DSP, for many engineers in the field, is a proprietary device used to implement digital-signal processing functions. These conventional pieces of silicon have been used for years to quickly and efficiently digitize analog signals, arithmetically alter that digital information in a desired way, then convert the digital back into something sensory that we humans can typically see, hear, or feel. These digital signal processors, or DSPs, are found throughout the world of electronic equipment, from cell phones to flat-screen TVs.

Dedicated DSP processors and cores are still a widely popular means of achieving digital signal processing needs. In little more than a few days time, a programmer can take a DSP chip and write an algorithm that will efficiently perform the digital signal processing task required. But as CPUs have incorporated arithmetic coprocessors and extensions that optimize digital signal processing tasks, and as other silicon alternatives have emerged, the necessity of a separate DSP chip is no longer viewed as inevitable. In fact, there seems to be increasing evidence that the negatives of discrete DSP chips might outweigh their benefits.

First, there is the issue of fixed costs. All DSP chips are viewed as IP (intellectual property), and as such, have per-use costs associated with them.While such considerations aren’t much of an impediment to a company producing a limited number of end products, many digital signal processing needs occur in high-volume products where steep discounting of retail prices over a short period of time requires the lowest per-unit cost possible.

Next, there is the issue of performance. Compared to either ASIC or FPGA hardware solutions, a DSP chip is by far the slowest option. ASIC chips run as fast as the technology allows. Similarly, an FPGA that has been optimized to perform a digital-signal-processing task, will run anywhere from 10 times to more than 1000 times faster than a single DSP chip.Whereas a DSP processor typically employs serial processing, the parallel capacities inherent to either ASIC or FPGA architecture will always give them both a significant edge over DSPs.