Multiprocessor Execution Profiler. (Reannouncement with New Availability Information).

摘要

Existing profiling tools generally have crude interfaces, are clumsy to use, and monitor only accumulated CPU time and function calls. After examining these programs, we concluded that they are insufficient aids for profiling a large scale multiprocessing system even if they are adequate for manual analysis of a single processor system. We developed a tool that collects the following information: where CPU time is expended, quality and quantity data passed between functions, how much global data is referenced and modified, and how these characteristics differ among processors on the network. To simplify data inspection the profiler has a mouse-driven graphical interface. We used the system on a number of single and multiprocessor Lisp programs. The profiler proved its usefulness in performance improvement and problem identification. This paper describes the design of the profiler and gives examples of its utilization.