core/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 "alnumeric.h"
  28 #include "defs.h"
  29
  30 struct SSE4Tag;
  31 struct LerpTag;
  32
  33
  34 #if defined(__GNUC__) && !defined(__clang__) && !defined(__SSE4_1__)
  35 #pragma GCC target("sse4.1")
  36 #endif
  37
  38 template<>
  39 float *Resample_<LerpTag,SSE4Tag>(const InterpState*, float *RESTRICT src, uint frac,
  40     uint increment, const al::span<float> dst)
  41 {
  42     const __m128i increment4{_mm_set1_epi32(static_cast<int>(increment*4))};
  43     const __m128 fracOne4{_mm_set1_ps(1.0f/MixerFracOne)};
  44     const __m128i fracMask4{_mm_set1_epi32(MixerFracMask)};
  45
  46     alignas(16) uint pos_[4], frac_[4];
  47     InitPosArrays(frac, increment, frac_, pos_);
  48     __m128i frac4{_mm_setr_epi32(static_cast<int>(frac_[0]), static_cast<int>(frac_[1]),
  49         static_cast<int>(frac_[2]), static_cast<int>(frac_[3]))};
  50     __m128i pos4{_mm_setr_epi32(static_cast<int>(pos_[0]), static_cast<int>(pos_[1]),
  51         static_cast<int>(pos_[2]), static_cast<int>(pos_[3]))};
  52
  53     auto dst_iter = dst.begin();
  54     for(size_t todo{dst.size()>>2};todo;--todo)
  55     {
  56         const int pos0{_mm_extract_epi32(pos4, 0)};
  57         const int pos1{_mm_extract_epi32(pos4, 1)};
  58         const int pos2{_mm_extract_epi32(pos4, 2)};
  59         const int pos3{_mm_extract_epi32(pos4, 3)};
  60         const __m128 val1{_mm_setr_ps(src[pos0  ], src[pos1  ], src[pos2  ], src[pos3  ])};
  61         const __m128 val2{_mm_setr_ps(src[pos0+1], src[pos1+1], src[pos2+1], src[pos3+1])};
  62
  63         /* val1 + (val2-val1)*mu */
  64         const __m128 r0{_mm_sub_ps(val2, val1)};
  65         const __m128 mu{_mm_mul_ps(_mm_cvtepi32_ps(frac4), fracOne4)};
  66         const __m128 out{_mm_add_ps(val1, _mm_mul_ps(mu, r0))};
  67
  68         _mm_store_ps(dst_iter, out);
  69         dst_iter += 4;
  70
  71         frac4 = _mm_add_epi32(frac4, increment4);
  72         pos4 = _mm_add_epi32(pos4, _mm_srli_epi32(frac4, MixerFracBits));
  73         frac4 = _mm_and_si128(frac4, fracMask4);
  74     }
  75
  76     if(size_t todo{dst.size()&3})
  77     {
  78         /* NOTE: These four elements represent the position *after* the last
  79          * four samples, so the lowest element is the next position to
  80          * resample.
  81          */
  82         src += static_cast<uint>(_mm_cvtsi128_si32(pos4));
  83         frac = static_cast<uint>(_mm_cvtsi128_si32(frac4));
  84
  85         do {
  86             *(dst_iter++) = lerp(src[0], src[1], static_cast<float>(frac) * (1.0f/MixerFracOne));
  87
  88             frac += increment;
  89             src  += frac>>MixerFracBits;
  90             frac &= MixerFracMask;
  91         } while(--todo);
  92     }
  93     return dst.data();
  94 }