mirror of
https://gitlab.com/libeigen/eigen.git
synced 2025-08-12 03:39:01 +08:00
refactored ei_kissfft_impl to maintain a cache of cpx fft plans
This commit is contained in:
Mark Borgerding
parent
210092d16c
commit
03ed6f9bfb
@ -24,112 +24,36 @@
|
||||
|
||||
#include <complex>
|
||||
#include <vector>
|
||||
#include <map>
|
||||
|
||||
namespace Eigen {
|
||||
|
||||
template <typename _Scalar>
|
||||
struct ei_kissfft_impl
|
||||
struct ei_kiss_cpx_fft
|
||||
{
|
||||
typedef _Scalar Scalar;
|
||||
typedef std::complex<Scalar> Complex;
|
||||
ei_kissfft_impl() : m_nfft(0) {}
|
||||
std::vector<Complex> m_twiddles;
|
||||
std::vector<int> m_stageRadix;
|
||||
std::vector<int> m_stageRemainder;
|
||||
bool m_inverse;
|
||||
|
||||
template <typename _Src>
|
||||
void fwd( Complex * dst,const _Src *src,int nfft)
|
||||
{
|
||||
prepare(nfft,false);
|
||||
work(0, dst, src, 1,1);
|
||||
}
|
||||
|
||||
// real-to-complex forward FFT
|
||||
// perform two FFTs of src even and src odd
|
||||
// then twiddle to recombine them into the half-spectrum format
|
||||
// then fill in the conjugate symmetric half
|
||||
void fwd( Complex * dst,const Scalar * src,int nfft)
|
||||
{
|
||||
if ( nfft&1 ) {
|
||||
// use generic mode for odd
|
||||
prepare(nfft,false);
|
||||
work(0, dst, src, 1,1);
|
||||
}else{
|
||||
int ncfft = nfft>>1;
|
||||
int ncfft2 = nfft>>2;
|
||||
// use optimized mode for even real
|
||||
fwd( dst, reinterpret_cast<const Complex*> (src),ncfft);
|
||||
make_real_twiddles(nfft);
|
||||
Complex dc = dst[0].real() + dst[0].imag();
|
||||
Complex nyquist = dst[0].real() - dst[0].imag();
|
||||
int k;
|
||||
for ( k=1;k <= ncfft2 ; ++k ) {
|
||||
Complex fpk = dst[k];
|
||||
Complex fpnk = conj(dst[ncfft-k]);
|
||||
Complex f1k = fpk + fpnk;
|
||||
Complex f2k = fpk - fpnk;
|
||||
//Complex tw = f2k * exp( Complex(0,-3.14159265358979323846264338327 * ((double) (k) / ncfft + .5) ) );
|
||||
Complex tw= f2k * m_realTwiddles[k-1];
|
||||
|
||||
dst[k] = (f1k + tw) * Scalar(.5);
|
||||
dst[ncfft-k] = conj(f1k -tw)*Scalar(.5);
|
||||
}
|
||||
|
||||
// place conjugate-symmetric half at the end for completeness
|
||||
// TODO: make this configurable ( opt-out )
|
||||
for ( k=1;k < ncfft ; ++k )
|
||||
dst[nfft-k] = conj(dst[k]);
|
||||
|
||||
dst[0] = dc;
|
||||
dst[ncfft] = nyquist;
|
||||
}
|
||||
}
|
||||
|
||||
// half-complex to scalar
|
||||
void inv( Scalar * dst,const Complex * src,int nfft)
|
||||
{
|
||||
// TODO add optimized version for even numbers
|
||||
std::vector<Complex> tmp(nfft);
|
||||
inv(&tmp[0],src,nfft);
|
||||
for (int k=0;k<nfft;++k)
|
||||
dst[k] = tmp[k].real();
|
||||
}
|
||||
|
||||
void inv(Complex * dst,const Complex *src,int nfft)
|
||||
{
|
||||
prepare(nfft,true);
|
||||
work(0, dst, src, 1,1);
|
||||
scale(dst, Scalar(1)/m_nfft );
|
||||
}
|
||||
|
||||
void prepare(int nfft,bool inverse)
|
||||
{
|
||||
make_twiddles(nfft,inverse);
|
||||
factorize(nfft);
|
||||
}
|
||||
|
||||
void make_real_twiddles(int nfft)
|
||||
{
|
||||
int ncfft2 = nfft>>2;
|
||||
if ( m_realTwiddles.size() != ncfft2) {
|
||||
m_realTwiddles.resize(ncfft2);
|
||||
int ncfft= nfft>>1;
|
||||
for (int k=1;k<=ncfft2;++k)
|
||||
m_realTwiddles[k-1] = exp( Complex(0,-3.14159265358979323846264338327 * ((double) (k) / ncfft + .5) ) );
|
||||
}
|
||||
}
|
||||
ei_kiss_cpx_fft() { }
|
||||
|
||||
void make_twiddles(int nfft,bool inverse)
|
||||
{
|
||||
if ( m_twiddles.size() == nfft) {
|
||||
// reuse the twiddles, conjugate if necessary
|
||||
if (inverse != m_inverse)
|
||||
for (int i=0;i<nfft;++i)
|
||||
m_twiddles[i] = conj( m_twiddles[i] );
|
||||
}else{
|
||||
m_inverse = inverse;
|
||||
m_twiddles.resize(nfft);
|
||||
Scalar phinc = (inverse?2:-2)* acos( (Scalar) -1) / nfft;
|
||||
for (int i=0;i<nfft;++i)
|
||||
m_twiddles[i] = exp( Complex(0,i*phinc) );
|
||||
}
|
||||
m_inverse = inverse;
|
||||
|
||||
void invert()
|
||||
{
|
||||
m_inverse = !m_inverse;
|
||||
for ( size_t i=0;i<m_twiddles.size() ;++i)
|
||||
m_twiddles[i] = conj( m_twiddles[i] );
|
||||
}
|
||||
|
||||
void factorize(int nfft)
|
||||
@ -157,17 +81,8 @@ namespace Eigen {
|
||||
m_stageRemainder.push_back(n);
|
||||
}while(n>1);
|
||||
}
|
||||
m_nfft = nfft;
|
||||
}
|
||||
|
||||
void scale(Complex *dst,Scalar s)
|
||||
{
|
||||
for (int k=0;k<m_nfft;++k)
|
||||
dst[k] *= s;
|
||||
}
|
||||
|
||||
private:
|
||||
|
||||
template <typename _Src>
|
||||
void work( int stage,Complex * xout, const _Src * xin, size_t fstride,size_t in_stride)
|
||||
{
|
||||
@ -338,7 +253,7 @@ namespace Eigen {
|
||||
int u,k,q1,q;
|
||||
Complex * twiddles = &m_twiddles[0];
|
||||
Complex t;
|
||||
int Norig = m_nfft;
|
||||
int Norig = m_twiddles.size();
|
||||
Complex * scratchbuf = (Complex*)alloca(p*sizeof(Complex) );
|
||||
|
||||
for ( u=0; u<m; ++u ) {
|
||||
@ -362,36 +277,141 @@ namespace Eigen {
|
||||
}
|
||||
}
|
||||
}
|
||||
};
|
||||
|
||||
int m_nfft;
|
||||
bool m_inverse;
|
||||
std::vector<Complex> m_twiddles;
|
||||
std::vector<Complex> m_realTwiddles;
|
||||
std::vector<int> m_stageRadix;
|
||||
std::vector<int> m_stageRemainder;
|
||||
|
||||
template <typename _Scalar>
|
||||
struct ei_kissfft_impl
|
||||
{
|
||||
typedef _Scalar Scalar;
|
||||
typedef std::complex<Scalar> Complex;
|
||||
ei_kissfft_impl() {}
|
||||
|
||||
void clear()
|
||||
{
|
||||
m_plans.clear();
|
||||
m_realTwiddles.clear();
|
||||
}
|
||||
|
||||
template <typename _Src>
|
||||
void fwd( Complex * dst,const _Src *src,int nfft)
|
||||
{
|
||||
get_plan(nfft,false).work(0, dst, src, 1,1);
|
||||
}
|
||||
|
||||
// real-to-complex forward FFT
|
||||
// perform two FFTs of src even and src odd
|
||||
// then twiddle to recombine them into the half-spectrum format
|
||||
// then fill in the conjugate symmetric half
|
||||
void fwd( Complex * dst,const Scalar * src,int nfft)
|
||||
{
|
||||
if ( nfft&3 ) {
|
||||
// use generic mode for odd
|
||||
get_plan(nfft,false).work(0, dst, src, 1,1);
|
||||
}else{
|
||||
int ncfft = nfft>>1;
|
||||
int ncfft2 = nfft>>2;
|
||||
Complex * rtw = real_twiddles(ncfft2);
|
||||
|
||||
// use optimized mode for even real
|
||||
fwd( dst, reinterpret_cast<const Complex*> (src), ncfft);
|
||||
Complex dc = dst[0].real() + dst[0].imag();
|
||||
Complex nyquist = dst[0].real() - dst[0].imag();
|
||||
int k;
|
||||
for ( k=1;k <= ncfft2 ; ++k ) {
|
||||
Complex fpk = dst[k];
|
||||
Complex fpnk = conj(dst[ncfft-k]);
|
||||
Complex f1k = fpk + fpnk;
|
||||
Complex f2k = fpk - fpnk;
|
||||
Complex tw= f2k * rtw[k-1];
|
||||
|
||||
dst[k] = (f1k + tw) * Scalar(.5);
|
||||
dst[ncfft-k] = conj(f1k -tw)*Scalar(.5);
|
||||
}
|
||||
|
||||
// place conjugate-symmetric half at the end for completeness
|
||||
// TODO: make this configurable ( opt-out )
|
||||
for ( k=1;k < ncfft ; ++k )
|
||||
dst[nfft-k] = conj(dst[k]);
|
||||
|
||||
dst[0] = dc;
|
||||
dst[ncfft] = nyquist;
|
||||
}
|
||||
}
|
||||
|
||||
// half-complex to scalar
|
||||
void inv( Scalar * dst,const Complex * src,int nfft)
|
||||
{
|
||||
// TODO add optimized version for even numbers
|
||||
std::vector<Complex> tmp(nfft);
|
||||
inv(&tmp[0],src,nfft);
|
||||
for (int k=0;k<nfft;++k)
|
||||
dst[k] = tmp[k].real();
|
||||
}
|
||||
|
||||
void inv(Complex * dst,const Complex *src,int nfft)
|
||||
{
|
||||
get_plan(nfft,true).work(0, dst, src, 1,1);
|
||||
scale(dst, nfft, Scalar(1)/nfft );
|
||||
}
|
||||
|
||||
private:
|
||||
|
||||
typedef ei_kiss_cpx_fft<Scalar> PlanData;
|
||||
|
||||
typedef std::map<int,PlanData> PlanMap;
|
||||
PlanMap m_plans;
|
||||
std::map<int, std::vector<Complex> > m_realTwiddles;
|
||||
|
||||
int PlanKey(int nfft,bool isinverse) const { return (nfft<<1) | isinverse; }
|
||||
|
||||
PlanData & get_plan(int nfft,bool inverse)
|
||||
{
|
||||
/*
|
||||
enum {FORWARD,INVERSE,REAL,COMPLEX};
|
||||
|
||||
struct PlanKey
|
||||
{
|
||||
PlanKey(int nfft,bool isinverse,bool iscomplex)
|
||||
{
|
||||
_key = (nfft<<2) | (isinverse<<1) | iscomplex;
|
||||
* for some reason this does not work
|
||||
*
|
||||
typedef typename std::map<int,PlanData>::iterator MapIt;
|
||||
MapIt it;
|
||||
it = m_plans.find( PlanKey(nfft,inverse) );
|
||||
if (it == m_plans.end() ) {
|
||||
// create new entry
|
||||
it = m_plans.insert( make_pair( PlanKey(nfft,inverse) , PlanData() ) );
|
||||
MapIt it2 = m_plans.find( PlanKey(nfft,!inverse) );
|
||||
if (it2 != m_plans.end() ) {
|
||||
it->second = it2.second;
|
||||
it->second.invert();
|
||||
}else{
|
||||
it->second.make_twiddles(nfft,inverse);
|
||||
it->second.factorize(nfft);
|
||||
}
|
||||
|
||||
bool operator<(const PlanKey & other) const
|
||||
{
|
||||
return this->_key < other._key;
|
||||
}
|
||||
int _key;
|
||||
};
|
||||
|
||||
struct PlanData
|
||||
{
|
||||
std::vector<Complex> m_twiddles;
|
||||
};
|
||||
|
||||
std::map<PlanKey,
|
||||
return it->second;
|
||||
*/
|
||||
PlanData & pd = m_plans[ PlanKey(nfft,inverse) ];
|
||||
if ( pd.m_twiddles.size() == 0 ) {
|
||||
pd.make_twiddles(nfft,inverse);
|
||||
pd.factorize(nfft);
|
||||
}
|
||||
return pd;
|
||||
}
|
||||
|
||||
Complex * real_twiddles(int ncfft2)
|
||||
{
|
||||
std::vector<Complex> & twidref = m_realTwiddles[ncfft2];// creates new if not there
|
||||
if ( (int)twidref.size() != ncfft2 ) {
|
||||
twidref.resize(ncfft2);
|
||||
int ncfft= ncfft2<<1;
|
||||
Scalar pi = acos( Scalar(-1) );
|
||||
for (int k=1;k<=ncfft2;++k)
|
||||
twidref[k-1] = exp( Complex(0,-pi * ((double) (k) / ncfft + .5) ) );
|
||||
}
|
||||
return &twidref[0];
|
||||
}
|
||||
|
||||
void scale(Complex *dst,int n,Scalar s)
|
||||
{
|
||||
for (int k=0;k<n;++k)
|
||||
dst[k] *= s;
|
||||
}
|
||||
};
|
||||
}
|
||||
|
||||
@ -44,7 +44,7 @@ complex<long double> promote(long double x) { return complex<long double>( x);
|
||||
{
|
||||
long double totalpower=0;
|
||||
long double difpower=0;
|
||||
cerr <<"idx\ttruth\t\tvalue\n";
|
||||
cerr <<"idx\ttruth\t\tvalue\t|dif|=\n";
|
||||
for (size_t k0=0;k0<fftbuf.size();++k0) {
|
||||
complex<long double> acc = 0;
|
||||
long double phinc = -2.*k0* M_PIl / timebuf.size();
|
||||
@ -55,7 +55,7 @@ complex<long double> promote(long double x) { return complex<long double>( x);
|
||||
complex<long double> x = promote(fftbuf[k0]);
|
||||
complex<long double> dif = acc - x;
|
||||
difpower += norm(dif);
|
||||
cerr << k0 << "\t" << acc << "\t" << x << endl;
|
||||
cerr << k0 << "\t" << acc << "\t" << x << "\t" << sqrt(norm(dif)) << endl;
|
||||
}
|
||||
cerr << "rmse:" << sqrt(difpower/totalpower) << endl;
|
||||
return sqrt(difpower/totalpower);
|
||||
@ -127,8 +127,9 @@ void test_FFT()
|
||||
#endif
|
||||
|
||||
#if 1
|
||||
CALL_SUBTEST( test_scalar<float>(45) ); CALL_SUBTEST( test_scalar<double>(45) ); CALL_SUBTEST( test_scalar<long double>(45) );
|
||||
CALL_SUBTEST( test_scalar<float>(32) ); CALL_SUBTEST( test_scalar<double>(32) ); CALL_SUBTEST( test_scalar<long double>(32) );
|
||||
CALL_SUBTEST( test_scalar<float>(45) ); CALL_SUBTEST( test_scalar<double>(45) ); CALL_SUBTEST( test_scalar<long double>(45) );
|
||||
CALL_SUBTEST( test_scalar<float>(50) ); CALL_SUBTEST( test_scalar<double>(50) ); CALL_SUBTEST( test_scalar<long double>(50) );
|
||||
CALL_SUBTEST( test_scalar<float>(256) ); CALL_SUBTEST( test_scalar<double>(256) ); CALL_SUBTEST( test_scalar<long double>(256) );
|
||||
CALL_SUBTEST( test_scalar<float>(2*3*4*5*7) ); CALL_SUBTEST( test_scalar<double>(2*3*4*5*7) ); CALL_SUBTEST( test_scalar<long double>(2*3*4*5*7) );
|
||||
#endif
|
||||
|
||||
Loading…
x
Reference in New Issue
Block a user