San Jose, Calif. -- For years, systems designers in high-end communications have faced a series of tough and complex chip choices among ASSPs, ASICs, DSPs, FPGAs and other devices.
While a number of cost and performance trade-offs are associated with those chip technologies, FPGA vendors have long argued that their respective devices have several advantages over complex and costly ASICs--faster time-to-market and more design flexibility, for example. However, FPGA cost/performance metrics have generally been perceived as weak compared with digital signal processors--until now.
FPGAs may now have a cost/performance advantage over standalone DSPs in a "significant class of DSP applications," according to a new and controversial benchmarking study from Berkeley Design Technology Inc. (BDTI).
In particular, FPGAs are better-suited than standalone DSPs "in performance-hungry applications with a lot of parallelism," such as high-end communications infrastructure equipment, said Jeff Bier, founder and president of BDTI (Berkeley, Calif.), a technology consultancy. "FPGAs will play an increasing role in DSP applications," Bier said.
FPGAs will not take over the world overnight, according to the report, which is both detailed and nuanced. Bier takes pains to stress that for many high-end telecom and other applications, FPGAs will continue to be used alongside existing devices in systems, such as application-specific standard products (ASSPs), ASICs, DSPs, general-purpose processors and other chips. But based on BDTI's study, there are also signs that FPGAs, traditionally associated with high cost, are actually less expensive than a DSP solution for certain applications in a design.