Introduction to OpenCL and what you should know about the differences between GPUs and CPUs
- Date and Time
Thursday, October 29th, 2009 at 13.30 (until 15.00)
Polacksbaken, room 1146, (or eventually room 1211, if needed)
OpenCL is a new standard for running C-like computation kernels across a wide range of devices. With it you can dramatically accelerate data-parallel processing on modern GPUs and easily take advantage of the vector units in modern CPUs.
However, to really take advantage of these devices you need to be aware of their architectural differences and the impact this can have on your application.
This seminar will give you an introduction to OpenCL, including the memory- and computation-models, some sample code to get started, and an overview of the differences between modern CPU and GPU architectures.