GPUlab
GPUlab is a competence center and laboratory for research in the use of Graphics Processing Units (GPUs) for visualization, scientific and high-performance computing. GPUlab is physically located at the Section for Scientific Computing, DTU Compute, Technical University of Denmark, see http://gpulab.compute.dtu.dk/. The purpose of GPUlab is to attract focal interests in the use of GPUs by engineering students, researchers, and Danish industry. Since 2010, the GPUlab has hosted several educational activities with support from two PhD schools, namely, ITMAN and DCAMM schools. Some of these activities have been supported from 2010-2013 by a national research grant received by Prof. Per Christian Hansen from the Danish Research Council for Technology and Production Sciences. Specific focus on the use of CUDA-enabled GPUs with attendance of more than 300 people of which more than 200 had a research background.
We are happy to announce that the GPUlab activities are now an integrate part of the DTU Computing Center and will be continued by the DCC team and key members from Scientific Computing from 2014 and forward.
If you are developing GPU applications you might want to have a look at our GPU programming Tips & Tricks section.
Hardware
At GPUlab a small hardware facility has been established which was initially maintained by the Scientific Computing research group. The hardware is now part of the general HPC system at the DTU Computing Center and includes the most recent GPUs from NVIDIA to facilitate use of CUDA-enabled GPUs in research projects.
As an official CUDA research center and CUDA teaching center, through the years 2013-2016, GPUlab has received regular donations of hardware from NVIDIA of generations Fermi, Kepler, and Maxwell. In addition, the DTU Computing Center has regularly acquired more GPUs in order to satisfy the increasing demand from researches and courses at DTU. Some of these GPUs are purchased in collaboration with research groups, who then have priority usage of them, but are still maintained by DTU Computing Center and open to all HPC users.
Getting access
If you are a student or an employee at DTU and already have an account on the HPC system, you can use GPUs as described here.
For any one else who would like to get access to the GPUlab hardware (for development and testing only), please contact us by email at support@hpc.dtu.dk.
Workshops/courses/seminars
GPU related activities for researchers in academia and industry are hosted every year. We have held several workshops, seminars and PhD schools with a clear combined educational and research scope regarding optimal use of modern many-core GPUs for massively parallel scientific computations in engineering applications.
Please visit our course pages for more info or contact us directly if your have special course requests for your research group or company.