A floating point DSP with optimizing C compiler

Abstract

A new digital signal processor (the DSP32C) has been developed which performs 32-bit floating point operations at a rate of 25 MFLOPs and can be programmed using a standard C compiler. This paper presents an overview of the architecture and instruction set with emphasis on the enhancements over its predecessor, the DSP32. The perfommce is expressed with common signal processor benchmarks. The application development environment is described which highlights the C compiler and hardware development system. Finally, an application example iuustrates the power and ease-of-use of this DSP.