Software

The Laboratory for Parallel Programming creates practical solutions for parallel computing. Specifically, we develop programming tools and software infrastructures that help their users effectively harness the power of massively parallel architectures.

Extra-P

Extra-P is an automatic performance-modeling tool that allows the quick identification of scalability bottlenecks in complex HPC applications. It uses performance measurements to generate human-readable models for the behavior of each function in an application, thus providing insights into present and future scalability issues and their causes. Extra-P is available under an open-source license as part of the Scalasca toolset.

More information and download: project website
Contact: Alexandru Calotoiu

Dynamic Torque/Maui Batch System

A batch system is an important middleware for managing supercomputing resources. Traditional batch systems only support the static allocation of resources to parallel jobs. Our laboratory develops an enhanced version of the widely used Torque/Maui batch system with support for dynamic resource allocation. Dynamic resource allocation helps improve system utilization, throughput, and fault tolerance.

Contact: Suraj Prabhakaran

DiscoPoP

Writing efficient parallel code is still an art mastered only by experienced programmers. In particular, the parallelization of the myriads of sequential legacy programs that exist today presents a major technical and economic challenge. To ease their parallelization, we develop DiscoPoP, a tool to automatically identify potential parallelism in sequential code and discover possible parallel design patterns.

Contact: Dr. Ali Jannesari

LWM2

LWM2 is a lightweight monitoring module that serves as a first-stop tool to quickly capture the performance of an application. Besides profiling individual applications, it also looks at external factors affecting application performance. Capabilities like monitoring I/O performance at the system level and, on torus networks, capturing network traffic at the node level allow such endeavors.

Contact: Aamer Shah

 

 

Contact

Technische Universität Darmstadt
Department of Computer Science

Laboratory for
Parallel Programming

Office: S4|14 3.1.11
Mornewegstraße 30
64293 Darmstadt

Phone:+49 6151 16-25626
Fax:+49 6151 16-23050
Email:wolf@cs.tu-darmstadt.de
A A A | Drucken Print | Impressum Impressum | Sitemap Sitemap | Suche Search | Kontakt Contact | Webseitenanalyse: Mehr Informationen
zum Seitenanfangzum Seitenanfang