Alc/mixer/mixer_sse41.cpp

   1 /**
   2  * OpenAL cross platform audio library
   3  * Copyright (C) 2014 by Timothy Arceri <t_arceri@yahoo.com.au>.
   4  * This library is free software; you can redistribute it and/or
   5  *  modify it under the terms of the GNU Library General Public
   6  *  License as published by the Free Software Foundation; either
   7  *  version 2 of the License, or (at your option) any later version.
   8  *
   9  * This library is distributed in the hope that it will be useful,
  10  *  but WITHOUT ANY WARRANTY; without even the implied warranty of
  11  *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  12  *  Library General Public License for more details.
  13  *
  14  * You should have received a copy of the GNU Library General Public
  15  *  License along with this library; if not, write to the
  16  *  Free Software Foundation, Inc.,
  17  *  51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
  18  * Or go to http://www.gnu.org/copyleft/lgpl.html
  19  */
  20
  21 #include "config.h"
  22
  23 #include <xmmintrin.h>
  24 #include <emmintrin.h>
  25 #include <smmintrin.h>
  26
  27 #include "alu.h"
  28 #include "defs.h"
  29
  30
  31 const ALfloat *Resample_lerp_SSE41(const InterpState* UNUSED(state),
  32   const ALfloat *RESTRICT src, ALsizei frac, ALint increment,
  33   ALfloat *RESTRICT dst, ALsizei numsamples)
  34 {
  35     const __m128i increment4 = _mm_set1_epi32(increment*4);
  36     const __m128 fracOne4 = _mm_set1_ps(1.0f/FRACTIONONE);
  37     const __m128i fracMask4 = _mm_set1_epi32(FRACTIONMASK);
  38     alignas(16) ALsizei pos_[4], frac_[4];
  39     __m128i frac4, pos4;
  40     ALsizei todo, pos, i;
  41
  42     ASSUME(numsamples > 0);
  43
  44     InitiatePositionArrays(frac, increment, frac_, pos_, 4);
  45     frac4 = _mm_setr_epi32(frac_[0], frac_[1], frac_[2], frac_[3]);
  46     pos4 = _mm_setr_epi32(pos_[0], pos_[1], pos_[2], pos_[3]);
  47
  48     todo = numsamples & ~3;
  49     for(i = 0;i < todo;i += 4)
  50     {
  51         const int pos0 = _mm_extract_epi32(pos4, 0);
  52         const int pos1 = _mm_extract_epi32(pos4, 1);
  53         const int pos2 = _mm_extract_epi32(pos4, 2);
  54         const int pos3 = _mm_extract_epi32(pos4, 3);
  55         const __m128 val1 = _mm_setr_ps(src[pos0  ], src[pos1  ], src[pos2  ], src[pos3  ]);
  56         const __m128 val2 = _mm_setr_ps(src[pos0+1], src[pos1+1], src[pos2+1], src[pos3+1]);
  57
  58         /* val1 + (val2-val1)*mu */
  59         const __m128 r0 = _mm_sub_ps(val2, val1);
  60         const __m128 mu = _mm_mul_ps(_mm_cvtepi32_ps(frac4), fracOne4);
  61         const __m128 out = _mm_add_ps(val1, _mm_mul_ps(mu, r0));
  62
  63         _mm_store_ps(&dst[i], out);
  64
  65         frac4 = _mm_add_epi32(frac4, increment4);
  66         pos4 = _mm_add_epi32(pos4, _mm_srli_epi32(frac4, FRACTIONBITS));
  67         frac4 = _mm_and_si128(frac4, fracMask4);
  68     }
  69
  70     /* NOTE: These four elements represent the position *after* the last four
  71      * samples, so the lowest element is the next position to resample.
  72      */
  73     pos = _mm_cvtsi128_si32(pos4);
  74     frac = _mm_cvtsi128_si32(frac4);
  75
  76     for(;i < numsamples;++i)
  77     {
  78         dst[i] = lerp(src[pos], src[pos+1], frac * (1.0f/FRACTIONONE));
  79
  80         frac += increment;
  81         pos  += frac>>FRACTIONBITS;
  82         frac &= FRACTIONMASK;
  83     }
  84     return dst;
  85 }