Recently, it became popular to utilise the many cores present in modern graphics cards (GPUs, graphics processing units) for general purpose computing applications. This development led to GPGPUs, or general-purpose computing on graphics processing units. With a peak performance of 1 TFlop/s, the new Tesla GPGPUs from Nvidia promise supercomputing power for the costs of a workstation.
GPGPUs are true many core processors (e.g. current Tesla computing processors have 240 cores). Since their archtitecture differs much from commodity x86 CPUs, new programming models and abstraction layers are needed. For this, Nvidia developed the Compute Unified Device Architecture (CUDA).
The workshop will give an introduction into GPGPU computing, the CUDA parallel programming model and real world applications. Additionally, Nvidia and FluiDyna will showcase Tesla GPGPU computing products.
The LRZ is proud to host this workshop on heterogeneous computing on GPGPUs, organized by Nvidia and FluiDyna.
The event will take place in the lecture halls at LRZ in Garching
- Date: February 11th, 2009, 13:00-17:00
- Location: LRZ Building, Garching Campus, near Munich
- Further information: http://www.lrz-muenchen.de/services/compute/aktuell/ali3359/