You are here:

Home Research Resource-constrained Computing Ressource-aware Parallel Computing for Biosensors



Ressource-aware Parallel Computing for Biosensors

Rapid Protopying of Multicore Image Processing Systems

Designing multicore systems in the field of image processing is tedious because the optimal number of need cores is often unkown a prioi. Amdahls Law states – with some restrictions - that a highly parallel application can be speedup with an increasing number of processing cores. This instance can be used to accelerate an application in order to meet a strict deadline of a real-time system.

Another point which makes the development of image processing systems difficult is the algorithm itself. In most case several methods with different requirements to platform resources and outcome qualities can be used. A flexible and easy to use rapid protopying system is therefore required

At the TU Dortmund an FPGA-based rapid protopying system is developed which enables a system designer to flexibly change the number of cores and the image processing algorithms to explore the design space of a multicore image processing system.

Multicore Rapid Prototyping System


Energy-aware General Purpose Computing on GPUs

Using a GPGPU-capable GPU for scientific and industrial general purpose applications is widely accepted for HPC (High Performance Computing) at the desktop or server level. On the other hand, GPGPUs (General Purpose Computing on Graphics Processing Units) can also be used in embedded systems without a graphical interface in order to accelerate parallel general purpose applications.

The appearance of low power graphics processors makes GPGPU computing extremely interesting for embedded system design. E.g. small optical biosensors with integrated GPU-based analytical units for mass screening at airports to curtail the global spreading in terms of virus infections can be pointed out.

Taking the energy in the software development process into account is mandatory in face of this very scare resource.

Links and Further Reading

Frank Weichert, Marcel Gaspar, Constantin Timm, Alexander Zybin, Evgeny Gurevich, Michael Engel, Heinrich Müller, and Peter Marwedel. Signal analysis and classification for plasmon assisted microscopy of nanoobjects. Technical Report 830, TU Dortmund, Faculty of Computer Science 12, 2010.
Constantin Timm, Andrej Gelenberg, Frank Weichert, and Peter Marwedel. Reducing the Energy Consumption of Embedded Systems by Integrating General Purpose GPUs. Technical Report 829, TU Dortmund, Faculty of Computer Science 12, 2010.
Frank Weichert, Marcel Gaspar, Alexander Zybin, Evgeny Gurevich, Alexander Görtz, Constantin Timm, Heinrich Müller, and Peter Marwedel. Plasmonen-unterstützte Mikroskopie zur Detektion von Viren. In Bildverarbeitung für die Medizin, Aachen / Germany, March 2010.