Re: [casper] how long can filter banks be?

[Andrew Martens]

Hi Gerry

If you put a real signal through an FFT you get a mirror image of one 
half of the channels in the other half. Our fft_wideband_real (which I 
am assuming that you will be using), removes half the channels 
automatically. So a 2^17 bin fft_wideband_real block will give you 2^16 
channels.

Regards
Andrew

Thanks Andrew.

I suppose you suggest 2^16 useable channels from a 2^17 FFT b/c of
aliasing?

Yes, we must give it a try to really know.

Gerry

On 1/20/2014 9:55 PM, Andrew Martens wrote:
Hi Gerry

Dan's idea of ganging together filter banks works well provided you can
make do with an FFT for the second filter bank i.e you don't need any of
the benefits of a PFB.

We currently are working on a design that includes 2 2^16 filter banks
(10 bits ADC input data, 8 tap PFB FIR, 18 bit data path) capable of
processing 880MHz each in a ROACH2. You should be able to comfortably
fit a 2^17 filter bank (2^16 effective channels) using a single FFT into
a ROACH2. 2^17 effective channels will be more difficult but might be
possible I think if you kept the number of taps in the pfb_fir low and
did the reordering of channels post-FFT in software. Best would be to
take a tutorial, change the parameters and see.

Regards
Andrew

On 21/01/2014 07:13, Dan Werthimer wrote:

hi gerry,


we haven't tried this, but i think the largest spectrometer you
could fit on a  roach2  is 256M points, implemented by a 16K point FFT,
followed by DRAM based corner turn and twiddle factors,
followed by another 16K point FFT.

if you have this many channels in your correlator,
you also be running up near the correlator X engine memory limits:

for instance, if you cross correlate in a Titan GPU, then you only have
5 or 6 GB of memory on each GPU card.

let's assume you have a max of 32 GPU's for your X engine.

then max frequency channels =

32 GPU's   x   6GB/GPU  x  42^2 baselinepols  x  4B/baseline

= 435M channels max for 32 GPU's  (round down to 256M max channels)


if you cross correlate in a CPU (eg: DiFX) then you can have more
memory,
but you'll need a lot more CPU's to keep up with the data rate, so CPU's
won't help.

be wary of readout rate too - that's a lot of data to read out :

256M channels  x  42^2 baselinepols  x  4B =  1 TB  every integration
time



best wishes,

dan



On Mon, Jan 20, 2014 at 7:53 PM, Gerry Harp <gh...@seti.org
<mailto:gh...@seti.org>> wrote:

     Hi

     Just for fun, how large of an FFT (filter bank) can fit into one of
     the Roach# boards? Has anyone ever successfully compiled a filter
     bank with length 2^17? We're interested in building a relatively
     narrow-band correlator so we need lots of channels. Any experience
     at large lengths or educated guesses are welcome. Also, how fast
did
     it go? Possible to keep up with 100 MSPS?

     It is proposal time, once more...

     Thanks

     Gerry Harp


     On 1/17/2014 11:56 AM, Dan Werthimer wrote:





                            Dear Casper Collaborators,


         We hope you can attend this year's Casper Worshop

                                in Berkeley, California

                          June 9 throuh June 13, 2014




         We'll have more information later about registration,
         travel, abstracts, etc, but for now, please reserve these
dates.


         Hoping you can participate,


         Dan and the Scientific and Local Organizing Committees







     --
     ----------------------
     Gerald R. Harp, Ph.D.
     Director, Center for SETI Research
     SETI Institute