PrintCustom Computing and Many-Cores
Two of the hottest trends in computer architecture are custom computing machines and many-cores. Custom computing machines use massively parallel, programmable hardware to build application-specific processing units that are tailored to the applications. Research has shown, that custom computing architectures based on field-programmable gate arrays (FPGAs) are able to accelerate computational intensive application kernels by up to several orders of magnitude over conventional CPUs. Besides FPGAs, we are also exploring the use of novel many-core processing architectures, such as general purpose graphic processing units (GP-GPUs), floating point arrays, or CELL processors.
Our research focus is the investigation of theoretical foundations and practical applications of custom computing machines and many-cores in the context of high performance computing. Specifically, we aim at improving the specification, design and implementation processes to make the potential of these innovative computer architectures more accessible to researchers and users with limited background and experience in computer systems. Currently, we work on the following projects:
- EPiCS: Engineering Proprioception in Computing Systems (EU FP7 IP)
- Custom Computing Architectures for Computational Nanophotonics
- Gomputer: Accelerating Monte-Carlo game tree search
- MM-RPU: A Multimode Reconfigurable Processing Unit: Architectural Integration and Programming
- MOVES: Multi-objective intrinsic evolution of embedded systems
- ParMRI: Fully Automated Tumor Diagnosis for MRI
- IMORC: Performance analysis and optimization for reconfigurable accelerators
The research in this area is supported by the core group of the Paderborn Center for Parallel Computing (PC˛) and the Computer Engineering Group of the University of Paderborn. We are constantly looking for new people joining our team. If you are interested in working in this area please contact Christian Plessl. We are also offering student projects in this area, please refer to the student projects pages of PC˛ and the Computer Engineering Group
