David Pointer at SRC Computer on the use of FPGAs for Filtered Backprojection (FBP), a critical component of medical imaging and other domains:
Monthly Archives: April 2008
Interesting news story out of the University of Arizona:
The article describes how FPGAs can be deployed for critical space applications, in which hardware/software flexibility must be balanced against performance requirements. According to Dr. Ali Akoglu, FPGAs are compelling because of their inherent parallelism and because of their potential for in-system, dynamic reconfiguration:
“So if you’re running a loop, and it is running 10,000 times, you can replicate the loop as a processing element in the FPGA ‘n’ number of times,” Akoglu explained. “That means you have an ‘n’ times speed-up.”
To solve issues of reliability, the system being developed at UA involves five or more independent, redundant processing units. If one fails, the system can be reconfigured, or “self-heal” by reconfiguring the system around the bad circuitry. If that node-level repair fails to resolve the problem, then the redundant circuits are smart enough to take over.
This project suggests where FPGAs are heading as partial reconfiguration becomes more practical and better supported in the tool flows. Partial reconfiguration (reprogramming only a part of the FPGA) can be used both to save FPGA space (you only need to “swap in” those functions you need at any given time) but also to improve reliability through intelligent reconfiguration when things go bad.
Here’s a cool article by one of our users, Paula Pingree at Jet Propulsion Laboratories:
(Gosh, this stuff really is rocket science.)