Cedric Nugteren – Eindhoven University of Technology

 

Cedric Nugteren

Cedric Nugteren has received his Bachelor and Master of Science degrees from the Eindhoven University of Technology. Already during his Master, he showed interest in parallel processors in general and GPUs in particular. His Master thesis was performed in the early days of general purpose usage of GPUs – not long after the release of the first version of CUDA. For his thesis, he investigated possibilities to improve register allocation in the CUDA tool-chain.

Cedric is now a third-year PhD-student at the Electrical Engineering department of the Eindhoven University of Technology, and still works with GPUs. He has focused his research on making GPUs and other parallel platforms accessible to a wide range of users. This has resulted in the design of a compiler based on algorithmic skeletons, a technique which allows for automatic generation of highly efficient program code. This compiler, named ‘Bones’, is publicly available.

At Eindhoven University of Technology, Cedric works in a small research team led by professor Henk Corporaal. This team, named “Parallel Architecture ReSearch Eindhoven” (PARsE), works on a wide variety of topics, including: compilation, GPUs, FPGAs, high-level-synthesis, low-power SIMD processor architecture, image processing, computer vision, and neural networks. Within the PARsE group, we have recently build a 16-GPU cluster from commodity hardware. The 16-GPUs, divided over 4 machines, can achieve a theoretical peak performance of 22TFLOPS using 4KW.

* Visit http://www.cedricnugteren.nl/ for contact details and a list of publications. Visit http://parse.ele.tue.nl/ for more information about our research group and the GPU cluster, and http://parse.ele.tue.nl/bones/ to obtain the Bones source-to-source compiler.

Our research group