Opportunities to implement parallelism exist at several levels within embedded devices. Consequently, there are multiple ways that it can be exploited—primarily to improve system performance. While some techniques are used successfully today in desktop and embedded architectures, many solutions offered constrain the designer to a fixed-function solution. Here we explain the key terminology around multiprocessing, and suggest a more flexible and generalized approach to adopting parallel technology that can provide benefits more appropriate for embedded applications applications—processing efficiency—namely scalable performance and low power.

Reprinted in its entirety from ARM IQ Vol. 2, No. 2