Instruction Selection Based on Object Code Sizes
Sikun Li
Abstract
Sikun Li
Abstract
Instruction selection of traditional compilers often adopts the dynamic programming technology to optimize the object code performance (reducing code running time). Because of the cost constraint, embedded systems always have very limited memory resources, requiring code sizes as small as possible. This paper provides an instruction selection method based on object code sizes for a type of processors which have variable instruction sizes, using dynamic programming. And relative to the compilers for performance, it can reduce the code size from 15% to 20%.
A significance statement is not available in the OpenAlex record.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
Instruction selection of traditional compilers often adopts the dynamic programming technology to optimize the object code performance (reducing code running time). Because of the cost constraint, embedded systems always have very limited memory resources, requiring code sizes as small as possible. This paper provides an instruction selection method based on object code sizes for a type of processors which have variable instruction sizes, using dynamic programming. And relative to the compilers for performance, it can reduce the code size from 15% to 20%.
Key concepts: Computer science, Compiler, Object code, Parallel computing, Code (set theory), Programming language, Selection (genetic algorithm), Object (grammar)