| The complexity and computational intensity of scientific computing has fueled research on parallel computing and performance analysis. The purpose of this thesis is to critically investigate the state of the art in performance analysis for scientific computing and then propose and demonstrate through implementation, the feasibility of a novel approach to performance analysis and experiment design. The core of this approach is an Instrumentation Database (IDB) that enables comparative analysis of parallel code performance across architectures and algorithms.; The basis of the IDB approach is scalable collection of performance data so that problem size and run-time environments do not affect the amount of information collected. This is achieved by uncoupling performance data from the underlying architecture and associating it with the control flow graph of the program. We consider the subset of nodes that are critical to performance: procedures, loops, calls, and communications/synchronization events. The resulting structure, a control flow hierarchy, is comprised of these nodes. Each node has statistical data collected for it; namely, a minimum, maximum, average, and standard deviation of its execution time, along with the number of times it was executed. Since these values are of fixed size, and the number of nodes is not a function of either problem size, system architecture or execution environments, data scalability is ensured. Another important contribution of the IDB approach is the use of database technology to map program structure onto relational schema that represent the control flow hierarchy, its corresponding statistical data, and static information that describes the execution environment.; To demonstrate the benefits of the proposed approach, we have implemented a POSIX compliant probe library, automated instrumentation tool, front-end visualization programs, database schema using an object-relational DBMS (PostgreSQL), and SQL queries. We also developed a methodology, based on these tools, for interactive performance analysis which we demonstrated by analyzing several different parallel scientific applications. |
