Preprocessing scheme of intelligent assembly for a high performance VLIW DSP

The time and energy needed for high-performance VLIW digital signal processors (DSPs) is dependent on the execution efficiency of the codes running on them. YHFT-Matrix DSP is a kind of high performance floating-point DSP for wireless communication and video image processing. It has strong ability to handle vector data and has abundant instruction set. To provide efficient codes for it, and meanwhile to ease the workload for programmers and to reduce programming difficulty, we use the intelligent assembly scheme, which combines the advantages of high-level language and assembly language. In this paper, the preprocessing for intelligent assembly is presented and the algorithm for lexical analysis and that for macro processing are investigated. For the former, a one time scan and step-by-step processing method is used to recognize kinds of lexical elements; for the latter, an algorithm based on a file relation graph is developed to settle macrosubstitution. The preprocessing results based on these algorithms can provide global information and information needed for subsequent process.