Assembly Code Analysis Using Stochastic Process Algebra

Currently compilers contain a large number of optimisations which are based on a set of heuristics that are not guaranteed to be effective to improve the performance metrics. In this paper, we propose a strategy which allows us the analysis and the choice of the best optimisation, by focusing on the hot part of an assembly code. In our approach, for each optimisation applied, the code of the hot loop is extracted and its dependency graph generated. Finally, and in order to select the best optimisation, the generated graphs are analytically analysed using stochastic process algebra.

• Type: Conference
• Year: 2008
• Where: EPEW
• Authors: Lamia Djoudi, Leïla Kloul
• Keywords: EPEW 2008, Internet Technology, Large Number, Performance Metrics, Stochastic Process Algebra