The Berkeley Parkes Swinburne Recorder (BPSR) is a high‑resolution digital filterbank data acquitision and processing system for the Parkes 21‑cm Multibeam receiver, developed in collaboration between Swinburne California at Berkeley. The BPSR instrument is currently hosted on the HIPSR Digital Signal Processor, which consists of 13 Reconfigurable Open Architecture Computing Hardware (ROACH) processing boards developed by the Collaboration Research (CASPER). Each ROACH board performs analog‑to‑digital conversion of a single dual‑polarization signal with a bandwidth of 400 MHz. The signals are sub‑divided into 1024 frequency channels using a polyphase Programmable Gate Array (FPGA) logic blocks. The data are then detected, integrated, decimated to 8 bits, and streamed to 8 server‑class nodes, 7 of which are directly connected to two ROACH boards via 10 GbE CX4.
The server‑class nodes were built to specication by Silicon Graphics Pty. Ltd. Each of the 8 workstations have dual 2.66 GHz Intel Xeon six‑core CPUs, 48 GB DDR3 memory, and dual Nvidia Tesla C2070 general‑purpose RAM, the data are routed, managed, and monitored using PSRDADA, then summed, normalized, decimated to 2 bits per sample, and written to disk in real‑time using DSPSR. These two Open Source software projects pulsar research groups around the world.
Compared to the previous analog filterbank system at Parkes, the major technical innovations of BPSR include orders of magnitude increases in time resolution (32 vs 125 microseconds), frequency resolution (1024 remote control and administration capability; and professionally engineered acquisition and analysis software, developed under an Open Source license as part of Swinburne's continuing dedication to foster international BPSR is an integral part of the High Time Resolution Universe Survey, currently underway at Parkes. HIPSR was funded by the ARC (LE110100212).
- The Parkes Pulsar Backends - Conference Proceedings - requires subscription
User Contributed Tags
Login to tag this record with meaningful keywords to make it easier to discover