--- /dev/null
+////////////////////////////////////////////////////////////////////////////////
+///
+/// gcc version of the 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.
+///
+/// This file is to be compiled on any platform with the GNU C compiler.
+/// Compiler. Please see 'mmx_win.cpp' for the x86 Windows version of this
+/// file.
+///
+/// Author : Copyright (c) Olli Parviainen
+/// Author e-mail : oparviai @ iki.fi
+/// SoundTouch WWW: http://www.iki.fi/oparviai/soundtouch
+///
+////////////////////////////////////////////////////////////////////////////////
+//
+// Last changed : $Date$
+// File revision : $Revision$
+//
+// $Id$
+//
+////////////////////////////////////////////////////////////////////////////////
+//
+// 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 <stdexcept>
+#include <string>
+#include "cpu_detect.h"
+
+#ifndef __GNUC__
+#error "wrong platform - this source code file is for the GNU C compiler."
+#endif
+
+using namespace std;
+using namespace soundtouch;
+
+
+#ifdef ALLOW_MMX
+// MMX routines available only with integer sample type
+
+//////////////////////////////////////////////////////////////////////////////
+//
+// implementation of MMX optimized functions of class 'TDStretch'
+//
+// NOTE: ebx in gcc 3.x is not preserved if -fPIC and -DPIC
+// gcc-3.4 correctly flags this error and wont let you continue.
+// gcc-2.95 preserves esi correctly
+//
+//////////////////////////////////////////////////////////////////////////////
+
+#include "TDStretch.h"
+#include <limits.h>
+
+// these are declared in 'TDStretch.cpp'
+extern int scanOffsets[4][24];
+
+// Calculates cross correlation of two buffers
+long TDStretchMMX::calcCrossCorrStereo(const short *pV1, const short *pV2) const
+{
+#ifdef __i386__
+ int corr;
+ uint local_overlapLength = overlapLength;
+ uint local_overlapDividerBits = overlapDividerBits;
+
+ asm volatile(
+ // Calculate cross-correlation between the tempOffset and tmpbid_buffer.
+
+ // Process 4 parallel batches of 2 * stereo samples each during one
+ // round to improve CPU-level parallellization.
+
+ // load address of sloped pV2 buffer to eax
+ // load address of mixing point of the sample data buffer to edi
+ // load counter to ecx = overlapLength / 8 - 1
+ // empty the mm0
+
+ // prepare to the first round by loading
+ // load mm1 = eax[0]
+ // load mm2 = eax[1];
+
+ "\n\tmovl %1, %%eax"
+ "\n\tmovl %2, %%edi"
+
+ "\n\tmovq (%%eax), %%mm1"
+ "\n\tmovl %3, %%ecx"
+
+ "\n\tmovq 8(%%eax), %%mm2"
+ "\n\tshr $3, %%ecx"
+
+ "\n\tpxor %%mm0, %%mm0"
+ "\n\tsub $1, %%ecx"
+
+ "\n\tmovd %4, %%mm5"
+
+ "\n1:"
+ // multiply-add mm1 = mm1 * edi[0]
+ // multiply-add mm2 = mm2 * edi[1]
+ //
+ // add mm2 += mm1
+ // mm2 >>= mm5 (=overlapDividerBits)
+ // add mm0 += mm2
+ //
+ // load mm3 = eax[2]
+ // multiply-add mm3 = mm3 * edi[2]
+ //
+ // load mm4 = eax[3]
+ // multiply-add mm4 = mm4 * edi[3]
+ //
+ // add mm3 += mm4
+ // mm3 >>= mm5 (=overlapDividerBits)
+ // add mm0 += mm3
+ //
+ // add eax += 4
+ // add edi += 4
+ // load mm1 = eax[0] (~eax[4])
+ // load mm2 = eax[1] (~eax[5])
+ //
+ // loop
+
+ "\n\tpmaddwd (%%edi), %%mm1" // qword ptr [edi]
+ "\n\tmovq 16(%%eax), %%mm3" // qword ptr [eax+16]
+
+ "\n\tpmaddwd 8(%%edi), %%mm2" // qword ptr [edi+8]
+ "\n\tmovq 24(%%eax), %%mm4" // qword ptr [eax+24]
+
+ "\n\tpmaddwd 16(%%edi), %%mm3" // qword ptr [edi+16]
+ "\n\tpaddd %%mm1, %%mm2"
+
+ "\n\tpmaddwd 24(%%edi), %%mm4" // qword ptr [edi+24]
+ "\n\tmovq 32(%%eax), %%mm1" // qword ptr [eax+32]
+
+ "\n\tpsrad %%mm5, %%mm2"
+ "\n\tadd $32, %%eax"
+
+ "\n\tpaddd %%mm4, %%mm3"
+ "\n\tpaddd %%mm2, %%mm0"
+
+ "\n\tmovq 8(%%eax), %%mm2" // qword ptr [eax+8]
+ "\n\tpsrad %%mm5, %%mm3"
+
+ "\n\tadd $32, %%edi"
+ "\n\tpaddd %%mm3, %%mm0"
+
+ "\n\tdec %%ecx"
+ "\n\tjnz 1b"
+
+ // Finalize the last partial loop:
+
+ "\n\tmovq 16(%%eax), %%mm3" // qword ptr [eax+16]
+ "\n\tpmaddwd (%%edi), %%mm1" // qword ptr [edi]
+
+ "\n\tmovq 24(%%eax), %%mm4" // qword ptr [eax+24]
+ "\n\tpmaddwd 8(%%edi), %%mm2" // qword ptr [edi+8]
+
+ "\n\tpmaddwd 16(%%edi), %%mm3" // qword ptr [edi+16]
+ "\n\tpaddd %%mm1, %%mm2"
+
+ "\n\tpmaddwd 24(%%edi), %%mm4" // qword ptr [edi+24]
+ "\n\tpsrad %%mm5, %%mm2"
+
+ "\n\tpaddd %%mm4, %%mm3"
+ "\n\tpaddd %%mm2, %%mm0"
+
+ "\n\tpsrad %%mm5, %%mm3"
+ "\n\tpaddd %%mm3, %%mm0"
+
+ // copy hi-dword of mm0 to lo-dword of mm1, then sum mmo+mm1
+ // and finally store the result into the variable "corr"
+
+ "\n\tmovq %%mm0, %%mm1"
+ "\n\tpsrlq $32, %%mm1"
+ "\n\tpaddd %%mm1, %%mm0"
+ "\n\tmovd %%mm0, %0"
+ : "=rm" (corr)
+ : "rim" (pV1), "rim" (pV2), "rim" (local_overlapLength),
+ "rim" (local_overlapDividerBits)
+ : "%ecx", "%eax", "%edi"
+ );
+ return corr;
+
+ // Note: Warning about the missing EMMS instruction is harmless
+ // as it'll be called elsewhere.
+#else
+ throw runtime_error("MMX not supported");
+#endif
+}
+
+void TDStretchMMX::clearCrossCorrState()
+{
+#ifdef __i386__
+ asm volatile("EMMS");
+#endif
+}
+
+// MMX-optimized version of the function overlapStereo
+void TDStretchMMX::overlapStereo(short *output, const short *input) const
+{
+#ifdef __i386__
+ short *local_midBuffer = pMidBuffer;
+ uint local_overlapLength = overlapLength;
+ uint local_overlapDividerBits = overlapDividerBits;
+
+ asm volatile(
+ "\n\t"
+ // load sliding mixing value counter to mm6 and mm7
+ // load counter value to ecx = overlapLength / 4
+ // load divider-shifter value to esi
+ // load mixing value adder to mm5
+ // load address of midBuffer to eax
+ // load address of inputBuffer added with ovlOffset to edi
+ // load address of end of the outputBuffer to edx
+ //
+ // We need to preserve esi, since gcc uses it for the
+ // stack frame.
+
+ "movl %0, %%eax\n\t" // ecx = 0x0000 OVL_
+ "movl $0x0002fffe, %%edi\n\t" // ecx = 0x0002 fffe
+
+ "movl %1, %%esi\n\t"
+ "movd %%eax, %%mm6\n\t" // mm6 = 0x0000 0000 0000 OVL_
+
+ "movl %%eax, %%ecx\n\t"
+ "sub $1, %%eax\n\t"
+
+ "punpckldq %%mm6, %%mm6\n\t" // mm6 = 0x0000 OVL_ 0000 OVL_
+
+ "or $0x00010000, %%eax\n\t" // eax = 0x0001 overlapLength-1
+
+ "movd %%edi, %%mm5\n\t" // mm5 = 0x0000 0000 0002 fffe
+ "movd %%eax, %%mm7\n\t" // mm7 = 0x0000 0000 0001 01ff
+
+ "movl %3, %%edi\n\t"
+
+ "movl %4, %%eax\n\t" // dword ptr local_midBuffer
+ "punpckldq %%mm5, %%mm5\n\t" // mm5 = 0x0002 fffe 0002 fffe
+
+ "shr $2, %%ecx\n\t" // ecx = overlapLength / 2
+ "punpckldq %%mm7, %%mm7\n\t" // mm7 = 0x0001 01ff 0001 01ff
+
+ "movl %2, %%edx\n"
+
+ "2:\n\t"
+ // Process two parallel batches of 2+2 stereo samples during each round
+ // to improve CPU-level parallellization.
+ //
+ // Load [eax] into mm0 and mm1
+ // Load [edi] into mm3
+ // unpack words of mm0, mm1 and mm3 into mm0 and mm1
+ // multiply-add mm0*mm6 and mm1*mm7, store results into mm0 and mm1
+ // divide mm0 and mm1 by 512 (=right-shift by overlapDividerBits)
+ // pack the result into mm0 and store into [edx]
+ //
+ // Load [eax+8] into mm2 and mm3
+ // Load [edi+8] into mm4
+ // unpack words of mm2, mm3 and mm4 into mm2 and mm3
+ // multiply-add mm2*mm6 and mm3*mm7, store results into mm2 and mm3
+ // divide mm2 and mm3 by 512 (=right-shift by overlapDividerBits)
+ // pack the result into mm2 and store into [edx+8]
+
+
+ "movq (%%eax), %%mm0\n\t" // mm0 = m1l m1r m0l m0r
+ "add $16, %%edx\n\t"
+
+ "movq (%%edi), %%mm3\n\t" // mm3 = i1l i1r i0l i0r
+ "movq %%mm0, %%mm1\n\t" // mm1 = m1l m1r m0l m0r
+
+ "movq 8(%%eax), %%mm2\n\t" // mm2 = m3l m3r m2l m2r
+ "punpcklwd %%mm3, %%mm0\n\t" // mm0 = i0l m0l i0r m0r
+
+ "movq 8(%%edi), %%mm4\n\t" // mm4 = i3l i3r i2l i2r
+ "punpckhwd %%mm3, %%mm1\n\t" // mm1 = i1l m1l i1r m1r
+
+ "movq %%mm2, %%mm3\n\t" // mm3 = m3l m3r m2l m2r
+ "punpcklwd %%mm4, %%mm2\n\t" // mm2 = i2l m2l i2r m2r
+
+ "pmaddwd %%mm6, %%mm0\n\t" // mm0 = i0l*m63+m0l*m62 i0r*m61+m0r*m60
+ "punpckhwd %%mm4, %%mm3\n\t" // mm3 = i3l m3l i3r m3r
+
+ "movd %%esi, %%mm4\n\t" // mm4 = overlapDividerBits
+
+ "pmaddwd %%mm7, %%mm1\n\t" // mm1 = i1l*m73+m1l*m72 i1r*m71+m1r*m70
+ "paddw %%mm5, %%mm6\n\t"
+
+ "paddw %%mm5, %%mm7\n\t"
+ "psrad %%mm4, %%mm0\n\t" // mmo >>= overlapDividerBits
+
+ "pmaddwd %%mm6, %%mm2\n\t" // mm2 = i2l*m63+m2l*m62 i2r*m61+m2r*m60
+ "psrad %%mm4, %%mm1\n\t" // mm1 >>= overlapDividerBits
+
+ "pmaddwd %%mm7, %%mm3\n\t" // mm3 = i3l*m73+m3l*m72 i3r*m71+m3r*m70
+ "psrad %%mm4, %%mm2\n\t" // mm2 >>= overlapDividerBits
+
+ "packssdw %%mm1, %%mm0\n\t" // mm0 = mm1h mm1l mm0h mm0l
+ "psrad %%mm4, %%mm3\n\t" // mm3 >>= overlapDividerBits
+
+ "add $16, %%eax\n\t"
+ "paddw %%mm5, %%mm6\n\t"
+
+ "packssdw %%mm3, %%mm2\n\t" // mm2 = mm2h mm2l mm3h mm3l
+ "paddw %%mm5, %%mm7\n\t"
+
+ "movq %%mm0, -16(%%edx)\n\t"
+ "add $16, %%edi\n\t"
+
+ "movq %%mm2, -8(%%edx)\n\t"
+ "dec %%ecx\n\t"
+
+ "jnz 2b\n\t"
+
+ "emms\n\t"
+
+ :
+ : "rim" (local_overlapLength),
+ "rim" (local_overlapDividerBits),
+ "rim" (output),
+ "rim" (input),
+ "rim" (local_midBuffer)
+ /* input */
+ : "%edi", "%ecx", "%edx", "%eax", "%esi" /* regs */
+ );
+#else
+ throw runtime_error("MMX not supported");
+#endif
+}
+
+
+//////////////////////////////////////////////////////////////////////////////
+//
+// implementation of MMX optimized functions of class 'FIRFilter'
+//
+//////////////////////////////////////////////////////////////////////////////
+
+#include "FIRFilter.h"
+
+FIRFilterMMX::FIRFilterMMX() : FIRFilter()
+{
+ filterCoeffsUnalign = NULL;
+}
+
+
+FIRFilterMMX::~FIRFilterMMX()
+{
+ delete[] filterCoeffsUnalign;
+}
+
+
+#if 1
+// (overloaded) Calculates filter coefficients for MMX routine
+void FIRFilterMMX::setCoefficients(const short *coeffs, uint newLength, uint uResultDivFactor)
+{
+#ifdef __i386__
+ 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 *)(((uint)filterCoeffsUnalign + 15) & -16);
+
+ // 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];
+ }
+#else
+ throw runtime_error("MMX not supported");
+#endif
+}
+
+
+
+// mmx-optimized version of the filter routine for stereo sound
+uint FIRFilterMMX::evaluateFilterStereo(short *dest, const short *src, const uint numSamples) const
+{
+#ifdef __i386__
+ // Create stack copies of the needed member variables for asm routines :
+ uint local_length = length;
+ uint local_lengthDiv8 = lengthDiv8;
+ uint local_resultDivider = resultDivFactor;
+ short *local_filterCoeffs = (short*)filterCoeffsAlign;
+ short *local_src = (short *)src;
+
+ asm volatile(
+ "\n\t"
+ // Load (num_samples-aa_filter_length)/2 to edi as a i
+ // Load a pointer to samples to esi
+ // Load a pointer to destination to edx
+
+ "movl %0, %%edi\n\t"
+ "subl %2, %%edi\n\t"
+ "movl %3, %%edx\n\t"
+ "sar $1, %%edi\n"
+
+ // Load filter length/8 to ecx
+ // Load pointer to samples from esi to ebx
+ // Load counter from edi to ecx
+ // Load [ebx] to mm3
+ // Load pointer to filter coefficients to eax
+ "3:\n\t"
+ "movl %1, %%esi\n\t"
+ "pxor %%mm0, %%mm0\n\t"
+
+ "movl %4, %%ecx\n\t"
+ "pxor %%mm7, %%mm7\n\t"
+
+ "movq (%%esi), %%mm1\n\t" // mm1 = l1 r1 l0 r0
+ "movl %5, %%eax\n"
+ "4:\n\t"
+
+ "movq 8(%%esi), %%mm2\n\t" // mm2 = l3 r3 l2 r2
+ "movq %%mm1, %%mm4\n\t" // mm4 = l1 r1 l0 r0
+
+ "movq 16(%%esi), %%mm3\n\t" // mm3 = l5 r5 l4 r4
+ "punpckhwd %%mm2, %%mm1\n\t" // mm1 = l3 l1 r3 r1
+
+ "movq %%mm2, %%mm6\n\t" // mm6 = l3 r3 l2 r2
+ "punpcklwd %%mm2, %%mm4\n\t" // mm4 = l2 l0 r2 r0
+
+ "movq (%%eax), %%mm2\n\t" // mm2 = f2 f0 f2 f0
+ "movq %%mm1, %%mm5\n\t" // mm5 = l3 l1 r3 r1
+
+ "punpcklwd %%mm3, %%mm6\n\t" // mm6 = l4 l2 r4 r2
+ "pmaddwd %%mm2, %%mm4\n\t" // mm4 = l2*f2+l0*f0 r2*f2+r0*f0
+
+ "pmaddwd %%mm2, %%mm5\n\t" // mm5 = l3*f2+l1*f0 r3*f2+l1*f0
+ "movq 8(%%eax), %%mm2\n\t" // mm2 = f3 f1 f3 f1
+
+ "paddd %%mm4, %%mm0\n\t" // mm0 += s02*f02
+ "movq %%mm3, %%mm4\n\t" // mm4 = l1 r1 l0 r0
+
+ "pmaddwd %%mm2, %%mm1\n\t" // mm1 = l3*f3+l1*f1 r3*f3+l1*f1
+ "paddd %%mm5, %%mm7\n\t" // mm7 += s13*f02
+
+ "pmaddwd %%mm2, %%mm6\n\t" // mm6 = l4*f3+l2*f1 r4*f3+f4*f1
+ "movq 24(%%esi), %%mm2\n\t" // mm2 = l3 r3 l2 r2
+
+ "paddd %%mm1, %%mm0\n\t" // mm0 += s31*f31
+ "movq 32(%%esi), %%mm1\n\t" // mm1 = l5 r5 l4 r4
+
+ "paddd %%mm6, %%mm7\n\t" // mm7 += s42*f31
+ "punpckhwd %%mm2, %%mm3\n\t" // mm3 = l3 l1 r3 r1
+
+ "movq %%mm2, %%mm6\n\t" // mm6 = l3 r3 l2 r2
+ "punpcklwd %%mm2, %%mm4\n\t" // mm4 = l2 l0 r2 r0
+
+ "movq 16(%%eax), %%mm2\n\t" // mm2 = f2 f0 f2 f0
+ "movq %%mm3, %%mm5\n\t" // mm5 = l3 l1 r3 r1
+
+ "punpcklwd %%mm1, %%mm6\n\t" // mm6 = l4 l2 r4 r2
+ "add $32, %%eax\n\t"
+
+ "pmaddwd %%mm2, %%mm4\n\t" // mm4 = l2*f2+l0*f0 r2*f2+r0*f0
+ "add $32, %%esi\n\t"
+
+ "pmaddwd %%mm2, %%mm5\n\t" // mm5 = l3*f2+l1*f0 r3*f2+l1*f0
+ "movq -8(%%eax), %%mm2\n\t" // mm2 = f3 f1 f3 f1
+
+ "paddd %%mm4, %%mm0\n\t" // mm0 += s02*f02
+ "pmaddwd %%mm2, %%mm3\n\t" // mm3 = l3*f3+l1*f1 r3*f3+l1*f1
+
+ "paddd %%mm5, %%mm7\n\t" // mm7 += s13*f02
+ "pmaddwd %%mm2, %%mm6\n\t" // mm6 = l4*f3+l2*f1 r4*f3+f4*f1
+
+ "paddd %%mm3, %%mm0\n\t" // mm0 += s31*f31
+ "paddd %%mm6, %%mm7\n\t" // mm7 += s42*f31
+
+ "dec %%ecx\n\t"
+ "jnz 4b\n\t"
+
+ // Divide mm0 and mm7 by 8192 (= right-shift by 13),
+ // pack and store to [edx]
+ "movd %6, %%mm4\n\t"
+
+ "psrad %%mm4, %%mm0\n\t" // divide the result
+
+ "add $8, %%edx\n\t"
+ "psrad %%mm4, %%mm7\n\t" // divide the result
+
+ "add $8, %1\n\t"
+ "packssdw %%mm7, %%mm0\n\t"
+
+ "movq %%mm0, -8(%%edx)\n\t"
+ "dec %%edi\n\t"
+
+ "jnz 3b\n\t"
+
+ "emms\n\t"
+
+ :
+ : "rim" (numSamples),
+ "rim" (local_src),
+ "rim" (local_length),
+ "rim" (dest),
+ "rim" (local_lengthDiv8),
+ "rim" (local_filterCoeffs),
+ "rim" (local_resultDivider) /* input */
+ : "%eax", "%ecx", "%edx", "%edi", "%esi" /* regs */
+ );
+ return (numSamples & 0xfffffffe) - local_length;
+#else
+ throw runtime_error("MMX not supported");
+ return 0;
+#endif
+}
+#endif
+
+#endif // ALLOW_MMX
projects / ardour.git / blobdiff