BMAT -- A Binary Matching Tool for Stale Profile Propagation

摘要: A major challenge of applying profile-based optimization on large real-world applications is how to capture adequate profile information. program, especially a GUI-based application, may be used in variety ways by different users machines. Extensive collection data necessary fully characterize this type program behavior. Unfortunately, realistic software production environment, many developers and testers need fast access the latest build, leaving little time for collecting profiles. To address dilemma, we would like re-use stale information from prior build. In paper present BMAT, effective tool that matches two versions binary without knowledge source code changes. BMAT enables propagation an older, extensively profiled build newer thus greatly reducing or even eliminating re-profiling. We use metrics evaluate quality results using propagated information: static branch prediction accuracy coverage. These measure well matching algorithm works frequently executed core across whole respectively. Experiments set DLLs Microsoft Windows 2000 Internet Explorer show compared freshly collected profiles, typically over 99% as 98% accurate coverage