gcp_fft_1d_r2.h File Reference

#include <gcell/gc_job_manager.h>
#include <complex>

Include dependency graph for gcp_fft_1d_r2.h:

Go to the source code of this file.

Functions
gc_job_desc_sptr	gcp_fft_1d_r2_submit (gc_job_manager_sptr mgr, unsigned int log2_fft_length, bool forward, bool shift, std::complex< float > *out, const std::complex< float > *in, const std::complex< float > *twiddle, const float *window)
	Submit a job that computes the forward or inverse FFT.
void	gcp_fft_1d_r2_twiddle (unsigned int log2_fft_length, std::complex< float > *W)
	Compute twiddle factors.

---

Function Documentation

gc_job_desc_sptr gcp_fft_1d_r2_submit	(	gc_job_manager_sptr	mgr,
		unsigned int	log2_fft_length,
		bool	forward,
		bool	shift,
		std::complex< float > *	out,
		const std::complex< float > *	in,
		const std::complex< float > *	twiddle,
		const float *	window
	)

Submit a job that computes the forward or inverse FFT.

Parameters:

	mgr	is the job manager instance
	log2_fft_length	is the log2 of the fft_length (4 <= x <= 12).
	forward	is true to compute the forward transform, else the inverse.
	shift	indicates if an "fftshift" should be applied to the output data
	out	is the fft_length output from FFT (must be 16-byte aligned).
	in	is the fft_length input to FFT (must be 16-byte aligned).
	twiddle	is fft_length/4 twiddle factor input to FFT (must be 16-byte aligned).
	window	is the window to be applied to the input data. The window length must be either 0 or fft_length (must be 16-byte aligned).

Returns a shared_ptr to a job descriptor which should be passed to wait_job*. Throws an exception in the event of a problem. This uses the FFTW conventions for scaling. That is, neither the forward nor inverse are scaled by 1/fft_length.

void gcp_fft_1d_r2_twiddle	(	unsigned int	log2_fft_length,
		std::complex< float > *	W
	)

Compute twiddle factors.

Parameters:

	log2_fft_length	is the log2 of the fft_length.
	W	is fft_length/4 twiddle factor output (must be 16-byte aligned).