Static Nonconcurrency Analysis of OpenMP Programs

Writing correct and eÆcient parallel programs is more diÆcult than doing so for sequential programs. One of the challenges comes from the nature of concurrent execution of a parallel program by di erent threads. Determining exact concurrency is NP-hard[10], and is impossible for real-world programs at compile time. OpenMP provides an easy and incremental way to write parallel programs. The well-structured OpenMP constructs and well-de ned semantics of OpenMP directives make compiler analyses more e ective on OpenMP programs than on more loosely structured parallel programs that are solely based on runtime libraries, such as MPI and Pthreads. In this paper, we present a static nonconcurrency analysis technique that detects, at compile time, whether two statements in an OpenMP program will not be executed concurrently by di erent threads in a team. Similar to the method presented in [5], ours is a close underestimation of the real nonconcurrency in a program. When our method determines that the executions of two statements are nonconcurrent, these two statements will not be executed concurrently. When the method fails, the two statements may, but need not, execute concurrently. Our nonconcurrency analysis models and uses the semantics of OpenMP directives. For example, in the following codes,