We describe an innovative, low-power, high-performance, programmable signal processor (DSP) for digital communications. The architecture of this processor is characterized by its explicit design for low-power implementations, its innovative ability to jointly exploit instruction-level parallelism and data-level parallelism to achieve high performance, its suitability as a target for an optimizing high-level language compiler, and its explicit replacement of hardware resources by compile-time practices. We describe the methodology used in the development of the processor, highlighting the techniques deployed to enable application/architecture/compiler/implementation co-development, and the optimization approach and metric used for power-performance evaluation and tradeoff analysis. We summarize the salient features of the architecture, provide a brief description of the hardware organization, and discuss the compiler techniques used to exercise these features. We also summarize the simulation environment and associated software development tools. Coding examples from two representative kernels in the digital communications domain are also provided. The resulting methodology, architecture, and compiler represent an advance of the state of the art in the area of low-power, domain-specific microprocessors.