Intravision/reactos
1
0
Fork 0
mirror of https://github.com/reactos/reactos.git synced 2024-11-10 08:43:28 +00:00
Code Issues Projects Releases Packages Wiki Activity
reactos/lib/3rdparty/libmpg123/synth_altivec.c

1058 lines
29 KiB
C
Raw Normal View History

Sync winemp3.acm with Wine HEAD. This one uses libmpg123 which was added in Version 1.12.3. Thx to Timo Kreuzer for telling me the magics of the crt="dll" switch. svn path=/trunk/; revision=48279
2010-07-26 15:04:32 +00:00
/*
decode.c: decoding samples...
copyright 1995-2009 by the mpg123 project - free software under the terms of the LGPL 2.1
see COPYING and AUTHORS files in distribution or http://mpg123.org
initially written by Michael Hipp
altivec optimization by tmkk
*/
#include "mpg123lib_intern.h"
#ifndef __APPLE__
#include <altivec.h>
#endif
/* A macro for normal synth functions */
#define SYNTH_ALTIVEC(B0STEP) \
v1 = vec_ld(0,window); \
v2 = vec_ld(16,window); \
v3 = vec_ld(32,window); \
v4 = vec_ld(48,window); \
v5 = vec_ld(64,window); \
v1 = vec_perm(v1,v2,vperm1); \
v6 = vec_ld(0,b0); \
v2 = vec_perm(v2,v3,vperm1); \
v7 = vec_ld(16,b0); \
v3 = vec_perm(v3,v4,vperm1); \
v8 = vec_ld(32,b0); \
v4 = vec_perm(v4,v5,vperm1); \
v9 = vec_ld(48,b0); \
\
vsum = vec_madd(v1,v6,vzero); \
vsum = vec_madd(v2,v7,vsum); \
vsum = vec_madd(v3,v8,vsum); \
vsum = vec_madd(v4,v9,vsum); \
\
window += 32; \
b0 += B0STEP; \
\
v1 = vec_ld(0,window); \
v2 = vec_ld(16,window); \
v3 = vec_ld(32,window); \
v4 = vec_ld(48,window); \
v5 = vec_ld(64,window); \
v1 = vec_perm(v1,v2,vperm1); \
v6 = vec_ld(0,b0); \
v2 = vec_perm(v2,v3,vperm1); \
v7 = vec_ld(16,b0); \
v3 = vec_perm(v3,v4,vperm1); \
v8 = vec_ld(32,b0); \
v4 = vec_perm(v4,v5,vperm1); \
v9 = vec_ld(48,b0); \
\
vsum2 = vec_madd(v1,v6,vzero); \
vsum2 = vec_madd(v2,v7,vsum2); \
vsum2 = vec_madd(v3,v8,vsum2); \
vsum2 = vec_madd(v4,v9,vsum2); \
\
window += 32; \
b0 += B0STEP; \
\
v1 = vec_ld(0,window); \
v2 = vec_ld(16,window); \
v3 = vec_ld(32,window); \
v4 = vec_ld(48,window); \
v5 = vec_ld(64,window); \
v1 = vec_perm(v1,v2,vperm1); \
v6 = vec_ld(0,b0); \
v2 = vec_perm(v2,v3,vperm1); \
v7 = vec_ld(16,b0); \
v3 = vec_perm(v3,v4,vperm1); \
v8 = vec_ld(32,b0); \
v4 = vec_perm(v4,v5,vperm1); \
v9 = vec_ld(48,b0); \
\
vsum3 = vec_madd(v1,v6,vzero); \
vsum3 = vec_madd(v2,v7,vsum3); \
vsum3 = vec_madd(v3,v8,vsum3); \
vsum3 = vec_madd(v4,v9,vsum3); \
\
window += 32; \
b0 += B0STEP; \
\
v1 = vec_ld(0,window); \
v2 = vec_ld(16,window); \
v3 = vec_ld(32,window); \
v4 = vec_ld(48,window); \
v5 = vec_ld(64,window); \
v1 = vec_perm(v1,v2,vperm1); \
v6 = vec_ld(0,b0); \
v2 = vec_perm(v2,v3,vperm1); \
v7 = vec_ld(16,b0); \
v3 = vec_perm(v3,v4,vperm1); \
v8 = vec_ld(32,b0); \
v4 = vec_perm(v4,v5,vperm1); \
v9 = vec_ld(48,b0); \
\
vsum4 = vec_madd(v1,v6,vzero); \
vsum4 = vec_madd(v2,v7,vsum4); \
vsum4 = vec_madd(v3,v8,vsum4); \
vsum4 = vec_madd(v4,v9,vsum4); \
\
window += 32; \
b0 += B0STEP; \
\
v1 = vec_mergeh(vsum,vsum3); \
v2 = vec_mergeh(vsum2,vsum4); \
v3 = vec_mergel(vsum,vsum3); \
v4 = vec_mergel(vsum2,vsum4); \
v5 = vec_mergeh(v1,v2); \
v6 = vec_mergel(v1,v2); \
v7 = vec_mergeh(v3,v4); \
v8 = vec_mergel(v3,v4);
/* A macro for stereo synth functions */
#define SYNTH_STEREO_ALTIVEC(B0STEP) \
v1 = vec_ld(0,window); \
v2 = vec_ld(16,window); \
v3 = vec_ld(32,window); \
v4 = vec_ld(48,window); \
v5 = vec_ld(64,window); \
v1 = vec_perm(v1,v2,vperm1); \
v6 = vec_ld(0,b0l); \
v10 = vec_ld(0,b0r); \
v2 = vec_perm(v2,v3,vperm1); \
v7 = vec_ld(16,b0l); \
v11 = vec_ld(16,b0r); \
v3 = vec_perm(v3,v4,vperm1); \
v8 = vec_ld(32,b0l); \
v12 = vec_ld(32,b0r); \
v4 = vec_perm(v4,v5,vperm1); \
v9 = vec_ld(48,b0l); \
v13 = vec_ld(48,b0r); \
\
vsum = vec_madd(v1,v6,vzero); \
vsum5 = vec_madd(v1,v10,vzero); \
vsum = vec_madd(v2,v7,vsum); \
vsum5 = vec_madd(v2,v11,vsum5); \
vsum = vec_madd(v3,v8,vsum); \
vsum5 = vec_madd(v3,v12,vsum5); \
vsum = vec_madd(v4,v9,vsum); \
vsum5 = vec_madd(v4,v13,vsum5); \
\
window += 32; \
b0l += B0STEP; \
b0r += B0STEP; \
\
v1 = vec_ld(0,window); \
v2 = vec_ld(16,window); \
v3 = vec_ld(32,window); \
v4 = vec_ld(48,window); \
v5 = vec_ld(64,window); \
v1 = vec_perm(v1,v2,vperm1); \
v6 = vec_ld(0,b0l); \
v10 = vec_ld(0,b0r); \
v2 = vec_perm(v2,v3,vperm1); \
v7 = vec_ld(16,b0l); \
v11 = vec_ld(16,b0r); \
v3 = vec_perm(v3,v4,vperm1); \
v8 = vec_ld(32,b0l); \
v12 = vec_ld(32,b0r); \
v4 = vec_perm(v4,v5,vperm1); \
v9 = vec_ld(48,b0l); \
v13 = vec_ld(48,b0r); \
\
vsum2 = vec_madd(v1,v6,vzero); \
vsum6 = vec_madd(v1,v10,vzero); \
vsum2 = vec_madd(v2,v7,vsum2); \
vsum6 = vec_madd(v2,v11,vsum6); \
vsum2 = vec_madd(v3,v8,vsum2); \
vsum6 = vec_madd(v3,v12,vsum6); \
vsum2 = vec_madd(v4,v9,vsum2); \
vsum6 = vec_madd(v4,v13,vsum6); \
\
window += 32; \
b0l += B0STEP; \
b0r += B0STEP; \
\
v1 = vec_ld(0,window); \
v2 = vec_ld(16,window); \
v3 = vec_ld(32,window); \
v4 = vec_ld(48,window); \
v5 = vec_ld(64,window); \
v1 = vec_perm(v1,v2,vperm1); \
v6 = vec_ld(0,b0l); \
v10 = vec_ld(0,b0r); \
v2 = vec_perm(v2,v3,vperm1); \
v7 = vec_ld(16,b0l); \
v11 = vec_ld(16,b0r); \
v3 = vec_perm(v3,v4,vperm1); \
v8 = vec_ld(32,b0l); \
v12 = vec_ld(32,b0r); \
v4 = vec_perm(v4,v5,vperm1); \
v9 = vec_ld(48,b0l); \
v13 = vec_ld(48,b0r); \
\
vsum3 = vec_madd(v1,v6,vzero); \
vsum7 = vec_madd(v1,v10,vzero); \
vsum3 = vec_madd(v2,v7,vsum3); \
vsum7 = vec_madd(v2,v11,vsum7); \
vsum3 = vec_madd(v3,v8,vsum3); \
vsum7 = vec_madd(v3,v12,vsum7); \
vsum3 = vec_madd(v4,v9,vsum3); \
vsum7 = vec_madd(v4,v13,vsum7); \
\
window += 32; \
b0l += B0STEP; \
b0r += B0STEP; \
\
v1 = vec_ld(0,window); \
v2 = vec_ld(16,window); \
v3 = vec_ld(32,window); \
v4 = vec_ld(48,window); \
v5 = vec_ld(64,window); \
v1 = vec_perm(v1,v2,vperm1); \
v6 = vec_ld(0,b0l); \
v10 = vec_ld(0,b0r); \
v2 = vec_perm(v2,v3,vperm1); \
v7 = vec_ld(16,b0l); \
v11 = vec_ld(16,b0r); \
v3 = vec_perm(v3,v4,vperm1); \
v8 = vec_ld(32,b0l); \
v12 = vec_ld(32,b0r); \
v4 = vec_perm(v4,v5,vperm1); \
v9 = vec_ld(48,b0l); \
v13 = vec_ld(48,b0r); \
\
vsum4 = vec_madd(v1,v6,vzero); \
vsum8 = vec_madd(v1,v10,vzero); \
vsum4 = vec_madd(v2,v7,vsum4); \
vsum8 = vec_madd(v2,v11,vsum8); \
vsum4 = vec_madd(v3,v8,vsum4); \
vsum8 = vec_madd(v3,v12,vsum8); \
vsum4 = vec_madd(v4,v9,vsum4); \
vsum8 = vec_madd(v4,v13,vsum8); \
\
window += 32; \
b0l += B0STEP; \
b0r += B0STEP; \
\
v1 = vec_mergeh(vsum,vsum3); \
v5 = vec_mergeh(vsum5,vsum7); \
v2 = vec_mergeh(vsum2,vsum4); \
v6 = vec_mergeh(vsum6,vsum8); \
v3 = vec_mergel(vsum,vsum3); \
v7 = vec_mergel(vsum5,vsum7); \
v4 = vec_mergel(vsum2,vsum4); \
v8 = vec_mergel(vsum6,vsum8); \
vsum = vec_mergeh(v1,v2); \
vsum5 = vec_mergeh(v5,v6); \
vsum2 = vec_mergel(v1,v2); \
vsum6 = vec_mergel(v5,v6); \
vsum3 = vec_mergeh(v3,v4); \
vsum7 = vec_mergeh(v7,v8); \
vsum4 = vec_mergel(v3,v4); \
vsum8 = vec_mergel(v7,v8);
int synth_1to1_altivec(real *bandPtr,int channel,mpg123_handle *fr, int final)
{
short *samples = (short *) (fr->buffer.data+fr->buffer.fill);
real *b0, **buf;
int clip;
int bo1;
if(fr->have_eq_settings) do_equalizer(bandPtr,channel,fr->equalizer);
if(!channel)
{
fr->bo--;
fr->bo &= 0xf;
buf = fr->real_buffs[0];
}
else
{
samples++;
buf = fr->real_buffs[1];
}
if(fr->bo & 0x1)
{
b0 = buf[0];
bo1 = fr->bo;
dct64_altivec(buf[1]+((fr->bo+1)&0xf),buf[0]+fr->bo,bandPtr);
}
else
{
b0 = buf[1];
bo1 = fr->bo+1;
dct64_altivec(buf[0]+fr->bo,buf[1]+fr->bo+1,bandPtr);
}
{
register int j;
real *window = fr->decwin + 16 - bo1;
ALIGNED(16) int clip_tmp[4];
vector float v1,v2,v3,v4,v5,v6,v7,v8,v9;
vector unsigned char vperm1,vperm2,vperm3,vperm4;
vector float vsum,vsum2,vsum3,vsum4,vmin,vmax,vzero;
vector signed int vclip;
vector signed short vsample1,vsample2;
vector unsigned int vshift;
vclip = vec_xor(vclip,vclip);
vzero = vec_xor(vzero,vzero);
vshift = vec_splat_u32(-1); /* 31 */
#ifdef __APPLE__
vmax = (vector float)(32767.0f);
vmin = (vector float)(-32768.0f);
vperm4 = (vector unsigned char)(0,1,18,19,2,3,22,23,4,5,26,27,6,7,30,31);
#else
vmax = (vector float){32767.0f,32767.0f,32767.0f,32767.0f};
vmin = (vector float){-32768.0f,-32768.0f,-32768.0f,-32768.0f};
vperm4 = (vector unsigned char){0,1,18,19,2,3,22,23,4,5,26,27,6,7,30,31};
#endif
vperm1 = vec_lvsl(0,window);
vperm2 = vec_lvsl(0,samples);
vperm3 = vec_lvsr(0,samples);
for (j=4;j;j--)
{
SYNTH_ALTIVEC(16);
vsum = vec_sub(v5,v6);
v9 = vec_sub(v7,v8);
vsum = vec_add(vsum,v9);
v3 = vec_round(vsum);
v3 = (vector float)vec_cts(v3,0);
v1 = (vector float)vec_cmpgt(vsum,vmax);
v2 = (vector float)vec_cmplt(vsum,vmin);
vsample1 = vec_ld(0,samples);
vsample2 = vec_ld(15,samples);
v3 = (vector float)vec_packs((vector signed int)v3,(vector signed int)v3);
v4 = (vector float)vec_perm(vsample1,vsample2,vperm2);
v5 = (vector float)vec_perm(v3,v4,vperm4);
v6 = (vector float)vec_perm(vsample2,vsample1,vperm2);
v7 = (vector float)vec_perm(v5,v6,vperm3);
v8 = (vector float)vec_perm(v6,v5,vperm3);
vec_st((vector signed short)v7,15,samples);
vec_st((vector signed short)v8,0,samples);
samples += 8;
v1 = (vector float)vec_sr((vector unsigned int)v1, vshift);
v2 = (vector float)vec_sr((vector unsigned int)v2, vshift);
v1 = (vector float)vec_add((vector unsigned int)v1,(vector unsigned int)v2);
vclip = vec_sums((vector signed int)v1,vclip);
}
for (j=4;j;j--)
{
SYNTH_ALTIVEC(-16);
vsum = vec_add(v5,v6);
v9 = vec_add(v7,v8);
vsum = vec_add(vsum,v9);
v3 = vec_round(vsum);
v3 = (vector float)vec_cts(v3,0);
v1 = (vector float)vec_cmpgt(vsum,vmax);
v2 = (vector float)vec_cmplt(vsum,vmin);
vsample1 = vec_ld(0,samples);
vsample2 = vec_ld(15,samples);
v3 = (vector float)vec_packs((vector signed int)v3,(vector signed int)v3);
v4 = (vector float)vec_perm(vsample1,vsample2,vperm2);
v5 = (vector float)vec_perm(v3,v4,vperm4);
v6 = (vector float)vec_perm(vsample2,vsample1,vperm2);
v7 = (vector float)vec_perm(v5,v6,vperm3);
v8 = (vector float)vec_perm(v6,v5,vperm3);
vec_st((vector signed short)v7,15,samples);
vec_st((vector signed short)v8,0,samples);
samples += 8;
v1 = (vector float)vec_sr((vector unsigned int)v1, vshift);
v2 = (vector float)vec_sr((vector unsigned int)v2, vshift);
v1 = (vector float)vec_add((vector unsigned int)v1,(vector unsigned int)v2);
vclip = vec_sums((vector signed int)v1,vclip);
}
vec_st(vclip,0,clip_tmp);
clip = clip_tmp[3];
}
if(final) fr->buffer.fill += 128;
return clip;
}
int synth_1to1_stereo_altivec(real *bandPtr_l, real *bandPtr_r, mpg123_handle *fr)
{
short *samples = (short *) (fr->buffer.data+fr->buffer.fill);
real *b0l, *b0r, **bufl, **bufr;
int clip;
int bo1;
if(fr->have_eq_settings)
{
do_equalizer(bandPtr_l,0,fr->equalizer);
do_equalizer(bandPtr_r,1,fr->equalizer);
}
fr->bo--;
fr->bo &= 0xf;
bufl = fr->real_buffs[0];
bufr = fr->real_buffs[1];
if(fr->bo & 0x1)
{
b0l = bufl[0];
b0r = bufr[0];
bo1 = fr->bo;
dct64_altivec(bufl[1]+((fr->bo+1)&0xf),bufl[0]+fr->bo,bandPtr_l);
dct64_altivec(bufr[1]+((fr->bo+1)&0xf),bufr[0]+fr->bo,bandPtr_r);
}
else
{
b0l = bufl[1];
b0r = bufr[1];
bo1 = fr->bo+1;
dct64_altivec(bufl[0]+fr->bo,bufl[1]+fr->bo+1,bandPtr_l);
dct64_altivec(bufr[0]+fr->bo,bufr[1]+fr->bo+1,bandPtr_r);
}
{
register int j;
real *window = fr->decwin + 16 - bo1;
ALIGNED(16) int clip_tmp[4];
vector float v1,v2,v3,v4,v5,v6,v7,v8,v9,v10,v11,v12,v13;
vector unsigned char vperm1,vperm2;
vector float vsum,vsum2,vsum3,vsum4,vsum5,vsum6,vsum7,vsum8,vmin,vmax,vzero;
vector signed int vclip;
vector unsigned int vshift;
vector signed short vprev;
vclip = vec_xor(vclip,vclip);
vzero = vec_xor(vzero,vzero);
vshift = vec_splat_u32(-1); /* 31 */
#ifdef __APPLE__
vmax = (vector float)(32767.0f);
vmin = (vector float)(-32768.0f);
#else
vmax = (vector float){32767.0f,32767.0f,32767.0f,32767.0f};
vmin = (vector float){-32768.0f,-32768.0f,-32768.0f,-32768.0f};
#endif
vperm1 = vec_lvsl(0,window);
vperm2 = vec_lvsr(0,samples);
vprev = vec_perm(vec_ld(0,samples),vec_ld(0,samples),vec_lvsl(0,samples));
for (j=4;j;j--)
{
SYNTH_STEREO_ALTIVEC(16);
vsum = vec_sub(vsum,vsum2);
vsum2 = vec_sub(vsum5,vsum6);
vsum3 = vec_sub(vsum3,vsum4);
vsum4 = vec_sub(vsum7,vsum8);
vsum = vec_add(vsum,vsum3);
vsum2 = vec_add(vsum2,vsum4);
v1 = vec_round(vsum);
v2 = vec_round(vsum2);
v1 = (vector float)vec_cts(v1,0);
v2 = (vector float)vec_cts(v2,0);
v3 = vec_mergeh(v1, v2);
v4 = vec_mergel(v1, v2);
v5 = (vector float)vec_packs((vector signed int)v3,(vector signed int)v4);
v6 = (vector float)vec_perm(vprev,(vector signed short)v5,vperm2);
vprev = (vector signed short)v5;
v1 = (vector float)vec_cmpgt(vsum,vmax);
v2 = (vector float)vec_cmplt(vsum,vmin);
v3 = (vector float)vec_cmpgt(vsum2,vmax);
v4 = (vector float)vec_cmplt(vsum2,vmin);
vec_st((vector signed short)v6,0,samples);
samples += 8;
v1 = (vector float)vec_sr((vector unsigned int)v1, vshift);
v2 = (vector float)vec_sr((vector unsigned int)v2, vshift);
v3 = (vector float)vec_sr((vector unsigned int)v3, vshift);
v4 = (vector float)vec_sr((vector unsigned int)v4, vshift);
v1 = (vector float)vec_add((vector unsigned int)v1,(vector unsigned int)v2);
v2 = (vector float)vec_add((vector unsigned int)v3,(vector unsigned int)v4);
vclip = vec_sums((vector signed int)v1,vclip);
vclip = vec_sums((vector signed int)v2,vclip);
}
for (j=4;j;j--)
{
SYNTH_STEREO_ALTIVEC(-16);
vsum = vec_add(vsum,vsum2);
vsum2 = vec_add(vsum5,vsum6);
vsum3 = vec_add(vsum3,vsum4);
vsum4 = vec_add(vsum7,vsum8);
vsum = vec_add(vsum,vsum3);
vsum2 = vec_add(vsum2,vsum4);
v1 = vec_round(vsum);
v2 = vec_round(vsum2);
v1 = (vector float)vec_cts(v1,0);
v2 = (vector float)vec_cts(v2,0);
v3 = vec_mergeh(v1, v2);
v4 = vec_mergel(v1, v2);
v5 = (vector float)vec_packs((vector signed int)v3,(vector signed int)v4);
v6 = (vector float)vec_perm(vprev,(vector signed short)v5,vperm2);
vprev = (vector signed short)v5;
v1 = (vector float)vec_cmpgt(vsum,vmax);
v2 = (vector float)vec_cmplt(vsum,vmin);
v3 = (vector float)vec_cmpgt(vsum2,vmax);
v4 = (vector float)vec_cmplt(vsum2,vmin);
vec_st((vector signed short)v6,0,samples);
samples += 8;
v1 = (vector float)vec_sr((vector unsigned int)v1, vshift);
v2 = (vector float)vec_sr((vector unsigned int)v2, vshift);
v3 = (vector float)vec_sr((vector unsigned int)v3, vshift);
v4 = (vector float)vec_sr((vector unsigned int)v4, vshift);
v1 = (vector float)vec_add((vector unsigned int)v1,(vector unsigned int)v2);
v2 = (vector float)vec_add((vector unsigned int)v3,(vector unsigned int)v4);
vclip = vec_sums((vector signed int)v1,vclip);
vclip = vec_sums((vector signed int)v2,vclip);
}
if((size_t)samples & 0xf)
{
v1 = (vector float)vec_perm(vec_ld(0,samples),vec_ld(0,samples),vec_lvsl(0,samples));
v2 = (vector float)vec_perm(vprev,(vector signed short)v1,vperm2);
vec_st((vector signed short)v2,0,samples);
}
vec_st(vclip,0,clip_tmp);
clip = clip_tmp[3];
}
fr->buffer.fill += 128;
return clip;
}
int synth_1to1_real_altivec(real *bandPtr,int channel,mpg123_handle *fr, int final)
{
real *samples = (real *) (fr->buffer.data+fr->buffer.fill);
real *b0, **buf;
int bo1;
if(fr->have_eq_settings) do_equalizer(bandPtr,channel,fr->equalizer);
if(!channel)
{
fr->bo--;
fr->bo &= 0xf;
buf = fr->real_buffs[0];
}
else
{
samples++;
buf = fr->real_buffs[1];
}
if(fr->bo & 0x1)
{
b0 = buf[0];
bo1 = fr->bo;
dct64_altivec(buf[1]+((fr->bo+1)&0xf),buf[0]+fr->bo,bandPtr);
}
else
{
b0 = buf[1];
bo1 = fr->bo+1;
dct64_altivec(buf[0]+fr->bo,buf[1]+fr->bo+1,bandPtr);
}
{
register int j;
real *window = fr->decwin + 16 - bo1;
vector float v1,v2,v3,v4,v5,v6,v7,v8,v9;
vector unsigned char vperm1,vperm2,vperm3,vperm4, vperm5;
vector float vsum,vsum2,vsum3,vsum4,vscale,vzero;
vector float vsample1,vsample2,vsample3;
vzero = vec_xor(vzero, vzero);
#ifdef __APPLE__
vscale = (vector float)(1.0f/32768.0f);
vperm4 = (vector unsigned char)(0,1,2,3,20,21,22,23,4,5,6,7,28,29,30,31);
vperm5 = (vector unsigned char)(8,9,10,11,20,21,22,23,12,13,14,15,28,29,30,31);
#else
vscale = (vector float){1.0f/32768.0f,1.0f/32768.0f,1.0f/32768.0f,1.0f/32768.0f};
vperm4 = (vector unsigned char){0,1,2,3,20,21,22,23,4,5,6,7,28,29,30,31};
vperm5 = (vector unsigned char){8,9,10,11,20,21,22,23,12,13,14,15,28,29,30,31};
#endif
vperm1 = vec_lvsl(0,window);
vperm2 = vec_lvsl(0,samples);
vperm3 = vec_lvsr(0,samples);
for (j=4;j;j--)
{
SYNTH_ALTIVEC(16);
vsum = vec_sub(v5,v6);
v9 = vec_sub(v7,v8);
vsum = vec_add(vsum,v9);
vsum = vec_madd(vsum, vscale, vzero);
vsample1 = vec_ld(0,samples);
vsample2 = vec_ld(16,samples);
vsample3 = vec_ld(31,samples);
v1 = vec_perm(vsample1, vsample2, vperm2);
v2 = vec_perm(vsample2, vsample3, vperm2);
v1 = vec_perm(vsum, v1, vperm4);
v2 = vec_perm(vsum, v2, vperm5);
v3 = vec_perm(vsample3, vsample2, vperm2);
v4 = vec_perm(vsample2, vsample1, vperm2);
v5 = vec_perm(v2, v3, vperm3);
v6 = vec_perm(v1, v2, vperm3);
v7 = vec_perm(v4, v1, vperm3);
vec_st(v5,31,samples);
vec_st(v6,16,samples);
vec_st(v7,0,samples);
samples += 8;
}
for (j=4;j;j--)
{
SYNTH_ALTIVEC(-16);
vsum = vec_add(v5,v6);
v9 = vec_add(v7,v8);
vsum = vec_add(vsum,v9);
vsum = vec_madd(vsum, vscale, vzero);
vsample1 = vec_ld(0,samples);
vsample2 = vec_ld(16,samples);
vsample3 = vec_ld(31,samples);
v1 = vec_perm(vsample1, vsample2, vperm2);
v2 = vec_perm(vsample2, vsample3, vperm2);
v1 = vec_perm(vsum, v1, vperm4);
v2 = vec_perm(vsum, v2, vperm5);
v3 = vec_perm(vsample3, vsample2, vperm2);
v4 = vec_perm(vsample2, vsample1, vperm2);
v5 = vec_perm(v2, v3, vperm3);
v6 = vec_perm(v1, v2, vperm3);
v7 = vec_perm(v4, v1, vperm3);
vec_st(v5,31,samples);
vec_st(v6,16,samples);
vec_st(v7,0,samples);
samples += 8;
}
}
if(final) fr->buffer.fill += 256;
return 0;
}
int synth_1to1_real_stereo_altivec(real *bandPtr_l, real *bandPtr_r, mpg123_handle *fr)
{
real *samples = (real *) (fr->buffer.data+fr->buffer.fill);
real *b0l, *b0r, **bufl, **bufr;
int bo1;
if(fr->have_eq_settings)
{
do_equalizer(bandPtr_l,0,fr->equalizer);
do_equalizer(bandPtr_r,1,fr->equalizer);
}
fr->bo--;
fr->bo &= 0xf;
bufl = fr->real_buffs[0];
bufr = fr->real_buffs[1];
if(fr->bo & 0x1)
{
b0l = bufl[0];
b0r = bufr[0];
bo1 = fr->bo;
dct64_altivec(bufl[1]+((fr->bo+1)&0xf),bufl[0]+fr->bo,bandPtr_l);
dct64_altivec(bufr[1]+((fr->bo+1)&0xf),bufr[0]+fr->bo,bandPtr_r);
}
else
{
b0l = bufl[1];
b0r = bufr[1];
bo1 = fr->bo+1;
dct64_altivec(bufl[0]+fr->bo,bufl[1]+fr->bo+1,bandPtr_l);
dct64_altivec(bufr[0]+fr->bo,bufr[1]+fr->bo+1,bandPtr_r);
}
{
register int j;
real *window = fr->decwin + 16 - bo1;
vector float v1,v2,v3,v4,v5,v6,v7,v8,v9,v10,v11,v12,v13;
vector unsigned char vperm1,vperm2;
vector float vsum,vsum2,vsum3,vsum4,vsum5,vsum6,vsum7,vsum8,vscale,vzero;
vector float vprev;
vzero = vec_xor(vzero,vzero);
#ifdef __APPLE__
vscale = (vector float)(1.0f/32768.0f);
#else
vscale = (vector float){1.0f/32768.0f,1.0f/32768.0f,1.0f/32768.0f,1.0f/32768.0f};
#endif
vperm1 = vec_lvsl(0,window);
vperm2 = vec_lvsr(0,samples);
vprev = vec_perm(vec_ld(0,samples),vec_ld(0,samples),vec_lvsl(0,samples));
for (j=4;j;j--)
{
SYNTH_STEREO_ALTIVEC(16);
vsum = vec_sub(vsum,vsum2);
vsum2 = vec_sub(vsum5,vsum6);
vsum3 = vec_sub(vsum3,vsum4);
vsum4 = vec_sub(vsum7,vsum8);
vsum = vec_add(vsum,vsum3);
vsum2 = vec_add(vsum2,vsum4);
vsum = vec_madd(vsum, vscale, vzero);
vsum2 = vec_madd(vsum2, vscale, vzero);
v1 = vec_mergeh(vsum, vsum2);
v2 = vec_mergel(vsum, vsum2);
v3 = vec_perm(vprev,v1,vperm2);
v4 = vec_perm(v1,v2,vperm2);
vprev = v2;
vec_st(v3,0,samples);
vec_st(v4,16,samples);
samples += 8;
}
for (j=4;j;j--)
{
SYNTH_STEREO_ALTIVEC(-16);
vsum = vec_add(vsum,vsum2);
vsum2 = vec_add(vsum5,vsum6);
vsum3 = vec_add(vsum3,vsum4);
vsum4 = vec_add(vsum7,vsum8);
vsum = vec_add(vsum,vsum3);
vsum2 = vec_add(vsum2,vsum4);
vsum = vec_madd(vsum, vscale, vzero);
vsum2 = vec_madd(vsum2, vscale, vzero);
v1 = vec_mergeh(vsum, vsum2);
v2 = vec_mergel(vsum, vsum2);
v3 = vec_perm(vprev,v1,vperm2);
v4 = vec_perm(v1,v2,vperm2);
vprev = v2;
vec_st(v3,0,samples);
vec_st(v4,16,samples);
samples += 8;
}
if((size_t)samples & 0xf)
{
v1 = (vector float)vec_perm(vec_ld(0,samples),vec_ld(0,samples),vec_lvsl(0,samples));
v2 = (vector float)vec_perm(vprev,v1,vperm2);
vec_st(v2,0,samples);
}
}
fr->buffer.fill += 256;
return 0;
}
int synth_1to1_s32_altivec(real *bandPtr,int channel,mpg123_handle *fr, int final)
{
int32_t *samples = (int32_t *) (fr->buffer.data+fr->buffer.fill);
real *b0, **buf;
int clip;
int bo1;
if(fr->have_eq_settings) do_equalizer(bandPtr,channel,fr->equalizer);
if(!channel)
{
fr->bo--;
fr->bo &= 0xf;
buf = fr->real_buffs[0];
}
else
{
samples++;
buf = fr->real_buffs[1];
}
if(fr->bo & 0x1)
{
b0 = buf[0];
bo1 = fr->bo;
dct64_altivec(buf[1]+((fr->bo+1)&0xf),buf[0]+fr->bo,bandPtr);
}
else
{
b0 = buf[1];
bo1 = fr->bo+1;
dct64_altivec(buf[0]+fr->bo,buf[1]+fr->bo+1,bandPtr);
}
{
register int j;
real *window = fr->decwin + 16 - bo1;
ALIGNED(16) int clip_tmp[4];
vector float v1,v2,v3,v4,v5,v6,v7,v8,v9;
vector unsigned char vperm1,vperm2,vperm3,vperm4,vperm5;
vector float vsum,vsum2,vsum3,vsum4,vmax,vmin,vzero;
vector signed int vsample1,vsample2,vsample3;
vector unsigned int vshift;
vector signed int vclip;
vzero = vec_xor(vzero, vzero);
vclip = vec_xor(vclip, vclip);
vshift = vec_splat_u32(-1); /* 31 */
#ifdef __APPLE__
vmax = (vector float)(32767.999f);
vmin = (vector float)(-32768.0f);
vperm4 = (vector unsigned char)(0,1,2,3,20,21,22,23,4,5,6,7,28,29,30,31);
vperm5 = (vector unsigned char)(8,9,10,11,20,21,22,23,12,13,14,15,28,29,30,31);
#else
vmax = (vector float){32767.999f,32767.999f,32767.999f,32767.999f};
vmin = (vector float){-32768.0f,-32768.0f,-32768.0f,-32768.0f};
vperm4 = (vector unsigned char){0,1,2,3,20,21,22,23,4,5,6,7,28,29,30,31};
vperm5 = (vector unsigned char){8,9,10,11,20,21,22,23,12,13,14,15,28,29,30,31};
#endif
vperm1 = vec_lvsl(0,window);
vperm2 = vec_lvsl(0,samples);
vperm3 = vec_lvsr(0,samples);
for (j=4;j;j--)
{
SYNTH_ALTIVEC(16);
vsum = vec_sub(v5,v6);
v9 = vec_sub(v7,v8);
v1 = vec_add(vsum,v9);
vsum = (vector float)vec_cts(v1,16);
v8 = (vector float)vec_cmpgt(v1,vmax);
v9 = (vector float)vec_cmplt(v1,vmin);
vsample1 = vec_ld(0,samples);
vsample2 = vec_ld(16,samples);
vsample3 = vec_ld(31,samples);
v1 = (vector float)vec_perm(vsample1, vsample2, vperm2);
v2 = (vector float)vec_perm(vsample2, vsample3, vperm2);
v1 = vec_perm(vsum, v1, vperm4);
v2 = vec_perm(vsum, v2, vperm5);
v3 = (vector float)vec_perm(vsample3, vsample2, vperm2);
v4 = (vector float)vec_perm(vsample2, vsample1, vperm2);
v5 = vec_perm(v2, v3, vperm3);
v6 = vec_perm(v1, v2, vperm3);
v7 = vec_perm(v4, v1, vperm3);
vec_st((vector signed int)v5,31,samples);
vec_st((vector signed int)v6,16,samples);
vec_st((vector signed int)v7,0,samples);
samples += 8;
v1 = (vector float)vec_sr((vector unsigned int)v8, vshift);
v2 = (vector float)vec_sr((vector unsigned int)v9, vshift);
v1 = (vector float)vec_add((vector unsigned int)v1,(vector unsigned int)v2);
vclip = vec_sums((vector signed int)v1,vclip);
}
for (j=4;j;j--)
{
SYNTH_ALTIVEC(-16);
vsum = vec_add(v5,v6);
v9 = vec_add(v7,v8);
v1 = vec_add(vsum,v9);
vsum = (vector float)vec_cts(v1,16);
v8 = (vector float)vec_cmpgt(v1,vmax);
v9 = (vector float)vec_cmplt(v1,vmin);
vsample1 = vec_ld(0,samples);
vsample2 = vec_ld(16,samples);
vsample3 = vec_ld(31,samples);
v1 = (vector float)vec_perm(vsample1, vsample2, vperm2);
v2 = (vector float)vec_perm(vsample2, vsample3, vperm2);
v1 = vec_perm(vsum, v1, vperm4);
v2 = vec_perm(vsum, v2, vperm5);
v3 = (vector float)vec_perm(vsample3, vsample2, vperm2);
v4 = (vector float)vec_perm(vsample2, vsample1, vperm2);
v5 = vec_perm(v2, v3, vperm3);
v6 = vec_perm(v1, v2, vperm3);
v7 = vec_perm(v4, v1, vperm3);
vec_st((vector signed int)v5,31,samples);
vec_st((vector signed int)v6,16,samples);
vec_st((vector signed int)v7,0,samples);
samples += 8;
v1 = (vector float)vec_sr((vector unsigned int)v8, vshift);
v2 = (vector float)vec_sr((vector unsigned int)v9, vshift);
v1 = (vector float)vec_add((vector unsigned int)v1,(vector unsigned int)v2);
vclip = vec_sums((vector signed int)v1,vclip);
}
vec_st(vclip,0,clip_tmp);
clip = clip_tmp[3];
}
if(final) fr->buffer.fill += 256;
return clip;
}
int synth_1to1_s32_stereo_altivec(real *bandPtr_l, real *bandPtr_r, mpg123_handle *fr)
{
int32_t *samples = (int32_t *) (fr->buffer.data+fr->buffer.fill);
real *b0l, *b0r, **bufl, **bufr;
int clip;
int bo1;
if(fr->have_eq_settings)
{
do_equalizer(bandPtr_l,0,fr->equalizer);
do_equalizer(bandPtr_r,1,fr->equalizer);
}
fr->bo--;
fr->bo &= 0xf;
bufl = fr->real_buffs[0];
bufr = fr->real_buffs[1];
if(fr->bo & 0x1)
{
b0l = bufl[0];
b0r = bufr[0];
bo1 = fr->bo;
dct64_altivec(bufl[1]+((fr->bo+1)&0xf),bufl[0]+fr->bo,bandPtr_l);
dct64_altivec(bufr[1]+((fr->bo+1)&0xf),bufr[0]+fr->bo,bandPtr_r);
}
else
{
b0l = bufl[1];
b0r = bufr[1];
bo1 = fr->bo+1;
dct64_altivec(bufl[0]+fr->bo,bufl[1]+fr->bo+1,bandPtr_l);
dct64_altivec(bufr[0]+fr->bo,bufr[1]+fr->bo+1,bandPtr_r);
}
{
register int j;
real *window = fr->decwin + 16 - bo1;
ALIGNED(16) int clip_tmp[4];
vector float v1,v2,v3,v4,v5,v6,v7,v8,v9,v10,v11,v12,v13;
vector unsigned char vperm1,vperm2;
vector float vsum,vsum2,vsum3,vsum4,vsum5,vsum6,vsum7,vsum8,vmax,vmin,vzero;
vector float vprev;
vector unsigned int vshift;
vector signed int vclip;
vzero = vec_xor(vzero, vzero);
vclip = vec_xor(vclip, vclip);
vshift = vec_splat_u32(-1); /* 31 */
#ifdef __APPLE__
vmax = (vector float)(32767.999f);
vmin = (vector float)(-32768.0f);
#else
vmax = (vector float){32767.999f,32767.999f,32767.999f,32767.999f};
vmin = (vector float){-32768.0f,-32768.0f,-32768.0f,-32768.0f};
#endif
vperm1 = vec_lvsl(0,window);
vperm2 = vec_lvsr(0,samples);
vprev = (vector float)vec_perm(vec_ld(0,samples),vec_ld(0,samples),vec_lvsl(0,samples));
for (j=4;j;j--)
{
SYNTH_STEREO_ALTIVEC(16);
vsum = vec_sub(vsum,vsum2);
vsum2 = vec_sub(vsum5,vsum6);
vsum3 = vec_sub(vsum3,vsum4);
vsum4 = vec_sub(vsum7,vsum8);
v1 = vec_add(vsum,vsum3);
v2 = vec_add(vsum2,vsum4);
vsum = (vector float)vec_cts(v1,16);
vsum2 = (vector float)vec_cts(v2,16);
v5 = (vector float)vec_cmpgt(v1,vmax);
v6 = (vector float)vec_cmplt(v1,vmin);
v7 = (vector float)vec_cmpgt(v2,vmax);
v8 = (vector float)vec_cmplt(v2,vmin);
v1 = vec_mergeh(vsum, vsum2);
v2 = vec_mergel(vsum, vsum2);
v3 = vec_perm(vprev,v1,vperm2);
v4 = vec_perm(v1,v2,vperm2);
vprev = v2;
vec_st((vector signed int)v3,0,samples);
vec_st((vector signed int)v4,16,samples);
samples += 8;
v1 = (vector float)vec_sr((vector unsigned int)v5, vshift);
v2 = (vector float)vec_sr((vector unsigned int)v6, vshift);
v3 = (vector float)vec_sr((vector unsigned int)v7, vshift);
v4 = (vector float)vec_sr((vector unsigned int)v8, vshift);
v1 = (vector float)vec_add((vector unsigned int)v1,(vector unsigned int)v2);
v2 = (vector float)vec_add((vector unsigned int)v3,(vector unsigned int)v4);
vclip = vec_sums((vector signed int)v1,vclip);
vclip = vec_sums((vector signed int)v2,vclip);
}
for (j=4;j;j--)
{
SYNTH_STEREO_ALTIVEC(-16);
vsum = vec_add(vsum,vsum2);
vsum2 = vec_add(vsum5,vsum6);
vsum3 = vec_add(vsum3,vsum4);
vsum4 = vec_add(vsum7,vsum8);
v1 = vec_add(vsum,vsum3);
v2 = vec_add(vsum2,vsum4);
vsum = (vector float)vec_cts(v1,16);
vsum2 = (vector float)vec_cts(v2,16);
v5 = (vector float)vec_cmpgt(v1,vmax);
v6 = (vector float)vec_cmplt(v1,vmin);
v7 = (vector float)vec_cmpgt(v2,vmax);
v8 = (vector float)vec_cmplt(v2,vmin);
v1 = vec_mergeh(vsum, vsum2);
v2 = vec_mergel(vsum, vsum2);
v3 = vec_perm(vprev,v1,vperm2);
v4 = vec_perm(v1,v2,vperm2);
vprev = v2;
vec_st((vector signed int)v3,0,samples);
vec_st((vector signed int)v4,16,samples);
samples += 8;
v1 = (vector float)vec_sr((vector unsigned int)v5, vshift);
v2 = (vector float)vec_sr((vector unsigned int)v6, vshift);
v3 = (vector float)vec_sr((vector unsigned int)v7, vshift);
v4 = (vector float)vec_sr((vector unsigned int)v8, vshift);
v1 = (vector float)vec_add((vector unsigned int)v1,(vector unsigned int)v2);
v2 = (vector float)vec_add((vector unsigned int)v3,(vector unsigned int)v4);
vclip = vec_sums((vector signed int)v1,vclip);
vclip = vec_sums((vector signed int)v2,vclip);
}
if((size_t)samples & 0xf)
{
v1 = (vector float)vec_perm(vec_ld(0,samples),vec_ld(0,samples),vec_lvsl(0,samples));
v2 = (vector float)vec_perm(vprev,v1,vperm2);
vec_st((vector signed int)v2,0,samples);
}
vec_st(vclip,0,clip_tmp);
clip = clip_tmp[3];
}
fr->buffer.fill += 256;
return clip;
}
Reference in a new issue Copy permalink