A graph based approach for MPI deadlock detection

Tobias Hilbrich, Bronis R. De Supinski, Martin Schulz, Matthias S. Müller

Publikation: Beitrag in Buch/Bericht/KonferenzbandKonferenzbeitragBegutachtung

45 Zitate (Scopus)

Abstract

The MPI standard defines several usage patterns that can lead to deadlock, some of which involve collective communications or non-deterministic operations such as wildcard receives. Further, some MPI programming deadlocks only occur for some MPI implementations or certain configurations. Many tools to detect MPI deadlocks exist; however, none precisely handles the increased complexity of deadlock detection created by the richness of theMPI standard, which requires a general deadlock model. We present the first general deadlock model for MPI including a novel necessary and sufficient criterion, the OR-Knot, for deadlock in MPI programs. This model enables visualization of MPI deadlocks and motivates the design of a new deadlock detection mechanism. We compare our implementation of this mechanism to the ad-hoc mechanism previously available in Umpire, which reflected MPI non-determinism and, thus, more completely detected MPI deadlocks than any other existing MPI deadlock detection tool. Overall, our results demonstrate that our mechanism improves performance by as much as two orders of magnitude while providing precise characterization of deadlocks.

OriginalspracheEnglisch
TitelICS'09 - Proceedings of the 23rd International Conference on Supercomputing
Seiten296-305
Seitenumfang10
DOIs
PublikationsstatusVeröffentlicht - 2009
Extern publiziertJa
Veranstaltung23rd International Conference on Supercomputing, ICS'09 - Yorktown Heights, NY, USA/Vereinigte Staaten
Dauer: 8 Juni 200912 Juni 2009

Publikationsreihe

NameProceedings of the International Conference on Supercomputing

Konferenz

Konferenz23rd International Conference on Supercomputing, ICS'09
Land/GebietUSA/Vereinigte Staaten
OrtYorktown Heights, NY
Zeitraum8/06/0912/06/09

Fingerprint

Untersuchen Sie die Forschungsthemen von „A graph based approach for MPI deadlock detection“. Zusammen bilden sie einen einzigartigen Fingerprint.

Dieses zitieren