse4_1( 42 const opus_int16 *x, 43 const opus_int16 *y, 44 opus_val32 sum[4], 45 int len); 46 #endif 47 48 #if defined(OPUS_X86_MAY_HAVE_SSE) && !defined(FIXED_POINT) 49 void xcorr_kernel_sse( 50 const opus_val16 *x, 51 const opus_val16 *y, 52 opus_val32 sum[4], 53 int len); 54 #endif 55 56 #if defined(OPUS_X86_PRESUME_SSE4_1) && defined(FIXED_POINT) 57 #define OVERRIDE_XCORR_KERNEL 58 #define xcorr_kernel(x, y, sum, len, arch) \ 59 ((void)arch, xcorr_kernel_sse4_1(x, y, sum, len)) 60 61 #elif defined(OPUS_X86_PRESUME_SSE) && !defined(FIXED_POINT) 62 #define OVERRIDE_XCORR_KERNEL 63 #define xcorr_kernel(x, y, sum, len, arch) \ 64 ((void)arch, xcorr_kernel_sse(x, y, sum, len)) 65 66 #elif defined(OPUS_HAVE_RTCD) && ((defined(OPUS_X86_MAY_HAVE_SSE4_1) && defined(FIXED_POINT)) || (defined(OPUS_X86_MAY_HAVE_SSE) && !defined(FIXED_POINT))) 67 68 extern void (*const XCORR_KERNEL_IMPL[OPUS_ARCHMASK + 1])( 69 const opus_val16 *x, 70 const opus_val16 *y, 71 opus_val32 sum[4], 72 int len); 73 74 #define OVERRIDE_XCORR_KERNEL 75 #define xcorr_kernel(x, y, sum, len, arch) \ 76 ((*XCORR_KERNEL_IMPL[(arch) & OPUS_ARCHMASK])(x, y, sum, len)) 77 78 #endif 79 80 #if defined(OPUS_X86_MAY_HAVE_SSE4_1) && defined(FIXED_POINT) 81 opus_val32 celt_inner_prod_sse4_1( 82 const opus_int16 *x, 83 const opus_int16 *y, 84 int N); 85 #endif 86 87 #if defined(OPUS_X86_MAY_HAVE_SSE2) && defined(FIXED_POINT) 88 opus_val32 celt_inner_prod_sse2( 89 const opus_int16 *x, 90 const opus_int16 *y, 91 int N); 92 #endif 93 94 #if defined(OPUS_X86_MAY_HAVE_SSE) && !defined(FIXED_POINT) 95 opus_val32 celt_inner_prod_sse( 96 const opus_val16 *x, 97 const opus_val16 *y, 98 int N); 99 #endif 100 101 102 #if defined(OPUS_X86_PRESUME_SSE4_1) && defined(FIXED_POINT) 103 #define OVERRIDE_CELT_INNER_PROD 104 #define celt_inner_prod(x, y, N, arch) \ 105 ((void)arch, celt_inner_prod_sse4_1(x, y, N)) 106 107 #elif defined(OPUS_X86_PRESUME_SSE2) && defined(FIXED_POINT) && !defined(OPUS_X86_MAY_HAVE_SSE4_1) 108 #define OVERRIDE_CELT_INNER_PROD 109 #define celt_inner_prod(x, y, N, arch) \ 110 ((void)arch, celt_inner_prod_sse2(x, y, N)) 111 112 #elif defined(OPUS_X86_PRESUME_SSE) && !defined(FIXED_POINT) 113 #define OVERRIDE_CELT_INNER_PROD 114 #define celt_inner_prod(x, y, N, arch) \ 115 ((void)arch, celt_inner_prod_sse(x, y, N)) 116 117 118 #elif defined(OPUS_HAVE_RTCD) && (((defined(OPUS_X86_MAY_HAVE_SSE4_1) || defined(OPUS_X86_MAY_HAVE_SSE2)) && defined(FIXED_POINT)) || \ 119 (defined(OPUS_X86_MAY_HAVE_SSE) && !defined(FIXED_POINT))) 120 121 extern opus_val32 (*const CELT_INNER_PROD_IMPL[OPUS_ARCHMASK + 1])( 122 const opus_val16 *x, 123 const opus_val16 *y, 124 int N); 125 126 #define OVERRIDE_CELT_INNER_PROD 127 #define celt_inner_prod(x, y, N, arch) \ 128 ((*CELT_INNER_PROD_IMPL[(arch) & OPUS_ARCHMASK])(x, y, N)) 129 130 #endif 131 132 #if defined(OPUS_X86_MAY_HAVE_SSE) && !defined(FIXED_POINT) 133 134 void dual_inner_prod_sse(const opus_val16 *x, 135 const opus_val16 *y01, 136 const opus_val16 *y02, 137 int N, 138 opus_val32 *xy1, 139 opus_val32 *xy2); 140 141 void comb_filter_const_sse(opus_val32 *y, 142 opus_val32 *x, 143 int T, 144 int N, 145 opus_val16 g10, 146 opus_val16 g11, 147 opus_val16 g12); 148 149 150 #if defined(OPUS_X86_PRESUME_SSE) 151 #define OVERRIDE_DUAL_INNER_PROD 152 #define OVERRIDE_COMB_FILTER_CONST 153 # define dual_inner_prod(x, y01, y02, N, xy1, xy2, arch) \ 154 ((void)(arch),dual_inner_prod_sse(x, y01, y02, N, xy1, xy2)) 155 156 # define comb_filter_const(y, x, T, N, g10, g11, g12, arch) \ 157 ((void)(arch),comb_filter_const_sse(y, x, T, N, g10, g11, g12)) 158 #elif defined(OPUS_HAVE_RTCD) 159 160 #define OVERRIDE_DUAL_INNER_PROD 161 #define OVERRIDE_COMB_FILTER_CONST 162 extern void (*const DUAL_INNER_PROD_IMPL[OPUS_ARCHMASK + 1])( 163 const opus_val16 *x, 164 const opus_val16 *y01, 165 const opus_val16 *y02, 166 int N, 167 opus_val32 *xy1, 168 opus_val32 *xy2); 169 170 #define dual_inner_prod(x, y01, y02, N, xy1, xy2, arch) \ 171 ((*DUAL_INNER_PROD_IMPL[(arch) & OPUS_ARCHMASK])(x, y01, y02, N, xy1, xy2)) 172 173 extern void (*const COMB_FILTER_CONST_IMPL[OPUS_ARCHMASK + 1])( 174 opus_val32 *y, 175 opus_val32 *x, 176 int T, 177 int N, 178 opus_val16 g10, 179 opus_val16 g11, 180 opus_val16 g12); 181 182 #define comb_filter_const(y, x, T, N, g10, g11, g12, arch) \ 183 ((*COMB_FILTER_CONST_IMPL[(arch) & OPUS_ARCHMASK])(y, x, T, N, g10, g11, g12)) 184 185 #define NON_STATIC_COMB_FILTER_CONST_C 186 187 #endif 188 189 void celt_pitch_xcorr_avx2(const float *_x, const float *_y, float *xcorr, int len, int max_pitch, int arch); 190 191 #if defined(OPUS_X86_PRESUME_AVX2) 192 193 #define OVERRIDE_PITCH_XCORR 194 # define celt_pitch_xcorr celt_pitch_xcorr_avx2 195 196 #elif defined(OPUS_HAVE_RTCD) && defined(OPUS_X86_MAY_HAVE_AVX2) 197 198 #define OVERRIDE_PITCH_XCORR 199 extern void (*const PITCH_XCORR_IMPL[OPUS_ARCHMASK + 1])( 200 const float *_x, 201 const float *_y, 202 float *xcorr, 203 int len, 204 int max_pitch, 205 int arch 206 ); 207 208 #define celt_pitch_xcorr(_x, _y, xcorr, len, max_pitch, arch) \ 209 ((*PITCH_XCORR_IMPL[(arch) & OPUS_ARCHMASK])(_x, _y, xcorr, len, max_pitch, arch)) 210 211 212 #endif /* OPUS_X86_PRESUME_AVX2 && !OPUS_HAVE_RTCD */ 213 214 #endif /* OPUS_X86_MAY_HAVE_SSE && !FIXED_POINT */ 215 216 #endif 217