Source code
Revision control
Copy as Markdown
Other Tools
////////////////////////////////////////////////////////////////////////////////
///
/// MMX optimized routines. All MMX optimized functions have been gathered into
/// this single source code file, regardless to their class or original source
/// code file, in order to ease porting the library to other compiler and
/// processor platforms.
///
/// The MMX-optimizations are programmed using MMX compiler intrinsics that
/// are supported both by Microsoft Visual C++ and GCC compilers, so this file
/// should compile with both toolsets.
///
/// NOTICE: If using Visual Studio 6.0, you'll need to install the "Visual C++
/// 6.0 processor pack" update to support compiler intrinsic syntax. The update
/// is available for download at Microsoft Developers Network, see here:
///
/// Author : Copyright (c) Olli Parviainen
/// Author e-mail : oparviai 'at' iki.fi
///
////////////////////////////////////////////////////////////////////////////////
//
// License :
//
// SoundTouch audio processing library
// Copyright (c) Olli Parviainen
//
// This library is free software; you can redistribute it and/or
// modify it under the terms of the GNU Lesser General Public
// License as published by the Free Software Foundation; either
// version 2.1 of the License, or (at your option) any later version.
//
// This library is distributed in the hope that it will be useful,
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
// Lesser General Public License for more details.
//
// You should have received a copy of the GNU Lesser General Public
// License along with this library; if not, write to the Free Software
// Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
//
////////////////////////////////////////////////////////////////////////////////
#include "STTypes.h"
#ifdef SOUNDTOUCH_ALLOW_MMX
// MMX routines available only with integer sample type
using namespace soundtouch;
//////////////////////////////////////////////////////////////////////////////
//
// implementation of MMX optimized functions of class 'TDStretchMMX'
//
//////////////////////////////////////////////////////////////////////////////
#include "TDStretch.h"
#include <mmintrin.h>
#include <limits.h>
#include <math.h>
// Calculates cross correlation of two buffers
double TDStretchMMX::calcCrossCorr(const short *pV1, const short *pV2, double &dnorm)
{
const __m64 *pVec1, *pVec2;
__m64 shifter;
__m64 accu, normaccu;
long corr, norm;
int i;
pVec1 = (__m64*)pV1;
pVec2 = (__m64*)pV2;
shifter = _m_from_int(overlapDividerBitsNorm);
normaccu = accu = _mm_setzero_si64();
// Process 4 parallel sets of 2 * stereo samples or 4 * mono samples
// during each round for improved CPU-level parallellization.
for (i = 0; i < channels * overlapLength / 16; i ++)
{
__m64 temp, temp2;
// dictionary of instructions:
// _m_pmaddwd : 4*16bit multiply-add, resulting two 32bits = [a0*b0+a1*b1 ; a2*b2+a3*b3]
// _mm_add_pi32 : 2*32bit add
// _m_psrad : 32bit right-shift
temp = _mm_add_pi32(_mm_sra_pi32(_mm_madd_pi16(pVec1[0], pVec2[0]), shifter),
_mm_sra_pi32(_mm_madd_pi16(pVec1[1], pVec2[1]), shifter));
temp2 = _mm_add_pi32(_mm_sra_pi32(_mm_madd_pi16(pVec1[0], pVec1[0]), shifter),
_mm_sra_pi32(_mm_madd_pi16(pVec1[1], pVec1[1]), shifter));
accu = _mm_add_pi32(accu, temp);
normaccu = _mm_add_pi32(normaccu, temp2);
temp = _mm_add_pi32(_mm_sra_pi32(_mm_madd_pi16(pVec1[2], pVec2[2]), shifter),
_mm_sra_pi32(_mm_madd_pi16(pVec1[3], pVec2[3]), shifter));
temp2 = _mm_add_pi32(_mm_sra_pi32(_mm_madd_pi16(pVec1[2], pVec1[2]), shifter),
_mm_sra_pi32(_mm_madd_pi16(pVec1[3], pVec1[3]), shifter));
accu = _mm_add_pi32(accu, temp);
normaccu = _mm_add_pi32(normaccu, temp2);
pVec1 += 4;
pVec2 += 4;
}
// copy hi-dword of mm0 to lo-dword of mm1, then sum mmo+mm1
// and finally store the result into the variable "corr"
accu = _mm_add_pi32(accu, _mm_srli_si64(accu, 32));
corr = _m_to_int(accu);
normaccu = _mm_add_pi32(normaccu, _mm_srli_si64(normaccu, 32));
norm = _m_to_int(normaccu);
// Clear MMS state
_m_empty();
if (norm > (long)maxnorm)
{
// modify 'maxnorm' inside critical section to avoid multi-access conflict if in OpenMP mode
#pragma omp critical
if (norm > (long)maxnorm)
{
maxnorm = norm;
}
}
// Normalize result by dividing by sqrt(norm) - this step is easiest
// done using floating point operation
dnorm = (double)norm;
return (double)corr / sqrt(dnorm < 1e-9 ? 1.0 : dnorm);
// Note: Warning about the missing EMMS instruction is harmless
// as it'll be called elsewhere.
}
/// Update cross-correlation by accumulating "norm" coefficient by previously calculated value
double TDStretchMMX::calcCrossCorrAccumulate(const short *pV1, const short *pV2, double &dnorm)
{
const __m64 *pVec1, *pVec2;
__m64 shifter;
__m64 accu;
long corr, lnorm;
int i;
// cancel first normalizer tap from previous round
lnorm = 0;
for (i = 1; i <= channels; i ++)
{
lnorm -= (pV1[-i] * pV1[-i]) >> overlapDividerBitsNorm;
}
pVec1 = (__m64*)pV1;
pVec2 = (__m64*)pV2;
shifter = _m_from_int(overlapDividerBitsNorm);
accu = _mm_setzero_si64();
// Process 4 parallel sets of 2 * stereo samples or 4 * mono samples
// during each round for improved CPU-level parallellization.
for (i = 0; i < channels * overlapLength / 16; i ++)
{
__m64 temp;
// dictionary of instructions:
// _m_pmaddwd : 4*16bit multiply-add, resulting two 32bits = [a0*b0+a1*b1 ; a2*b2+a3*b3]
// _mm_add_pi32 : 2*32bit add
// _m_psrad : 32bit right-shift
temp = _mm_add_pi32(_mm_sra_pi32(_mm_madd_pi16(pVec1[0], pVec2[0]), shifter),
_mm_sra_pi32(_mm_madd_pi16(pVec1[1], pVec2[1]), shifter));
accu = _mm_add_pi32(accu, temp);
temp = _mm_add_pi32(_mm_sra_pi32(_mm_madd_pi16(pVec1[2], pVec2[2]), shifter),
_mm_sra_pi32(_mm_madd_pi16(pVec1[3], pVec2[3]), shifter));
accu = _mm_add_pi32(accu, temp);
pVec1 += 4;
pVec2 += 4;
}
// copy hi-dword of mm0 to lo-dword of mm1, then sum mmo+mm1
// and finally store the result into the variable "corr"
accu = _mm_add_pi32(accu, _mm_srli_si64(accu, 32));
corr = _m_to_int(accu);
// Clear MMS state
_m_empty();
// update normalizer with last samples of this round
pV1 = (short *)pVec1;
for (int j = 1; j <= channels; j ++)
{
lnorm += (pV1[-j] * pV1[-j]) >> overlapDividerBitsNorm;
}
dnorm += (double)lnorm;
if (lnorm > (long)maxnorm)
{
maxnorm = lnorm;
}
// Normalize result by dividing by sqrt(norm) - this step is easiest
// done using floating point operation
return (double)corr / sqrt((dnorm < 1e-9) ? 1.0 : dnorm);
}
void TDStretchMMX::clearCrossCorrState()
{
// Clear MMS state
_m_empty();
//_asm EMMS;
}
// MMX-optimized version of the function overlapStereo
void TDStretchMMX::overlapStereo(short *output, const short *input) const
{
const __m64 *pVinput, *pVMidBuf;
__m64 *pVdest;
__m64 mix1, mix2, adder, shifter;
int i;
pVinput = (const __m64*)input;
pVMidBuf = (const __m64*)pMidBuffer;
pVdest = (__m64*)output;
// mix1 = mixer values for 1st stereo sample
// mix1 = mixer values for 2nd stereo sample
// adder = adder for updating mixer values after each round
mix1 = _mm_set_pi16(0, overlapLength, 0, overlapLength);
adder = _mm_set_pi16(1, -1, 1, -1);
mix2 = _mm_add_pi16(mix1, adder);
adder = _mm_add_pi16(adder, adder);
// Overlaplength-division by shifter. "+1" is to account for "-1" deduced in
// overlapDividerBits calculation earlier.
shifter = _m_from_int(overlapDividerBitsPure + 1);
for (i = 0; i < overlapLength / 4; i ++)
{
__m64 temp1, temp2;
// load & shuffle data so that input & mixbuffer data samples are paired
temp1 = _mm_unpacklo_pi16(pVMidBuf[0], pVinput[0]); // = i0l m0l i0r m0r
temp2 = _mm_unpackhi_pi16(pVMidBuf[0], pVinput[0]); // = i1l m1l i1r m1r
// temp = (temp .* mix) >> shifter
temp1 = _mm_sra_pi32(_mm_madd_pi16(temp1, mix1), shifter);
temp2 = _mm_sra_pi32(_mm_madd_pi16(temp2, mix2), shifter);
pVdest[0] = _mm_packs_pi32(temp1, temp2); // pack 2*2*32bit => 4*16bit
// update mix += adder
mix1 = _mm_add_pi16(mix1, adder);
mix2 = _mm_add_pi16(mix2, adder);
// --- second round begins here ---
// load & shuffle data so that input & mixbuffer data samples are paired
temp1 = _mm_unpacklo_pi16(pVMidBuf[1], pVinput[1]); // = i2l m2l i2r m2r
temp2 = _mm_unpackhi_pi16(pVMidBuf[1], pVinput[1]); // = i3l m3l i3r m3r
// temp = (temp .* mix) >> shifter
temp1 = _mm_sra_pi32(_mm_madd_pi16(temp1, mix1), shifter);
temp2 = _mm_sra_pi32(_mm_madd_pi16(temp2, mix2), shifter);
pVdest[1] = _mm_packs_pi32(temp1, temp2); // pack 2*2*32bit => 4*16bit
// update mix += adder
mix1 = _mm_add_pi16(mix1, adder);
mix2 = _mm_add_pi16(mix2, adder);
pVinput += 2;
pVMidBuf += 2;
pVdest += 2;
}
_m_empty(); // clear MMS state
}
//////////////////////////////////////////////////////////////////////////////
//
// implementation of MMX optimized functions of class 'FIRFilter'
//
//////////////////////////////////////////////////////////////////////////////
#include "FIRFilter.h"
FIRFilterMMX::FIRFilterMMX() : FIRFilter()
{
filterCoeffsAlign = NULL;
filterCoeffsUnalign = NULL;
}
FIRFilterMMX::~FIRFilterMMX()
{
delete[] filterCoeffsUnalign;
}
// (overloaded) Calculates filter coefficients for MMX routine
void FIRFilterMMX::setCoefficients(const short *coeffs, uint newLength, uint uResultDivFactor)
{
uint i;
FIRFilter::setCoefficients(coeffs, newLength, uResultDivFactor);
// Ensure that filter coeffs array is aligned to 16-byte boundary
delete[] filterCoeffsUnalign;
filterCoeffsUnalign = new short[2 * newLength + 8];
filterCoeffsAlign = (short *)SOUNDTOUCH_ALIGN_POINTER_16(filterCoeffsUnalign);
// rearrange the filter coefficients for mmx routines
for (i = 0;i < length; i += 4)
{
filterCoeffsAlign[2 * i + 0] = coeffs[i + 0];
filterCoeffsAlign[2 * i + 1] = coeffs[i + 2];
filterCoeffsAlign[2 * i + 2] = coeffs[i + 0];
filterCoeffsAlign[2 * i + 3] = coeffs[i + 2];
filterCoeffsAlign[2 * i + 4] = coeffs[i + 1];
filterCoeffsAlign[2 * i + 5] = coeffs[i + 3];
filterCoeffsAlign[2 * i + 6] = coeffs[i + 1];
filterCoeffsAlign[2 * i + 7] = coeffs[i + 3];
}
}
// mmx-optimized version of the filter routine for stereo sound
uint FIRFilterMMX::evaluateFilterStereo(short *dest, const short *src, uint numSamples) const
{
// Create stack copies of the needed member variables for asm routines :
uint i, j;
__m64 *pVdest = (__m64*)dest;
if (length < 2) return 0;
for (i = 0; i < (numSamples - length) / 2; i ++)
{
__m64 accu1;
__m64 accu2;
const __m64 *pVsrc = (const __m64*)src;
const __m64 *pVfilter = (const __m64*)filterCoeffsAlign;
accu1 = accu2 = _mm_setzero_si64();
for (j = 0; j < lengthDiv8 * 2; j ++)
{
__m64 temp1, temp2;
temp1 = _mm_unpacklo_pi16(pVsrc[0], pVsrc[1]); // = l2 l0 r2 r0
temp2 = _mm_unpackhi_pi16(pVsrc[0], pVsrc[1]); // = l3 l1 r3 r1
accu1 = _mm_add_pi32(accu1, _mm_madd_pi16(temp1, pVfilter[0])); // += l2*f2+l0*f0 r2*f2+r0*f0
accu1 = _mm_add_pi32(accu1, _mm_madd_pi16(temp2, pVfilter[1])); // += l3*f3+l1*f1 r3*f3+r1*f1
temp1 = _mm_unpacklo_pi16(pVsrc[1], pVsrc[2]); // = l4 l2 r4 r2
accu2 = _mm_add_pi32(accu2, _mm_madd_pi16(temp2, pVfilter[0])); // += l3*f2+l1*f0 r3*f2+r1*f0
accu2 = _mm_add_pi32(accu2, _mm_madd_pi16(temp1, pVfilter[1])); // += l4*f3+l2*f1 r4*f3+r2*f1
// accu1 += l2*f2+l0*f0 r2*f2+r0*f0
// += l3*f3+l1*f1 r3*f3+r1*f1
// accu2 += l3*f2+l1*f0 r3*f2+r1*f0
// l4*f3+l2*f1 r4*f3+r2*f1
pVfilter += 2;
pVsrc += 2;
}
// accu >>= resultDivFactor
accu1 = _mm_srai_pi32(accu1, resultDivFactor);
accu2 = _mm_srai_pi32(accu2, resultDivFactor);
// pack 2*2*32bits => 4*16 bits
pVdest[0] = _mm_packs_pi32(accu1, accu2);
src += 4;
pVdest ++;
}
_m_empty(); // clear emms state
return (numSamples & 0xfffffffe) - length;
}
#else
// workaround to not complain about empty module
bool _dontcomplain_mmx_empty;
#endif // SOUNDTOUCH_ALLOW_MMX