port_array <input, kPortsPerTile*kParallel_factor> in
The input data to the function. I/O is two parallel streams each TT_DATA type.
port_array <output, kOutputPorts> out
The output data from the function. I/O is two parallel streams each TT_DATA type.
kernel m_combInKernel [kParallel_factor]
FFT recursive decomposition. Widget kernel Widgets are used to reorder data as per FFT algorithm requirements.
kernel m_r2Comb [kParallel_factor]
FFT recursive decomposition. R2Combiner kernel R2combiner kernels connect 2 subframe processors with next stage’s r2combiner kernels
kernel m_combOutKernel [kParallel_factor]
FFT recursive decomposition. Widget kernel Widgets are used to reorder data as per FFT algorithm requirements.
fft_ifft_dit_1ch_graph <TT_DATA, TT_TWIDDLE, (TP_POINT_SIZE>> 1), TP_FFT_NIFFT, kFFTsubShift, TP_CASC_LEN, TP_DYN_PT_SIZE, (TP_WINDOW_VSIZE>> 1), TP_API, kNextParallelPower, TP_USE_WIDGETS, TP_RND, TP_SAT, TP_INDEX, TP_ORIG_PAR_POWER> FFTsubframe0
FFT recursive decomposition. Subframe0 FFT is split into 2 subframe processors with a stage of r2cominers connecting subframe processors. This is a recursive call with decrementing TP_PARALLEL_POWER template parameter.
fft_ifft_dit_1ch_graph <TT_DATA, TT_TWIDDLE, (TP_POINT_SIZE>> 1), TP_FFT_NIFFT, kFFTsubShift, TP_CASC_LEN, TP_DYN_PT_SIZE, (TP_WINDOW_VSIZE>> 1), TP_API, kNextParallelPower, TP_USE_WIDGETS, TP_RND, TP_SAT, TP_INDEX+kParallel_factor/2, TP_ORIG_PAR_POWER> FFTsubframe1
FFT recursive decomposition. Subframe1. FFT is split into 2 subframe processors with a stage of r2cominers connecting subframe processors. This is a recursive call with decrementing TP_PARALLEL_POWER template parameter.