摘要：A major challenge of applying profile-based optimization on large real-world applications ishow to capture adequate profile information. A large program, especially a GUI-based application,may be used in a large variety of ways by different users on different machines. Extensivecollection of profile data is necessary to fully characterize this type of program behavior.Unfortunately, in a realistic software production environment, many developers and testers needfast access to the latest build, leaving little time for collecting profiles. To address this dilemma,we would like to re-use stale profile information from a prior program build. In this paper wepresent BMAT, a fast and effective tool that matches two versions of a binary program withoutknowledge of source code changes. BMAT enables the propagation of profile information from anolder, extensively profiled build to a newer build, thus greatly reducing or even eliminating theneed for re-profiling. We use two metrics to evaluate the quality of the results using propagatedprofile information: static branch prediction and the accuracy of code coverage. These metricsmeasure how well the matching algorithm works for the frequently executed core code and acrossthe whole program, respectively. Experiments on a set of large DLLs from Microsoft Windows2000 and Internet Explorer show that compared to freshly collected profiles, propagatedinformation using BMAT is typically over 99% as effective in branch prediction and over 98% asaccurate in code coverage information
