A low power DSP engine for wireless communications

This paper describes the architecture and the performance of a new programmable 16-bit digital signal processor (DSP) engine. It is developed specifically for next generation wireless digital systems and speech applications. Besides providing a basic instruction set, similar to current day 16-bit DSPs, it contains extra architectural features and unique instructions, which make the engine highly efficient for compute-intensive tasks such as vector quantization and Viterbi operations. The datapath contains two multiply-accumulate units and one ALU. The external memory bandwidth is kept to two data buses and two corresponding address buses. Still, the internal bus network is designed such that all three units are operating in parallel. This parallelism is reflected in the performance benchmarks. For example, an FIR filter of N taps will take N/2 instruction cycles compared to N for a general purpose 16-bit DSP, and it requires only half the number of memory accesses of a general purpose DSP.