> > > hi tom, > > for a 1400 MHz bandwidth single pol, you need > one roach and one 3 Gsps ADC board for pulsar/transient > search machine and kurtosis spectrometer. > > for coherent dedispersion pulsar timing and for seti, > you could use the same system to break the 1400 MHz > band into sub-bands and send the sub-bands to a bank of GPU/CPU units. > each GPU/CPU can handle 200 MHz bandwidth, single pol. > if you have more than 4 CPU/GPU units, then you need a 10Gbe switch, > because roach only has four 10Gbit ports. > > if you want 8 GPU/CPU units, to handle 1600 MHz single pol, > then it might be slightly cheaper to split the band into two 800 MHz > chunks, > and use a pair of roaches, rather than purchasing a 12 port switch > and a single roach and a 10Gbe switch.
I think you need to be careful about total bandwidth available. One of the strengths of GUPPI is its 8 bit dynamic range. That means that for 2 pols, 800 MHz bandwidth, you need ~25 gigabits per second, or slightly more than 2 10 GbE ports. But at 1400 MHz, that'll be 2800 GS/s * 2 pols *8 bits/pol, or 41,600 Gbits/second for an 8 bit dual pol system. Obviously if you reduce the number of bits you can stuff it all into 4 10 GbE links, but you really need to consider your RFI environment. All that said, I'd recommend that you spring for the 10 Gb switch if you try to build one of these. It makes things easier in so many ways. Troubleshooting, testing, etc. is easier without point-to-point links that you have to connect and disconnect to reroute stuff. You can get rid of it in the end when you install the system if there are space or power constraints. John > > dan > > > > On 1/29/2010 4:17 PM, Tom Kuiper wrote: >> Dan Werthimer wrote: >>> each GPU can handle 100 to 200 MHz dual pol depending on whether >>> you are doing coherent dedispersion (timing), or spectroscopy >>> (searching). >>> matthew and jonathan are the experts at reading data from ibob/roach >>> and >>> using CPU cluster to do pulsar/transient search. >>> john ford, paul demorest, scott ransom et al are the experts at using >>> ibob/bee2 >>> to packetize data (800 MHz dual pol) for GPU based pulsar cluster >>> (see their fantastic GUPPI instrument). >> We could have up to 1400 MHz at once, 8200-8600 and 31,500-32,500 MHz >> but I think only one polarization. I saw that John Ford is using 8 >> GPUs for 800 MHz. Can you get several GPUs on the single bus of a >> multi-core host or does that cause too much of a bottle-neck? I also >> should think about doing the various piggy-back tasks in parallel. >> I'm guessing that setispec on a ROACH is a tight fit. How about two? >> The kurtosis is a very light task, I think, so can some of the >> left-over resources be used to expand the SETI bandwidth or refine the >> resolution? >> >> Anyway, for now it's some high-level wishing so I'll scope one unit at >> three dual-channel ADCs, three ROACHes, two 4 core hosts, and 8 GPUs. >> Does that seem reasonable? About $40K? (We have to pay Xilinx :-( .) >> >> Thanks for your help >> >> Tom >> > >
