Achtung:

Sie haben Javascript deaktiviert!
Sie haben versucht eine Funktion zu nutzen, die nur mit Javascript möglich ist. Um sämtliche Funktionalitäten unserer Internetseite zu nutzen, aktivieren Sie bitte Javascript in Ihrem Browser.

Info-Icon Diese Seite ist nicht in Deutsch verfügbar
Data Center Building O Bildinformationen anzeigen

Data Center Building O

Research Areas

The development of computing systems with high energy-efficiency is currently the prime challenge in computing systems research striving to overcome the stagnating progress in performance and efficiency for individual CPU cores.

Research of the core research team at PC²

The core research team of PC² lead by Prof. Plessl is currently investigating three approaches to address this challenge: 1. The use of novel, massively parallel and customizable reconfigurable accelerators; 2. a more intelligent and adaptive use of all available computing resources in heterogeneous computing systems; and 3. the use of alternative computational models.

1. Custom Computing with FPGAs

Field-programable gate arrays (FPGAs) are massively parallel, software programmable hardware devices that can be tailored to implement highly customized application-specific accelerators. These accelerators are able to substantially outperform conventional CPU implementations in terms of performance and energy for many applications. Our research in this area focuses on developing high-level programming methods and domain-specific compilation approaches to simplify the programming and use of FPGAs. We further collaborate with researchers in computational science and engineering to evaluate the merits of this technology in practical applications.

This topic is or has been addressed in these projects:

2. Runtime-systems for Heterogeneous Computing Systems

The concurrent use of different processor elements has become widespread across the computer system spectrum ranging from embedded over general-purpose up to high-performance computing systems. While these heterogeneous computer systems offer significant efficiency benefits by dispatching computationally demanding application parts to the most suitable computing resource at runtime, the execution models and runtime systems have so far remained rather static. That is, application partitioning decisions are performed at compile time rather than runtime. In our research we study the opportunities for moving application mapping and scheduling decisions to runtime in order to provide benefits at the application level (performance, latency) and system level (energy consumption, resource utilization).

This topic is or has been addressed in these projects:

3. Emerging Computing Paradigms and Technologies

Meeting the ambitious targets for energy-efficiency for exascale computer systems and beyond, has spurred an increasing interest in unconventional computing paradigms, covering all layers from the computational elements and fabric up to the application level. In our research, we study Approximate Computing, which is one of the most promising yet practical emerging computing approaches. The key objective of our work is studying how approximate computing can be used for scientific computing, where a defined quality of results is required.

Collaborative research with other groups

In addition, PC² is also involved in collaborative research projects with other research groups at Paderborn University, in particular with the groups represented in the management board of PC². On the one hand, we perform foundational and applied computing systems research with groups from the computer architecture domain. On the other hand, we performa technology evaluation projects to study the practical suitability of emerging hard and software technologies for the scientific applications in the main research domains of our users.

Die Universität der Informationsgesellschaft