Tools and methods to support the correctness of parallel programs

Prof. Dr. Matthias S. Müller, RWTH Aachen University

10 Jul 2017, 16:30–18:00; Location: S2|02-C110

Parallel programming for large scale computer architectures is a complex and error-prone task. Automatic tools can assist application developers in the detection and removal of such errors. This talk will present the scalable runtime tool MUST, the methods behind the tool (graph analysis, push down automaton) and the limitations that result from the chosen approach. Initially MUST was designed for programs using the Message Passing Interface (MPI). Motivated by the increasing number of hybrid programs mixing MPI with OpenMP the tool was extended to support hybrid programs. One key factor is the analysis of the memory access in conjunction with the synchronization enforced by the programming standard used.  The talk will show how MUST detects and analyzes deadlocks and data races in large scale parallel programs.

Category: CE Seminar


Technische Universität Darmstadt

Graduate School CE
Dolivostraße 15
D-64293 Darmstadt

Phone+49 6151/16-24401
Fax -24404

to assistants' office

Open BSc/MSc Theses

Show a list of open BSc/MSc topics at GSC CE.

 Print |  Impressum |  Sitemap |  Search |  Contact |  Privacy Policy
zum Seitenanfangzum Seitenanfang