Neue Rechnerarchitekturen

Das Institut forscht an neuartigen FPGA-basierten Architekturen für spezifische rechenintensive Anwendungen sowie an netzwerkartig verbundenen FPGA- und Prozessorstrukturen, die als Basis für Parallelrechnerarchitekturen dienen.

Parallelrechner ER-1 (1994)Parallelrechner ER-2 (1998), Parallelrechner ER-3 (2011), Parallelrechner ER-4 (2014)

Algebraic Engineering Group

Linear Codes
are studied as polynomial ideals using Gröbner bases. This allows to apply methods from algebraic geometry and commutative algebra. Aims are to develop invariants and to find global and local structure (Mahwish, Dück, Leppert).

Swarm-Based Computing
Simple software agents interacting locally with each other or with the environment can comprise a population that behaves (almost) like a biological system. Goals are to employ artificial swarms to efficiently solve scientific problems, like text understanding (Sallam).

General-Purpose Graphics Card Processing
GPUs are capable to perform general-purpose computations that are traditionally treated by CPUs. Our aim is to parallelize computationally intense algorithms for GPUs mostly from graph theory and bioinformatics (Muhammad, Srinivasa).

FPGA Based Scientific Computing
FPGAs received a major boost during the last years and are amenable to perform greedy data-driven computations. Our objective is to design algorithms for FPGAs mostly from DNA computing and bioinformatics (Brandt).

Symbolisch-numerisches Rechnen

wird verwendet, um die Bestimmung exakter mathematischer Prädikate zu beschleunigen. Die algebraische Beschreibung isolierter Situationen wird in die kontinuierliche Gesamtlage eingebettet, um eine stabile Grundlage für ingenieurliche Konstruktions- und Entwurfsverfahren zu bilden. Wir forschen an schnellen, nie versagenden Verfahren.