Hi Randy, I think the issue is that you are processing 16 samples per FPGA clock for VEGAS (if I remember right), so I expect you'll need to hand craft an FFT out of the FFT direct block and twiddle blocks. It might be easier to use a xilinx block in this case, but I haven't actually looked at what's available. Glenn
On Mon, Jul 18, 2016 at 1:04 PM, Randy McCullough <rmccu...@nrao.edu> wrote: > All, > > We're currently in the process of adding incoherent and coherent > de-dispersion > pulsar modes to our ROACH 2-based VEGAS backend and have hit a stumbling > block. We've built all designs ranging from 8,192 channels down to 64 > channels, > but when attempting to implement a 32-channel design the standard Wideband > Real FFT block breaks. Has anyone successfully implemented a 32 channel > PFB/FFT > pair on a ROACH 2 platform? > > Randy > >