Accelerating Data Race Detection with Minimal Hardware Support

EuroPar 2011 (International Conference on Parallel and Distributed Computing) |

Published by Springer

We propose a high performance hybrid hardware/software solution to race detection that uses minimal hardware support. This hardware extension consists of a single extra instruction, StateChk, that simply returns the coherence state of a cache block without requiring any complex traps to handlers. To leverage this support, we propose a new algorithm for race detection. This detection algorithm uses StateChk to eliminate many expensive operations. We also propose a new execution schedule manipulation heuristic to achieve high coverage rapidly. This approach is capable of detecting virtually all data races detected by a traditional happened-before data race detection approach, but at significantly lower space and performance overhead.