[Pafgbt] PAF beamformer size and cost

[John Ford]

> This is an incomplete question, but maybe we can beat it into something
> answerable:  Do we know enough about existing applications on CASPER
> hardware to make a conservative estimate of what it would cost to build a
> PAF beamformer with a given set of specs?  I'm looking for at least two
> estimates.  What is a realistic set of specs for the first science PAF
> beamformer, and what would the dream machine that would make a big
> scientific impact cost?  You're welcome to define the specs that go with
> either of these two questions or I'll start defining them by thinking "out
> loud".  The first science beamformer will guide the initial system design,
> and the dream machine will help get a handle on longer range expectations.

There are quite a few beamformer designs underway in the CASPER world. 
Check this one out: http://arxiv.org/abs/0912.0380v1

I can take a crack at a block diagram and real cost estimate once we iron
out some basic requirements/desires.

Some strawman specs I (sort of) got from Rick's paper the other day:

Analog bandwidth: =< 500 MHz (lets us use 4 channels per ibob/roach using
the first generation iADC at a <1 GHz sample rate)  These have enough
analog bandwidth to digitize at RF if desired.

1024 frequency channels in the front-end channelizer (enough?) (how many
taps in the pfb?)

7 simultaneous beams

38 dual pol elements in the feed array

So for starters, we'd need:

38 iadc boards  (~2K each)

(38*2)/4 = 19 roach boards for the digitizing.  ~5K each

We'd need a 10 Gb Ethernet switch with at least 28 ports on it as the
signal distribution system. ~15K

Some roach boards for beamforming.  Say one for each beam.  I think that's
excessive, as we can probably form multiple beams on each roach.

so the ballpark hardware cost for the DSP stuff is looking like

38 ADC boards:     95k
26 roach boards:   130K
1 Ethernet Switch: 15K
1 host:            20K
1 lot cables, power supplies, etc.
                   15K

                   ~275K  Call it 300K for good measure.

We could also build some ADC modules that have more channels on each
board, reducing the number of processing boards we need as well as the
number of ADC boards.

Maybe we can start with this and talk about whether this is a Cadillac or
a Yugo.  Then we can dig in further!

John

>
> Cheers,
> Rick
>
> _______________________________________________
> Pafgbt mailing list
> Pafgbt at listmgr.cv.nrao.edu
> http://listmgr.cv.nrao.edu/mailman/listinfo/pafgbt
>