Efficient thread labeling for on-the-fly detection of races in OpenMP programs

Abstract

It is important to detect data races in OpenMP programs, because they may lead to unpredictable results from an execution of the programs. To detect data races during an execution of program that may have nested parallelism, it is important to maintain thread information called label which is used to determine the logical concurrency between threads. This paper presents an efficient thread labeling, called eNR labeling, which creates thread labels on every thread operation in a constant amount of time and space even in its worst case. Compared with three state-of-the-arts labeling schemes, our empirical results using OpenMP benchmarks show that eNR labeling reduces both time overhead by 30% and space overhead by more than 90% for on-the-fly data race detection. ? 2013 International Information Institute.