virtualenv/FFmpeg
1
0
Fork 0
mirror of https://github.com/FFmpeg/FFmpeg.git synced 2025-02-04 06:08:26 +02:00
Code Issues Releases Activity

optimize imdct_half:

Browse Source 
remove tmp buffer.
skip fft reinterleave pass, leaving data in a format more convenient for simd.
merge post-rotate with post-reorder.

Originally committed as revision 14700 to svn://svn.ffmpeg.org/ffmpeg/trunk
This commit is contained in:
Loren Merritt 2008-08-12 00:33:34 +00:00
parent 49c0dd754c
commit 46803f4f67
6 changed files with 248 additions and 359 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

14
libavcodec/dsputil.h
View File

@ -645,7 +645,7 @@ typedef struct FFTContext {
void (*imdct_calc)(struct MDCTContext *s, FFTSample *output, void (*imdct_calc)(struct MDCTContext *s, FFTSample *output,
const FFTSample *input, FFTSample *tmp); const FFTSample *input, FFTSample *tmp);
void (*imdct_half)(struct MDCTContext *s, FFTSample *output, void (*imdct_half)(struct MDCTContext *s, FFTSample *output,
const FFTSample *input, FFTSample *tmp); const FFTSample *input);
} FFTContext; } FFTContext;
int ff_fft_init(FFTContext *s, int nbits, int inverse); int ff_fft_init(FFTContext *s, int nbits, int inverse);
@ -696,16 +696,16 @@ void ff_sine_window_init(float *window, int n);
int ff_mdct_init(MDCTContext *s, int nbits, int inverse); int ff_mdct_init(MDCTContext *s, int nbits, int inverse);
void ff_imdct_calc(MDCTContext *s, FFTSample *output, void ff_imdct_calc(MDCTContext *s, FFTSample *output,
const FFTSample *input, FFTSample *tmp); const FFTSample *input, FFTSample *tmp);
void ff_imdct_half(MDCTContext *s, FFTSample *output, void ff_imdct_half(MDCTContext *s, FFTSample *output, const FFTSample *input);
const FFTSample *input, FFTSample *tmp); void ff_imdct_calc_3dn(MDCTContext *s, FFTSample *output,
const FFTSample *input, FFTSample *tmp);
void ff_imdct_half_3dn(MDCTContext *s, FFTSample *output, const FFTSample *input);
void ff_imdct_calc_3dn2(MDCTContext *s, FFTSample *output, void ff_imdct_calc_3dn2(MDCTContext *s, FFTSample *output,
const FFTSample *input, FFTSample *tmp); const FFTSample *input, FFTSample *tmp);
void ff_imdct_half_3dn2(MDCTContext *s, FFTSample *output, void ff_imdct_half_3dn2(MDCTContext *s, FFTSample *output, const FFTSample *input);
const FFTSample *input, FFTSample *tmp);
void ff_imdct_calc_sse(MDCTContext *s, FFTSample *output, void ff_imdct_calc_sse(MDCTContext *s, FFTSample *output,
const FFTSample *input, FFTSample *tmp); const FFTSample *input, FFTSample *tmp);
void ff_imdct_half_sse(MDCTContext *s, FFTSample *output, void ff_imdct_half_sse(MDCTContext *s, FFTSample *output, const FFTSample *input);
const FFTSample *input, FFTSample *tmp);
void ff_mdct_calc(MDCTContext *s, FFTSample *out, void ff_mdct_calc(MDCTContext *s, FFTSample *out,
const FFTSample *input, FFTSample *tmp); const FFTSample *input, FFTSample *tmp);
void ff_mdct_end(MDCTContext *s); void ff_mdct_end(MDCTContext *s);

2
libavcodec/fft.c
View File

@ -106,6 +106,8 @@ int ff_fft_init(FFTContext *s, int nbits, int inverse)
s->fft_calc = ff_fft_calc_3dn2; s->fft_calc = ff_fft_calc_3dn2;
} else if (has_vectors & MM_3DNOW) { } else if (has_vectors & MM_3DNOW) {
/* 3DNow! for K6-2/3 */ /* 3DNow! for K6-2/3 */
s->imdct_calc = ff_imdct_calc_3dn;
s->imdct_half = ff_imdct_half_3dn;
s->fft_calc = ff_fft_calc_3dn; s->fft_calc = ff_fft_calc_3dn;
} }
#elif defined HAVE_ALTIVEC && !defined ALTIVEC_USE_REFERENCE_C_CODE #elif defined HAVE_ALTIVEC && !defined ALTIVEC_USE_REFERENCE_C_CODE

193
libavcodec/i386/fft_3dn2.c
View File

@ -1,7 +1,6 @@
/* /*
* FFT/MDCT transform with Extended 3DNow! optimizations * FFT/MDCT transform with Extended 3DNow! optimizations
* Copyright (c) 2006 Zuxy MENG Jie, Loren Merritt * Copyright (c) 2006-2008 Zuxy MENG Jie, Loren Merritt
* Based on fft_sse.c copyright (c) 2002 Fabrice Bellard.
* *
* This file is part of FFmpeg. * This file is part of FFmpeg.
* *
@ -23,16 +22,23 @@
#include "libavutil/x86_cpu.h" #include "libavutil/x86_cpu.h"
#include "libavcodec/dsputil.h" #include "libavcodec/dsputil.h"
DECLARE_ALIGNED_8(static const int, m1m1[2]) = { 1<<31, 1<<31 };
#ifdef EMULATE_3DNOWEXT #ifdef EMULATE_3DNOWEXT
#define PSWAPD(s,d)\
"movq "#s","#d"\n"\
"psrlq $32,"#d"\n"\
"punpckldq "#s","#d"\n"
#define ff_fft_calc_3dn2 ff_fft_calc_3dn #define ff_fft_calc_3dn2 ff_fft_calc_3dn
#define ff_fft_dispatch_3dn2 ff_fft_dispatch_3dn #define ff_fft_dispatch_3dn2 ff_fft_dispatch_3dn
#define ff_fft_dispatch_interleave_3dn2 ff_fft_dispatch_interleave_3dn #define ff_fft_dispatch_interleave_3dn2 ff_fft_dispatch_interleave_3dn
#define ff_imdct_calc_3dn2 ff_imdct_calc_3dn #define ff_imdct_calc_3dn2 ff_imdct_calc_3dn
#define ff_imdct_half_3dn2 ff_imdct_half_3dn #define ff_imdct_half_3dn2 ff_imdct_half_3dn
#else
#define PSWAPD(s,d) "pswapd "#s","#d"\n"
#endif #endif
void ff_fft_dispatch_3dn2(FFTComplex *z, int nbits); void ff_fft_dispatch_3dn2(FFTComplex *z, int nbits);
void ff_fft_dispatch_interleave_3dn2(FFTComplex *z, int nbits);
void ff_fft_calc_3dn2(FFTContext *s, FFTComplex *z) void ff_fft_calc_3dn2(FFTContext *s, FFTComplex *z)
{ {
@ -45,35 +51,45 @@ void ff_fft_calc_3dn2(FFTContext *s, FFTComplex *z)
FFSWAP(FFTSample, z[i].im, z[i+1].re); FFSWAP(FFTSample, z[i].im, z[i+1].re);
} }
static void imdct_3dn2(MDCTContext *s, const FFTSample *input, FFTSample *tmp) void ff_imdct_half_3dn2(MDCTContext *s, FFTSample *output, const FFTSample *input)
{ {
long n4, n2, n; x86_reg j, k;
x86_reg k; long n = 1 << s->nbits;
long n2 = n >> 1;
long n4 = n >> 2;
long n8 = n >> 3;
const uint16_t *revtab = s->fft.revtab; const uint16_t *revtab = s->fft.revtab;
const FFTSample *tcos = s->tcos; const FFTSample *tcos = s->tcos;
const FFTSample *tsin = s->tsin; const FFTSample *tsin = s->tsin;
const FFTSample *in1, *in2; const FFTSample *in1, *in2;
FFTComplex *z = (FFTComplex *)tmp; FFTComplex *z = (FFTComplex *)output;
n = 1 << s->nbits;
n2 = n >> 1;
n4 = n >> 2;
/* pre rotation */ /* pre rotation */
in1 = input; in1 = input;
in2 = input + n2 - 1; in2 = input + n2 - 1;
#ifdef EMULATE_3DNOWEXT
asm volatile("movd %0, %%mm7" ::"r"(1<<31));
#endif
for(k = 0; k < n4; k++) { for(k = 0; k < n4; k++) {
// FIXME a single block is faster, but gcc 2.95 and 3.4.x on 32bit can't compile it // FIXME a single block is faster, but gcc 2.95 and 3.4.x on 32bit can't compile it
asm volatile( asm volatile(
"movd %0, %%mm0 \n\t" "movd %0, %%mm0 \n"
"movd %2, %%mm1 \n\t" "movd %2, %%mm1 \n"
"punpckldq %1, %%mm0 \n\t" "punpckldq %1, %%mm0 \n"
"punpckldq %3, %%mm1 \n\t" "punpckldq %3, %%mm1 \n"
"movq %%mm0, %%mm2 \n\t" "movq %%mm0, %%mm2 \n"
"pfmul %%mm1, %%mm0 \n\t" PSWAPD( %%mm1, %%mm3 )
"pswapd %%mm1, %%mm1 \n\t" "pfmul %%mm1, %%mm0 \n"
"pfmul %%mm1, %%mm2 \n\t" "pfmul %%mm3, %%mm2 \n"
"pfpnacc %%mm2, %%mm0 \n\t" #ifdef EMULATE_3DNOWEXT
"movq %%mm0, %%mm1 \n"
"punpckhdq %%mm2, %%mm0 \n"
"punpckldq %%mm2, %%mm1 \n"
"pxor %%mm7, %%mm0 \n"
"pfadd %%mm1, %%mm0 \n"
#else
"pfpnacc %%mm2, %%mm0 \n"
#endif
::"m"(in2[-2*k]), "m"(in1[2*k]), ::"m"(in2[-2*k]), "m"(in1[2*k]),
"m"(tcos[k]), "m"(tsin[k]) "m"(tcos[k]), "m"(tsin[k])
); );
@ -83,101 +99,74 @@ static void imdct_3dn2(MDCTContext *s, const FFTSample *input, FFTSample *tmp)
); );
} }
ff_fft_calc_3dn2(&s->fft, z); ff_fft_dispatch_3dn2(z, s->fft.nbits);
/* post rotation + reordering */ #define CMUL(j,mm0,mm1)\
for(k = 0; k < n4; k++) { "movq (%2,"#j",2), %%mm6 \n"\
asm volatile( "movq 8(%2,"#j",2), "#mm0"\n"\
"movq %0, %%mm0 \n\t" "movq %%mm6, "#mm1"\n"\
"movd %1, %%mm1 \n\t" "movq "#mm0",%%mm7 \n"\
"punpckldq %2, %%mm1 \n\t" "pfmul (%3,"#j"), %%mm6 \n"\
"movq %%mm0, %%mm2 \n\t" "pfmul (%4,"#j"), "#mm0"\n"\
"pfmul %%mm1, %%mm0 \n\t" "pfmul (%4,"#j"), "#mm1"\n"\
"pswapd %%mm1, %%mm1 \n\t" "pfmul (%3,"#j"), %%mm7 \n"\
"pfmul %%mm1, %%mm2 \n\t" "pfsub %%mm6, "#mm0"\n"\
"pfpnacc %%mm2, %%mm0 \n\t" "pfadd %%mm7, "#mm1"\n"
"movq %%mm0, %0 \n\t"
:"+m"(z[k]) /* post rotation */
:"m"(tcos[k]), "m"(tsin[k]) j = -n2;
); k = n2-8;
} asm volatile(
"1: \n"
CMUL(%0, %%mm0, %%mm1)
CMUL(%1, %%mm2, %%mm3)
"movd %%mm0, (%2,%0,2) \n"
"movd %%mm1,12(%2,%1,2) \n"
"movd %%mm2, (%2,%1,2) \n"
"movd %%mm3,12(%2,%0,2) \n"
"psrlq $32, %%mm0 \n"
"psrlq $32, %%mm1 \n"
"psrlq $32, %%mm2 \n"
"psrlq $32, %%mm3 \n"
"movd %%mm0, 8(%2,%0,2) \n"
"movd %%mm1, 4(%2,%1,2) \n"
"movd %%mm2, 8(%2,%1,2) \n"
"movd %%mm3, 4(%2,%0,2) \n"
"sub $8, %1 \n"
"add $8, %0 \n"
"jl 1b \n"
:"+r"(j), "+r"(k)
:"r"(z+n8), "r"(tcos+n8), "r"(tsin+n8)
:"memory"
);
asm volatile("femms");
} }
void ff_imdct_calc_3dn2(MDCTContext *s, FFTSample *output, void ff_imdct_calc_3dn2(MDCTContext *s, FFTSample *output,
const FFTSample *input, FFTSample *tmp) const FFTSample *input, FFTSample *tmp)
{
x86_reg k;
long n8, n2, n;
FFTComplex *z = (FFTComplex *)tmp;
n = 1 << s->nbits;
n2 = n >> 1;
n8 = n >> 3;
imdct_3dn2(s, input, tmp);
k = n-8;
asm volatile("movd %0, %%mm7" ::"r"(1<<31));
asm volatile(
"1: \n\t"
"movq (%4,%0), %%mm0 \n\t" // z[n8+k]
"neg %0 \n\t"
"pswapd -8(%4,%0), %%mm1 \n\t" // z[n8-1-k]
"movq %%mm0, %%mm2 \n\t"
"pxor %%mm7, %%mm2 \n\t"
"punpckldq %%mm1, %%mm2 \n\t"
"pswapd %%mm2, %%mm3 \n\t"
"punpckhdq %%mm1, %%mm0 \n\t"
"pswapd %%mm0, %%mm4 \n\t"
"pxor %%mm7, %%mm0 \n\t"
"pxor %%mm7, %%mm4 \n\t"
"movq %%mm3, -8(%3,%0) \n\t" // output[n-2-2*k] = { z[n8-1-k].im, -z[n8+k].re }
"movq %%mm4, -8(%2,%0) \n\t" // output[n2-2-2*k]= { -z[n8-1-k].re, z[n8+k].im }
"neg %0 \n\t"
"movq %%mm0, (%1,%0) \n\t" // output[2*k] = { -z[n8+k].im, z[n8-1-k].re }
"movq %%mm2, (%2,%0) \n\t" // output[n2+2*k] = { -z[n8+k].re, z[n8-1-k].im }
"sub $8, %0 \n\t"
"jge 1b \n\t"
:"+r"(k)
:"r"(output), "r"(output+n2), "r"(output+n), "r"(z+n8)
:"memory"
);
asm volatile("femms");
}
void ff_imdct_half_3dn2(MDCTContext *s, FFTSample *output,
const FFTSample *input, FFTSample *tmp)
{ {
x86_reg j, k; x86_reg j, k;
long n8, n4, n; long n = 1 << s->nbits;
FFTComplex *z = (FFTComplex *)tmp; long n4 = n >> 2;
n = 1 << s->nbits; ff_imdct_half_3dn2(s, output+n4, input);
n4 = n >> 2;
n8 = n >> 3;
imdct_3dn2(s, input, tmp);
j = -n; j = -n;
k = n-8; k = n-8;
asm volatile("movd %0, %%mm7" ::"r"(1<<31));
asm volatile( asm volatile(
"1: \n\t" "movq %4, %%mm7 \n"
"movq (%3,%1), %%mm0 \n\t" // z[n8+k] "1: \n"
"pswapd (%3,%0), %%mm1 \n\t" // z[n8-1-k] PSWAPD((%2,%1), %%mm0)
"movq %%mm0, %%mm2 \n\t" PSWAPD((%3,%0), %%mm1)
"punpckldq %%mm1, %%mm0 \n\t" "pxor %%mm7, %%mm0 \n"
"punpckhdq %%mm2, %%mm1 \n\t" "movq %%mm1, (%3,%1) \n"
"pxor %%mm7, %%mm0 \n\t" "movq %%mm0, (%2,%0) \n"
"pxor %%mm7, %%mm1 \n\t" "sub $8, %1 \n"
"movq %%mm0, (%2,%1) \n\t" // output[n4+2*k] = { -z[n8+k].re, z[n8-1-k].im } "add $8, %0 \n"
"movq %%mm1, (%2,%0) \n\t" // output[n4-2-2*k] = { -z[n8-1-k].re, z[n8+k].im } "jl 1b \n"
"sub $8, %1 \n\t"
"add $8, %0 \n\t"
"jl 1b \n\t"
:"+r"(j), "+r"(k) :"+r"(j), "+r"(k)
:"r"(output+n4), "r"(z+n8) :"r"(output+n4), "r"(output+n4*3),
:"memory" "m"(*m1m1)
); );
asm volatile("femms"); asm volatile("femms");
} }

295
libavcodec/i386/fft_sse.c
View File

@ -1,6 +1,6 @@
/* /*
* FFT/MDCT transform with SSE optimizations * FFT/MDCT transform with SSE optimizations
* Copyright (c) 2002 Fabrice Bellard. * Copyright (c) 2008 Loren Merritt
* *
* This file is part of FFmpeg. * This file is part of FFmpeg.
* *
@ -22,9 +22,6 @@
#include "libavutil/x86_cpu.h" #include "libavutil/x86_cpu.h"
#include "libavcodec/dsputil.h" #include "libavcodec/dsputil.h"
static const int p1m1p1m1[4] __attribute__((aligned(16))) =
{ 0, 1 << 31, 0, 1 << 31 };
static const int m1m1m1m1[4] __attribute__((aligned(16))) = static const int m1m1m1m1[4] __attribute__((aligned(16))) =
{ 1 << 31, 1 << 31, 1 << 31, 1 << 31 }; { 1 << 31, 1 << 31, 1 << 31, 1 << 31 };
@ -73,212 +70,134 @@ void ff_fft_permute_sse(FFTContext *s, FFTComplex *z)
memcpy(z, s->tmp_buf, n*sizeof(FFTComplex)); memcpy(z, s->tmp_buf, n*sizeof(FFTComplex));
} }
static void imdct_sse(MDCTContext *s, const FFTSample *input, FFTSample *tmp) void ff_imdct_half_sse(MDCTContext *s, FFTSample *output, const FFTSample *input)
{ {
x86_reg k; av_unused x86_reg i, j, k, l;
long n4, n2, n; long n = 1 << s->nbits;
const uint16_t *revtab = s->fft.revtab; long n2 = n >> 1;
long n4 = n >> 2;
long n8 = n >> 3;
const uint16_t *revtab = s->fft.revtab + n8;
const FFTSample *tcos = s->tcos; const FFTSample *tcos = s->tcos;
const FFTSample *tsin = s->tsin; const FFTSample *tsin = s->tsin;
const FFTSample *in1, *in2; FFTComplex *z = (FFTComplex *)output;
FFTComplex *z = (FFTComplex *)tmp;
n = 1 << s->nbits;
n2 = n >> 1;
n4 = n >> 2;
#ifdef ARCH_X86_64
asm volatile ("movaps %0, %%xmm8\n\t"::"m"(*p1m1p1m1));
#define P1M1P1M1 "%%xmm8"
#else
#define P1M1P1M1 "%4"
#endif
/* pre rotation */ /* pre rotation */
in1 = input; for(k=n8-2; k>=0; k-=2) {
in2 = input + n2 - 4; asm volatile(
"movaps (%2,%1,2), %%xmm0 \n" // { z[k].re, z[k].im, z[k+1].re, z[k+1].im }
/* Complex multiplication */ "movaps -16(%2,%0,2), %%xmm1 \n" // { z[-k-2].re, z[-k-2].im, z[-k-1].re, z[-k-1].im }
for (k = 0; k < n4; k += 4) { "movaps %%xmm0, %%xmm2 \n"
asm volatile ( "shufps $0x88, %%xmm1, %%xmm0 \n" // { z[k].re, z[k+1].re, z[-k-2].re, z[-k-1].re }
"movaps %0, %%xmm0 \n\t" // xmm0 = r0 X r1 X : in2 "shufps $0x77, %%xmm2, %%xmm1 \n" // { z[-k-1].im, z[-k-2].im, z[k+1].im, z[k].im }
"movaps %1, %%xmm3 \n\t" // xmm3 = X i1 X i0: in1 "movlps (%3,%1), %%xmm4 \n"
"movaps -16+1*%0, %%xmm4 \n\t" // xmm4 = r0 X r1 X : in2 "movlps (%4,%1), %%xmm5 \n"
"movaps 16+1*%1, %%xmm7 \n\t" // xmm7 = X i1 X i0: in1 "movhps -8(%3,%0), %%xmm4 \n" // { cos[k], cos[k+1], cos[-k-2], cos[-k-1] }
"movlps %2, %%xmm1 \n\t" // xmm1 = X X R1 R0: tcos "movhps -8(%4,%0), %%xmm5 \n" // { sin[k], sin[k+1], sin[-k-2], sin[-k-1] }
"movlps %3, %%xmm2 \n\t" // xmm2 = X X I1 I0: tsin "movaps %%xmm0, %%xmm2 \n"
"movlps 8+1*%2, %%xmm5 \n\t" // xmm5 = X X R1 R0: tcos "movaps %%xmm1, %%xmm3 \n"
"movlps 8+1*%3, %%xmm6 \n\t" // xmm6 = X X I1 I0: tsin "mulps %%xmm5, %%xmm0 \n" // re*sin
"shufps $95, %%xmm0, %%xmm0 \n\t" // xmm0 = r1 r1 r0 r0 "mulps %%xmm4, %%xmm1 \n" // im*cos
"shufps $160,%%xmm3, %%xmm3 \n\t" // xmm3 = i1 i1 i0 i0 "mulps %%xmm4, %%xmm2 \n" // re*cos
"shufps $95, %%xmm4, %%xmm4 \n\t" // xmm4 = r1 r1 r0 r0 "mulps %%xmm5, %%xmm3 \n" // im*sin
"shufps $160,%%xmm7, %%xmm7 \n\t" // xmm7 = i1 i1 i0 i0 "subps %%xmm0, %%xmm1 \n" // -> re
"unpcklps %%xmm2, %%xmm1 \n\t" // xmm1 = I1 R1 I0 R0 "addps %%xmm3, %%xmm2 \n" // -> im
"unpcklps %%xmm6, %%xmm5 \n\t" // xmm5 = I1 R1 I0 R0 "movaps %%xmm1, %%xmm0 \n"
"movaps %%xmm1, %%xmm2 \n\t" // xmm2 = I1 R1 I0 R0 "unpcklps %%xmm2, %%xmm1 \n" // { z[k], z[k+1] }
"movaps %%xmm5, %%xmm6 \n\t" // xmm6 = I1 R1 I0 R0 "unpckhps %%xmm2, %%xmm0 \n" // { z[-k-2], z[-k-1] }
"xorps "P1M1P1M1", %%xmm2 \n\t" // xmm2 = -I1 R1 -I0 R0 ::"r"(-4*k), "r"(4*k),
"xorps "P1M1P1M1", %%xmm6 \n\t" // xmm6 = -I1 R1 -I0 R0 "r"(input+n4), "r"(tcos+n8), "r"(tsin+n8)
"mulps %%xmm1, %%xmm0 \n\t" // xmm0 = rI rR rI rR );
"mulps %%xmm5, %%xmm4 \n\t" // xmm4 = rI rR rI rR #ifdef ARCH_X86_64
"shufps $177,%%xmm2, %%xmm2 \n\t" // xmm2 = R1 -I1 R0 -I0 // if we have enough regs, don't let gcc make the luts latency-bound
"shufps $177,%%xmm6, %%xmm6 \n\t" // xmm6 = R1 -I1 R0 -I0 // but if not, latency is faster than spilling
"mulps %%xmm2, %%xmm3 \n\t" // xmm3 = Ri -Ii Ri -Ii asm("movlps %%xmm0, %0 \n"
"mulps %%xmm6, %%xmm7 \n\t" // xmm7 = Ri -Ii Ri -Ii "movhps %%xmm0, %1 \n"
"addps %%xmm3, %%xmm0 \n\t" // xmm0 = result "movlps %%xmm1, %2 \n"
"addps %%xmm7, %%xmm4 \n\t" // xmm4 = result "movhps %%xmm1, %3 \n"
::"m"(in2[-2*k]), "m"(in1[2*k]), :"=m"(z[revtab[-k-2]]),
"m"(tcos[k]), "m"(tsin[k]) "=m"(z[revtab[-k-1]]),
#ifndef ARCH_X86_64 "=m"(z[revtab[ k ]]),
,"m"(*p1m1p1m1) "=m"(z[revtab[ k+1]])
);
#else
asm("movlps %%xmm0, %0" :"=m"(z[revtab[-k-2]]));
asm("movhps %%xmm0, %0" :"=m"(z[revtab[-k-1]]));
asm("movlps %%xmm1, %0" :"=m"(z[revtab[ k ]]));
asm("movhps %%xmm1, %0" :"=m"(z[revtab[ k+1]]));
#endif #endif
);
/* Should be in the same block, hack for gcc2.95 & gcc3 */
asm (
"movlps %%xmm0, %0 \n\t"
"movhps %%xmm0, %1 \n\t"
"movlps %%xmm4, %2 \n\t"
"movhps %%xmm4, %3 \n\t"
:"=m"(z[revtab[k]]), "=m"(z[revtab[k + 1]]),
"=m"(z[revtab[k + 2]]), "=m"(z[revtab[k + 3]])
);
} }
ff_fft_calc_sse(&s->fft, z); ff_fft_dispatch_sse(z, s->fft.nbits);
#ifndef ARCH_X86_64 /* post rotation + reinterleave + reorder */
#undef P1M1P1M1
#define P1M1P1M1 "%3"
#endif
/* post rotation + reordering */ #define CMUL(j,xmm0,xmm1)\
for (k = 0; k < n4; k += 4) { "movaps (%2,"#j",2), %%xmm6 \n"\
asm ( "movaps 16(%2,"#j",2), "#xmm0"\n"\
"movaps %0, %%xmm0 \n\t" // xmm0 = i1 r1 i0 r0: z "movaps %%xmm6, "#xmm1"\n"\
"movaps 16+1*%0, %%xmm4 \n\t" // xmm4 = i1 r1 i0 r0: z "movaps "#xmm0",%%xmm7 \n"\
"movlps %1, %%xmm1 \n\t" // xmm1 = X X R1 R0: tcos "mulps (%3,"#j"), %%xmm6 \n"\
"movlps 8+1*%1, %%xmm5 \n\t" // xmm5 = X X R1 R0: tcos "mulps (%4,"#j"), "#xmm0"\n"\
"movaps %%xmm0, %%xmm3 \n\t" // xmm3 = i1 r1 i0 r0 "mulps (%4,"#j"), "#xmm1"\n"\
"movaps %%xmm4, %%xmm7 \n\t" // xmm7 = i1 r1 i0 r0 "mulps (%3,"#j"), %%xmm7 \n"\
"movlps %2, %%xmm2 \n\t" // xmm2 = X X I1 I0: tsin "subps %%xmm6, "#xmm0"\n"\
"movlps 8+1*%2, %%xmm6 \n\t" // xmm6 = X X I1 I0: tsin "addps %%xmm7, "#xmm1"\n"
"shufps $160,%%xmm0, %%xmm0 \n\t" // xmm0 = r1 r1 r0 r0
"shufps $245,%%xmm3, %%xmm3 \n\t" // xmm3 = i1 i1 i0 i0 j = -n2;
"shufps $160,%%xmm4, %%xmm4 \n\t" // xmm4 = r1 r1 r0 r0 k = n2-16;
"shufps $245,%%xmm7, %%xmm7 \n\t" // xmm7 = i1 i1 i0 i0 asm volatile(
"unpcklps %%xmm2, %%xmm1 \n\t" // xmm1 = I1 R1 I0 R0 "1: \n"
"unpcklps %%xmm6, %%xmm5 \n\t" // xmm5 = I1 R1 I0 R0 CMUL(%0, %%xmm0, %%xmm1)
"movaps %%xmm1, %%xmm2 \n\t" // xmm2 = I1 R1 I0 R0 CMUL(%1, %%xmm4, %%xmm5)
"movaps %%xmm5, %%xmm6 \n\t" // xmm6 = I1 R1 I0 R0 "shufps $0x1b, %%xmm1, %%xmm1 \n"
"xorps "P1M1P1M1", %%xmm2 \n\t" // xmm2 = -I1 R1 -I0 R0 "shufps $0x1b, %%xmm5, %%xmm5 \n"
"mulps %%xmm1, %%xmm0 \n\t" // xmm0 = rI rR rI rR "movaps %%xmm4, %%xmm6 \n"
"xorps "P1M1P1M1", %%xmm6 \n\t" // xmm6 = -I1 R1 -I0 R0 "unpckhps %%xmm1, %%xmm4 \n"
"mulps %%xmm5, %%xmm4 \n\t" // xmm4 = rI rR rI rR "unpcklps %%xmm1, %%xmm6 \n"
"shufps $177,%%xmm2, %%xmm2 \n\t" // xmm2 = R1 -I1 R0 -I0 "movaps %%xmm0, %%xmm2 \n"
"shufps $177,%%xmm6, %%xmm6 \n\t" // xmm6 = R1 -I1 R0 -I0 "unpcklps %%xmm5, %%xmm0 \n"
"mulps %%xmm2, %%xmm3 \n\t" // xmm3 = Ri -Ii Ri -Ii "unpckhps %%xmm5, %%xmm2 \n"
"mulps %%xmm6, %%xmm7 \n\t" // xmm7 = Ri -Ii Ri -Ii "movaps %%xmm6, (%2,%1,2) \n"
"addps %%xmm3, %%xmm0 \n\t" // xmm0 = result "movaps %%xmm4, 16(%2,%1,2) \n"
"addps %%xmm7, %%xmm4 \n\t" // xmm4 = result "movaps %%xmm0, (%2,%0,2) \n"
"movaps %%xmm0, %0 \n\t" "movaps %%xmm2, 16(%2,%0,2) \n"
"movaps %%xmm4, 16+1*%0\n\t" "sub $16, %1 \n"
:"+m"(z[k]) "add $16, %0 \n"
:"m"(tcos[k]), "m"(tsin[k]) "jl 1b \n"
#ifndef ARCH_X86_64 :"+&r"(j), "+&r"(k)
,"m"(*p1m1p1m1) :"r"(z+n8), "r"(tcos+n8), "r"(tsin+n8)
#endif :"memory"
); );
}
} }
void ff_imdct_calc_sse(MDCTContext *s, FFTSample *output, void ff_imdct_calc_sse(MDCTContext *s, FFTSample *output,
const FFTSample *input, FFTSample *tmp) const FFTSample *input, FFTSample *tmp)
{
x86_reg k;
long n8, n2, n;
FFTComplex *z = (FFTComplex *)tmp;
n = 1 << s->nbits;
n2 = n >> 1;
n8 = n >> 3;
imdct_sse(s, input, tmp);
/*
Mnemonics:
0 = z[k].re
1 = z[k].im
2 = z[k + 1].re
3 = z[k + 1].im
4 = z[-k - 2].re
5 = z[-k - 2].im
6 = z[-k - 1].re
7 = z[-k - 1].im
*/
k = 16-n;
asm volatile("movaps %0, %%xmm7 \n\t"::"m"(*m1m1m1m1));
asm volatile(
"1: \n\t"
"movaps -16(%4,%0), %%xmm1 \n\t" // xmm1 = 4 5 6 7 = z[-2-k]
"neg %0 \n\t"
"movaps (%4,%0), %%xmm0 \n\t" // xmm0 = 0 1 2 3 = z[k]
"xorps %%xmm7, %%xmm0 \n\t" // xmm0 = -0 -1 -2 -3
"movaps %%xmm0, %%xmm2 \n\t" // xmm2 = -0 -1 -2 -3
"shufps $141,%%xmm1, %%xmm0 \n\t" // xmm0 = -1 -3 4 6
"shufps $216,%%xmm1, %%xmm2 \n\t" // xmm2 = -0 -2 5 7
"shufps $156,%%xmm0, %%xmm0 \n\t" // xmm0 = -1 6 -3 4 !
"shufps $156,%%xmm2, %%xmm2 \n\t" // xmm2 = -0 7 -2 5 !
"movaps %%xmm0, (%1,%0) \n\t" // output[2*k]
"movaps %%xmm2, (%2,%0) \n\t" // output[n2+2*k]
"neg %0 \n\t"
"shufps $27, %%xmm0, %%xmm0 \n\t" // xmm0 = 4 -3 6 -1
"xorps %%xmm7, %%xmm0 \n\t" // xmm0 = -4 3 -6 1 !
"shufps $27, %%xmm2, %%xmm2 \n\t" // xmm2 = 5 -2 7 -0 !
"movaps %%xmm0, -16(%2,%0) \n\t" // output[n2-4-2*k]
"movaps %%xmm2, -16(%3,%0) \n\t" // output[n-4-2*k]
"add $16, %0 \n\t"
"jle 1b \n\t"
:"+r"(k)
:"r"(output), "r"(output+n2), "r"(output+n), "r"(z+n8)
:"memory"
);
}
void ff_imdct_half_sse(MDCTContext *s, FFTSample *output,
const FFTSample *input, FFTSample *tmp)
{ {
x86_reg j, k; x86_reg j, k;
long n8, n4, n; long n = 1 << s->nbits;
FFTComplex *z = (FFTComplex *)tmp; long n4 = n >> 2;
n = 1 << s->nbits; ff_imdct_half_sse(s, output+n4, input);
n4 = n >> 2;
n8 = n >> 3;
imdct_sse(s, input, tmp);
j = -n; j = -n;
k = n-16; k = n-16;
asm volatile("movaps %0, %%xmm7 \n\t"::"m"(*m1m1m1m1));
asm volatile( asm volatile(
"1: \n\t" "movaps %4, %%xmm7 \n"
"movaps (%3,%1), %%xmm0 \n\t" "1: \n"
"movaps (%3,%0), %%xmm1 \n\t" "movaps (%2,%1), %%xmm0 \n"
"xorps %%xmm7, %%xmm0 \n\t" "movaps (%3,%0), %%xmm1 \n"
"movaps %%xmm0, %%xmm2 \n\t" "shufps $0x1b, %%xmm0, %%xmm0 \n"
"shufps $141,%%xmm1, %%xmm0 \n\t" "shufps $0x1b, %%xmm1, %%xmm1 \n"
"shufps $216,%%xmm1, %%xmm2 \n\t" "xorps %%xmm7, %%xmm0 \n"
"shufps $54, %%xmm0, %%xmm0 \n\t" "movaps %%xmm1, (%3,%1) \n"
"shufps $156,%%xmm2, %%xmm2 \n\t" "movaps %%xmm0, (%2,%0) \n"
"xorps %%xmm7, %%xmm0 \n\t" "sub $16, %1 \n"
"movaps %%xmm2, (%2,%1) \n\t" "add $16, %0 \n"
"movaps %%xmm0, (%2,%0) \n\t" "jl 1b \n"
"sub $16, %1 \n\t"
"add $16, %0 \n\t"
"jl 1b \n\t"
:"+r"(j), "+r"(k) :"+r"(j), "+r"(k)
:"r"(output+n4), "r"(z+n8) :"r"(output+n4), "r"(output+n4*3),
:"memory" "m"(*m1m1m1m1)
); );
} }

93
libavcodec/mdct.c
View File

@ -100,18 +100,25 @@ int ff_mdct_init(MDCTContext *s, int nbits, int inverse)
(pim) = _are * _bim + _aim * _bre;\ (pim) = _are * _bim + _aim * _bre;\
} }
static void imdct_c(MDCTContext *s, const FFTSample *input, FFTSample *tmp) /**
* Compute the middle half of the inverse MDCT of size N = 2^nbits,
* thus excluding the parts that can be derived by symmetry
* @param output N/2 samples
* @param input N/2 samples
*/
void ff_imdct_half(MDCTContext *s, FFTSample *output, const FFTSample *input)
{ {
int k, n4, n2, n, j; int k, n8, n4, n2, n, j;
const uint16_t *revtab = s->fft.revtab; const uint16_t *revtab = s->fft.revtab;
const FFTSample *tcos = s->tcos; const FFTSample *tcos = s->tcos;
const FFTSample *tsin = s->tsin; const FFTSample *tsin = s->tsin;
const FFTSample *in1, *in2; const FFTSample *in1, *in2;
FFTComplex *z = (FFTComplex *)tmp; FFTComplex *z = (FFTComplex *)output;
n = 1 << s->nbits; n = 1 << s->nbits;
n2 = n >> 1; n2 = n >> 1;
n4 = n >> 2; n4 = n >> 2;
n8 = n >> 3;
/* pre rotation */ /* pre rotation */
in1 = input; in1 = input;
@ -125,9 +132,15 @@ static void imdct_c(MDCTContext *s, const FFTSample *input, FFTSample *tmp)
ff_fft_calc(&s->fft, z); ff_fft_calc(&s->fft, z);
/* post rotation + reordering */ /* post rotation + reordering */
/* XXX: optimize */ output += n4;
for(k = 0; k < n4; k++) { for(k = 0; k < n8; k++) {
CMUL(z[k].re, z[k].im, z[k].re, z[k].im, tcos[k], tsin[k]); FFTSample r0, i0, r1, i1;
CMUL(r0, i1, z[n8-k-1].im, z[n8-k-1].re, tsin[n8-k-1], tcos[n8-k-1]);
CMUL(r1, i0, z[n8+k ].im, z[n8+k ].re, tsin[n8+k ], tcos[n8+k ]);
z[n8-k-1].re = r0;
z[n8-k-1].im = i0;
z[n8+k ].re = r1;
z[n8+k ].im = i1;
} }
} }
@ -140,52 +153,16 @@ static void imdct_c(MDCTContext *s, const FFTSample *input, FFTSample *tmp)
void ff_imdct_calc(MDCTContext *s, FFTSample *output, void ff_imdct_calc(MDCTContext *s, FFTSample *output,
const FFTSample *input, FFTSample *tmp) const FFTSample *input, FFTSample *tmp)
{ {
int k, n8, n2, n; int k;
FFTComplex *z = (FFTComplex *)tmp; int n = 1 << s->nbits;
n = 1 << s->nbits; int n2 = n >> 1;
n2 = n >> 1; int n4 = n >> 2;
n8 = n >> 3;
imdct_c(s, input, tmp); ff_imdct_half(s, output+n4, input);
for(k = 0; k < n8; k++) { for(k = 0; k < n4; k++) {
output[2*k] = -z[n8 + k].im; output[k] = -output[n2-k-1];
output[n2-1-2*k] = z[n8 + k].im; output[n-k-1] = output[n2+k];
output[2*k+1] = z[n8-1-k].re;
output[n2-1-2*k-1] = -z[n8-1-k].re;
output[n2 + 2*k]=-z[k+n8].re;
output[n-1- 2*k]=-z[k+n8].re;
output[n2 + 2*k+1]=z[n8-k-1].im;
output[n-2 - 2 * k] = z[n8-k-1].im;
}
}
/**
* Compute the middle half of the inverse MDCT of size N = 2^nbits,
* thus excluding the parts that can be derived by symmetry
* @param output N/2 samples
* @param input N/2 samples
* @param tmp N/2 samples
*/
void ff_imdct_half(MDCTContext *s, FFTSample *output,
const FFTSample *input, FFTSample *tmp)
{
int k, n8, n4, n;
FFTComplex *z = (FFTComplex *)tmp;
n = 1 << s->nbits;
n4 = n >> 2;
n8 = n >> 3;
imdct_c(s, input, tmp);
for(k = 0; k < n8; k++) {
output[n4-1-2*k] = z[n8+k].im;
output[n4-1-2*k-1] = -z[n8-k-1].re;
output[n4 + 2*k] = -z[n8+k].re;
output[n4 + 2*k+1] = z[n8-k-1].im;
} }
} }
@ -203,7 +180,7 @@ void ff_mdct_calc(MDCTContext *s, FFTSample *out,
const uint16_t *revtab = s->fft.revtab; const uint16_t *revtab = s->fft.revtab;
const FFTSample *tcos = s->tcos; const FFTSample *tcos = s->tcos;
const FFTSample *tsin = s->tsin; const FFTSample *tsin = s->tsin;
FFTComplex *x = (FFTComplex *)tmp; FFTComplex *x = (FFTComplex *)out;
n = 1 << s->nbits; n = 1 << s->nbits;
n2 = n >> 1; n2 = n >> 1;
@ -227,12 +204,14 @@ void ff_mdct_calc(MDCTContext *s, FFTSample *out,
ff_fft_calc(&s->fft, x); ff_fft_calc(&s->fft, x);
/* post rotation */ /* post rotation */
for(i=0;i<n4;i++) { for(i=0;i<n8;i++) {
re = x[i].re; FFTSample r0, i0, r1, i1;
im = x[i].im; CMUL(i1, r0, x[n8-i-1].re, x[n8-i-1].im, -tsin[n8-i-1], -tcos[n8-i-1]);
CMUL(re1, im1, re, im, -tsin[i], -tcos[i]); CMUL(i0, r1, x[n8+i ].re, x[n8+i ].im, -tsin[n8+i ], -tcos[n8+i ]);
out[2*i] = im1; x[n8-i-1].re = r0;
out[n2-1-2*i] = re1; x[n8-i-1].im = i0;
x[n8+i ].re = r1;
x[n8+i ].im = i1;
} }
} }

10
libavcodec/vorbis_dec.c
View File

@ -1517,18 +1517,18 @@ static int vorbis_parse_audio_packet(vorbis_context *vc) {
// MDCT, overlap/add, save data for next overlapping FPMATH // MDCT, overlap/add, save data for next overlapping FPMATH
retlen = (blocksize + vc->blocksize[previous_window])/4; retlen = (blocksize + vc->blocksize[previous_window])/4;
dir = retlen <= blocksize/2; // pick an order so that ret[] can reuse residues[] without stepping on any data we need dir = retlen <= blocksize/2; // pick an order so that ret[] can reuse floors[] without stepping on any data we need
for(j=dir?0:vc->audio_channels-1; (unsigned)j<vc->audio_channels; j+=dir*2-1) { for(j=dir?0:vc->audio_channels-1; (unsigned)j<vc->audio_channels; j+=dir*2-1) {
uint_fast16_t bs0=vc->blocksize[0]; uint_fast16_t bs0=vc->blocksize[0];
uint_fast16_t bs1=vc->blocksize[1]; uint_fast16_t bs1=vc->blocksize[1];
float *residue=vc->channel_residues+res_chan[j]*blocksize/2; float *residue=vc->channel_residues+res_chan[j]*blocksize/2;
float *floor=vc->channel_floors+j*blocksize/2; float *floor=vc->channel_floors+j*blocksize/2;
float *saved=vc->saved+j*bs1/4; float *saved=vc->saved+j*bs1/4;
float *ret=vc->channel_residues+j*retlen; float *ret=vc->channel_floors+j*retlen;
float *buf=floor; float *buf=residue;
const float *win=vc->win[blockflag&previous_window]; const float *win=vc->win[blockflag&previous_window];
vc->mdct[0].fft.imdct_half(&vc->mdct[blockflag], buf, floor, residue); vc->mdct[0].fft.imdct_half(&vc->mdct[blockflag], buf, floor);
if(blockflag == previous_window) { if(blockflag == previous_window) {
vc->dsp.vector_fmul_window(ret, saved, buf, win, fadd_bias, blocksize/4); vc->dsp.vector_fmul_window(ret, saved, buf, win, fadd_bias, blocksize/4);
@ -1583,7 +1583,7 @@ static int vorbis_decode_frame(AVCodecContext *avccontext,
AV_DEBUG("parsed %d bytes %d bits, returned %d samples (*ch*bits) \n", get_bits_count(gb)/8, get_bits_count(gb)%8, len); AV_DEBUG("parsed %d bytes %d bits, returned %d samples (*ch*bits) \n", get_bits_count(gb)/8, get_bits_count(gb)%8, len);
for(i=0; i<vc->audio_channels; i++) for(i=0; i<vc->audio_channels; i++)
channel_ptrs[i] = vc->channel_residues+i*len; channel_ptrs[i] = vc->channel_floors+i*len;
vc->dsp.float_to_int16_interleave(data, channel_ptrs, len, vc->audio_channels); vc->dsp.float_to_int16_interleave(data, channel_ptrs, len, vc->audio_channels);
*data_size=len*2*vc->audio_channels; *data_size=len*2*vc->audio_channels;