mirror of https://git.ffmpeg.org/ffmpeg.git
lavu/tx: refactor power-of-two FFT
This commit refactors the power-of-two FFT, making it faster and halving the size of all tables, making the code much smaller on all systems. This removes the big/small pass split, because on modern systems the "big" pass is always faster, and even on older machines there is no measurable speed difference.
This commit is contained in:
Lynne
parent
aa910a7ecd
commit
89da62f2fc
|
|
@ -105,7 +105,7 @@ typedef void FFTComplex;
|
||||||
CMUL((c).re, (c).im, (a).re, (a).im, (b).re, (b).im)
|
CMUL((c).re, (c).im, (a).re, (a).im, (b).re, (b).im)
|
||||||
|
|
||||||
#define COSTABLE(size) \
|
#define COSTABLE(size) \
|
||||||
DECLARE_ALIGNED(32, FFTSample, TX_NAME(ff_cos_##size))[size/2]
|
DECLARE_ALIGNED(32, FFTSample, TX_NAME(ff_cos_##size))[size/4 + 1]
|
||||||
|
|
||||||
/* Used by asm, reorder with care */
|
/* Used by asm, reorder with care */
|
||||||
struct AVTXContext {
|
struct AVTXContext {
|
||||||
|
|
|
||||||
|
|
@ -1,6 +1,8 @@
|
||||||
/*
|
/*
|
||||||
* Copyright (c) 2019 Lynne <dev@lynne.ee>
|
* Copyright (c) Lynne
|
||||||
|
*
|
||||||
* Power of two FFT:
|
* Power of two FFT:
|
||||||
|
* Copyright (c) Lynne
|
||||||
* Copyright (c) 2008 Loren Merritt
|
* Copyright (c) 2008 Loren Merritt
|
||||||
* Copyright (c) 2002 Fabrice Bellard
|
* Copyright (c) 2002 Fabrice Bellard
|
||||||
* Partly based on libdjbfft by D. J. Bernstein
|
* Partly based on libdjbfft by D. J. Bernstein
|
||||||
|
|
@ -65,10 +67,11 @@ static av_always_inline void init_cos_tabs_idx(int index)
|
||||||
int m = 1 << index;
|
int m = 1 << index;
|
||||||
double freq = 2*M_PI/m;
|
double freq = 2*M_PI/m;
|
||||||
FFTSample *tab = cos_tabs[index];
|
FFTSample *tab = cos_tabs[index];
|
||||||
for(int i = 0; i <= m/4; i++)
|
|
||||||
tab[i] = RESCALE(cos(i*freq));
|
for (int i = 0; i < m/4; i++)
|
||||||
for(int i = 1; i < m/4; i++)
|
*tab++ = RESCALE(cos(i*freq));
|
||||||
tab[m/2 - i] = tab[i];
|
|
||||||
|
*tab = 0;
|
||||||
}
|
}
|
||||||
|
|
||||||
#define INIT_FF_COS_TABS_FUNC(index, size) \
|
#define INIT_FF_COS_TABS_FUNC(index, size) \
|
||||||
|
|
@ -214,76 +217,60 @@ static av_always_inline void fft15(FFTComplex *out, FFTComplex *in,
|
||||||
fft5_m3(out, tmp + 10, stride);
|
fft5_m3(out, tmp + 10, stride);
|
||||||
}
|
}
|
||||||
|
|
||||||
#define BUTTERFLIES(a0,a1,a2,a3) {\
|
#define BUTTERFLIES(a0,a1,a2,a3) \
|
||||||
BF(t3, t5, t5, t1);\
|
do { \
|
||||||
BF(a2.re, a0.re, a0.re, t5);\
|
r0=a0.re; \
|
||||||
BF(a3.im, a1.im, a1.im, t3);\
|
i0=a0.im; \
|
||||||
BF(t4, t6, t2, t6);\
|
r1=a1.re; \
|
||||||
BF(a3.re, a1.re, a1.re, t4);\
|
i1=a1.im; \
|
||||||
BF(a2.im, a0.im, a0.im, t6);\
|
BF(t3, t5, t5, t1); \
|
||||||
}
|
BF(a2.re, a0.re, r0, t5); \
|
||||||
|
BF(a3.im, a1.im, i1, t3); \
|
||||||
|
BF(t4, t6, t2, t6); \
|
||||||
|
BF(a3.re, a1.re, r1, t4); \
|
||||||
|
BF(a2.im, a0.im, i0, t6); \
|
||||||
|
} while (0)
|
||||||
|
|
||||||
// force loading all the inputs before storing any.
|
#define TRANSFORM(a0,a1,a2,a3,wre,wim) \
|
||||||
// this is slightly slower for small data, but avoids store->load aliasing
|
do { \
|
||||||
// for addresses separated by large powers of 2.
|
CMUL(t1, t2, a2.re, a2.im, wre, -wim); \
|
||||||
#define BUTTERFLIES_BIG(a0,a1,a2,a3) {\
|
CMUL(t5, t6, a3.re, a3.im, wre, wim); \
|
||||||
FFTSample r0=a0.re, i0=a0.im, r1=a1.re, i1=a1.im;\
|
BUTTERFLIES(a0, a1, a2, a3); \
|
||||||
BF(t3, t5, t5, t1);\
|
} while (0)
|
||||||
BF(a2.re, a0.re, r0, t5);\
|
|
||||||
BF(a3.im, a1.im, i1, t3);\
|
|
||||||
BF(t4, t6, t2, t6);\
|
|
||||||
BF(a3.re, a1.re, r1, t4);\
|
|
||||||
BF(a2.im, a0.im, i0, t6);\
|
|
||||||
}
|
|
||||||
|
|
||||||
#define TRANSFORM(a0,a1,a2,a3,wre,wim) {\
|
|
||||||
CMUL(t1, t2, a2.re, a2.im, wre, -wim);\
|
|
||||||
CMUL(t5, t6, a3.re, a3.im, wre, wim);\
|
|
||||||
BUTTERFLIES(a0,a1,a2,a3)\
|
|
||||||
}
|
|
||||||
|
|
||||||
#define TRANSFORM_ZERO(a0,a1,a2,a3) {\
|
|
||||||
t1 = a2.re;\
|
|
||||||
t2 = a2.im;\
|
|
||||||
t5 = a3.re;\
|
|
||||||
t6 = a3.im;\
|
|
||||||
BUTTERFLIES(a0,a1,a2,a3)\
|
|
||||||
}
|
|
||||||
|
|
||||||
/* z[0...8n-1], w[1...2n-1] */
|
/* z[0...8n-1], w[1...2n-1] */
|
||||||
#define PASS(name)\
|
static void split_radix_combine(FFTComplex *z, const FFTSample *cos, int n)
|
||||||
static void name(FFTComplex *z, const FFTSample *wre, unsigned int n)\
|
{
|
||||||
{\
|
int o1 = 2*n;
|
||||||
FFTSample t1, t2, t3, t4, t5, t6;\
|
int o2 = 4*n;
|
||||||
int o1 = 2*n;\
|
int o3 = 6*n;
|
||||||
int o2 = 4*n;\
|
const FFTSample *wim = cos + o1 - 7;
|
||||||
int o3 = 6*n;\
|
FFTSample t1, t2, t3, t4, t5, t6, r0, i0, r1, i1;
|
||||||
const FFTSample *wim = wre+o1;\
|
|
||||||
n--;\
|
for (int i = 0; i < n; i += 4) {
|
||||||
\
|
TRANSFORM(z[0], z[o1 + 0], z[o2 + 0], z[o3 + 0], cos[0], wim[7]);
|
||||||
TRANSFORM_ZERO(z[0],z[o1],z[o2],z[o3]);\
|
TRANSFORM(z[2], z[o1 + 2], z[o2 + 2], z[o3 + 2], cos[2], wim[5]);
|
||||||
TRANSFORM(z[1],z[o1+1],z[o2+1],z[o3+1],wre[1],wim[-1]);\
|
TRANSFORM(z[4], z[o1 + 4], z[o2 + 4], z[o3 + 4], cos[4], wim[3]);
|
||||||
do {\
|
TRANSFORM(z[6], z[o1 + 6], z[o2 + 6], z[o3 + 6], cos[6], wim[1]);
|
||||||
z += 2;\
|
|
||||||
wre += 2;\
|
TRANSFORM(z[1], z[o1 + 1], z[o2 + 1], z[o3 + 1], cos[1], wim[6]);
|
||||||
wim -= 2;\
|
TRANSFORM(z[3], z[o1 + 3], z[o2 + 3], z[o3 + 3], cos[3], wim[4]);
|
||||||
TRANSFORM(z[0],z[o1],z[o2],z[o3],wre[0],wim[0]);\
|
TRANSFORM(z[5], z[o1 + 5], z[o2 + 5], z[o3 + 5], cos[5], wim[2]);
|
||||||
TRANSFORM(z[1],z[o1+1],z[o2+1],z[o3+1],wre[1],wim[-1]);\
|
TRANSFORM(z[7], z[o1 + 7], z[o2 + 7], z[o3 + 7], cos[7], wim[0]);
|
||||||
} while(--n);\
|
|
||||||
|
z += 2*4;
|
||||||
|
cos += 2*4;
|
||||||
|
wim -= 2*4;
|
||||||
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
PASS(pass)
|
#define DECL_FFT(n, n2, n4) \
|
||||||
#undef BUTTERFLIES
|
static void fft##n(FFTComplex *z) \
|
||||||
#define BUTTERFLIES BUTTERFLIES_BIG
|
{ \
|
||||||
PASS(pass_big)
|
fft##n2(z); \
|
||||||
|
fft##n4(z + n4*2); \
|
||||||
#define DECL_FFT(n,n2,n4)\
|
fft##n4(z + n4*3); \
|
||||||
static void fft##n(FFTComplex *z)\
|
split_radix_combine(z, TX_NAME(ff_cos_##n), n4/2); \
|
||||||
{\
|
|
||||||
fft##n2(z);\
|
|
||||||
fft##n4(z+n4*2);\
|
|
||||||
fft##n4(z+n4*3);\
|
|
||||||
pass(z,TX_NAME(ff_cos_##n),n4/2);\
|
|
||||||
}
|
}
|
||||||
|
|
||||||
static void fft2(FFTComplex *z)
|
static void fft2(FFTComplex *z)
|
||||||
|
|
@ -310,7 +297,7 @@ static void fft4(FFTComplex *z)
|
||||||
|
|
||||||
static void fft8(FFTComplex *z)
|
static void fft8(FFTComplex *z)
|
||||||
{
|
{
|
||||||
FFTSample t1, t2, t3, t4, t5, t6;
|
FFTSample t1, t2, t3, t4, t5, t6, r0, i0, r1, i1;
|
||||||
|
|
||||||
fft4(z);
|
fft4(z);
|
||||||
|
|
||||||
|
|
@ -319,24 +306,30 @@ static void fft8(FFTComplex *z)
|
||||||
BF(t5, z[7].re, z[6].re, -z[7].re);
|
BF(t5, z[7].re, z[6].re, -z[7].re);
|
||||||
BF(t6, z[7].im, z[6].im, -z[7].im);
|
BF(t6, z[7].im, z[6].im, -z[7].im);
|
||||||
|
|
||||||
BUTTERFLIES(z[0],z[2],z[4],z[6]);
|
BUTTERFLIES(z[0], z[2], z[4], z[6]);
|
||||||
TRANSFORM(z[1],z[3],z[5],z[7],RESCALE(M_SQRT1_2),RESCALE(M_SQRT1_2));
|
TRANSFORM(z[1], z[3], z[5], z[7], RESCALE(M_SQRT1_2), RESCALE(M_SQRT1_2));
|
||||||
}
|
}
|
||||||
|
|
||||||
static void fft16(FFTComplex *z)
|
static void fft16(FFTComplex *z)
|
||||||
{
|
{
|
||||||
FFTSample t1, t2, t3, t4, t5, t6;
|
FFTSample t1, t2, t3, t4, t5, t6, r0, i0, r1, i1;
|
||||||
FFTSample cos_16_1 = TX_NAME(ff_cos_16)[1];
|
FFTSample cos_16_1 = TX_NAME(ff_cos_16)[1];
|
||||||
|
FFTSample cos_16_2 = TX_NAME(ff_cos_16)[2];
|
||||||
FFTSample cos_16_3 = TX_NAME(ff_cos_16)[3];
|
FFTSample cos_16_3 = TX_NAME(ff_cos_16)[3];
|
||||||
|
|
||||||
fft8(z);
|
fft8(z + 0);
|
||||||
fft4(z+8);
|
fft4(z + 8);
|
||||||
fft4(z+12);
|
fft4(z + 12);
|
||||||
|
|
||||||
TRANSFORM_ZERO(z[0],z[4],z[8],z[12]);
|
t1 = z[ 8].re;
|
||||||
TRANSFORM(z[2],z[6],z[10],z[14],RESCALE(M_SQRT1_2),RESCALE(M_SQRT1_2));
|
t2 = z[ 8].im;
|
||||||
TRANSFORM(z[1],z[5],z[9],z[13],cos_16_1,cos_16_3);
|
t5 = z[12].re;
|
||||||
TRANSFORM(z[3],z[7],z[11],z[15],cos_16_3,cos_16_1);
|
t6 = z[12].im;
|
||||||
|
BUTTERFLIES(z[0], z[4], z[8], z[12]);
|
||||||
|
|
||||||
|
TRANSFORM(z[ 2], z[ 6], z[10], z[14], cos_16_2, cos_16_2);
|
||||||
|
TRANSFORM(z[ 1], z[ 5], z[ 9], z[13], cos_16_1, cos_16_3);
|
||||||
|
TRANSFORM(z[ 3], z[ 7], z[11], z[15], cos_16_3, cos_16_1);
|
||||||
}
|
}
|
||||||
|
|
||||||
DECL_FFT(32,16,8)
|
DECL_FFT(32,16,8)
|
||||||
|
|
@ -344,7 +337,6 @@ DECL_FFT(64,32,16)
|
||||||
DECL_FFT(128,64,32)
|
DECL_FFT(128,64,32)
|
||||||
DECL_FFT(256,128,64)
|
DECL_FFT(256,128,64)
|
||||||
DECL_FFT(512,256,128)
|
DECL_FFT(512,256,128)
|
||||||
#define pass pass_big
|
|
||||||
DECL_FFT(1024,512,256)
|
DECL_FFT(1024,512,256)
|
||||||
DECL_FFT(2048,1024,512)
|
DECL_FFT(2048,1024,512)
|
||||||
DECL_FFT(4096,2048,1024)
|
DECL_FFT(4096,2048,1024)
|
||||||
|
|
@ -386,8 +378,8 @@ DECL_COMP_FFT(3)
|
||||||
DECL_COMP_FFT(5)
|
DECL_COMP_FFT(5)
|
||||||
DECL_COMP_FFT(15)
|
DECL_COMP_FFT(15)
|
||||||
|
|
||||||
static void monolithic_fft(AVTXContext *s, void *_out, void *_in,
|
static void split_radix_fft(AVTXContext *s, void *_out, void *_in,
|
||||||
ptrdiff_t stride)
|
ptrdiff_t stride)
|
||||||
{
|
{
|
||||||
FFTComplex *in = _in;
|
FFTComplex *in = _in;
|
||||||
FFTComplex *out = _out;
|
FFTComplex *out = _out;
|
||||||
|
|
@ -730,7 +722,7 @@ int TX_NAME(ff_tx_init_mdct_fft)(AVTXContext *s, av_tx_fn *tx,
|
||||||
n == 5 ? inv ? compound_imdct_5xM : compound_mdct_5xM :
|
n == 5 ? inv ? compound_imdct_5xM : compound_mdct_5xM :
|
||||||
inv ? compound_imdct_15xM : compound_mdct_15xM;
|
inv ? compound_imdct_15xM : compound_mdct_15xM;
|
||||||
} else { /* Direct transform case */
|
} else { /* Direct transform case */
|
||||||
*tx = monolithic_fft;
|
*tx = split_radix_fft;
|
||||||
if (is_mdct)
|
if (is_mdct)
|
||||||
*tx = inv ? monolithic_imdct : monolithic_mdct;
|
*tx = inv ? monolithic_imdct : monolithic_mdct;
|
||||||
}
|
}
|
||||||
|
|
|
||||||
Loading…
Reference in New Issue