MPI runtime error detection with MUST: Advances in deadlock detection

Issue title: Selected Papers from Super Computing 2012

Article type: Research Article

Authors: Hilbrich, Tobias^; | Protze, Joachim^{; ;} | Schulz, Martin | de Supinski, Bronis R. | Müller, Matthias S.^{; ;}

Affiliations: Technische Universität Dresden, Dresden, Germany. E-mail: tobias.hilbrich@tu-dresden.de | Lawrence Livermore National Laboratory, Livermore, CA, USA. E-mails: {bronis, schulzm}@llnl.gov | RWTH Aachen University, Aachen, Germany | JARA – High Performance Computing, Aachen, Germany. E-mails: {mueller, protze}@rz.rwth-aachen.de

Note: [] Corresponding author: Tobias Hilbrich, Technische Universität Dresden, D-01062 Dresden, Germany. E-mail: tobias.hilbrich@tu-dresden.de

Abstract: The widely used Message Passing Interface (MPI) is complex and rich. As a result, application developers require automated tools to avoid and to detect MPI programming errors. We present the Marmot Umpire Scalable Tool (MUST) that detects such errors with significantly increased scalability. We present improvements to our graph-based deadlock detection approach for MPI, which cover future MPI extensions. Our enhancements also check complex MPI constructs that no previous graph-based detection approach handled correctly. Finally, we present optimizations for the processing of MPI operations that reduce runtime deadlock detection overheads. Existing approaches often require 𝒪(p) analysis time per MPI operation, for p processes. We empirically observe that our improvements lead to sub-linear or better analysis time per operation for a wide range of real world applications.

Keywords: Deadlock detection, message passing interface, correctness checking

DOI: 10.3233/SPR-130368

Journal: Scientific Programming, vol. 21, no. 3-4, pp. 109-121, 2013