DMCA
Compiler-Directed Instruction Stream Compression
BibTeX
@MISC{Greene_compiler-directedinstruction,
author = {David A. Greene and Charles R. Lefurgy and Trevor N. Mudge},
title = {Compiler-Directed Instruction Stream Compression},
year = {}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
Embedded systems are becoming increasingly important, appearing in every aspect of daily life. The cost of these systems is strongly influenced by the cost of the processor chip. The chip cost, in turn, is dominated by the die size. By compressing embedded software the space devoted to instruction memory can be reduced, lowering costs. Alternatively, more fully-featured software can be provided without increasing costs. We present a compiler-controlled instruction stream compression method. Our implementation uses similar code templates to create, in essence, lightweight functions. These functions may have arguments which fill the gaps in the template code. While our initial compression results are not terribly successful, we attribute this to deficiencies in the current compiler implementation, in both compression and non-compression related areas. This paper explores our results and presents suggestions for compiler improvements to achieve better compression ratios. 1
Keyphrases
compiler-directed instruction stream compression initial compression result lightweight function fully-featured software compiler improvement chip cost compiler-controlled instruction stream compression method non-compression related area daily life processor chip similar code template compression ratio die size present suggestion current compiler implementation template code
