OpenCL-U
Course Description
This course will concentrate on teaching students the fundamentals of OpenCL, an open standard for programming heterogenous multi-core processors using a parallel programming language (based on ANSI-C99).
The goal of the course is to give students the opportunity to:
- Gain an understanding of OpenCL as a technology framework for enabling heterogenous parallel processing.
- Become familiar with the tools and techniques required to develop an efficient OpenCL enabled application.
- Learn techniques for adapting existing algorithms to a data and/or task parallel programming model.
- Understand the different characteristics of modern multi-core hardware architectures (e.g. CPUs vs. GPUs).
Code Examples
 |
// template for reversing a buffer of text w/platform selection |
 |
/* example showing one possible solution for the buffer reversal problem which uses vector types to/from global memory for buffers of arbitrary size */ |
 |
// template for a 2d floating-point matrix transpose |
 |
// template for matching a pattern in a search buffer |
Documentation
- Munshi, A. 2010. The OpenCL 1.1 Specification. http://www.khronos.org/registry/cl/specs/opencl-1.1.pdf
- Munshi, A. 2010. The OpenCL 1.1 Man Pages. http://www.khronos.org/registry/cl/sdk/1.1/docs/man/xhtml/
Implementations
- AMD: http://developer.amd.com/gpu/ATIStreamSDK/Pages/default.aspx
- APPLE: http://developer.apple.com/search/index.php?q=opencl
- IBM: http://www.alphaworks.ibm.com/tech/opencl
- INTEL: http://whatif.intel.com/
- NVIDIA: http://developer.nvidia.com/object/opencl.html
- SNU-SAMSUNG: http://opencl.snu.ac.kr/
Developer Tools
- Gremedy gDEBugger: http://www.gremedy.com/
- NVidia Parallel NSight: http://www.nvidia.com/object/parallel-nsight.html
- NVidia Visual Profiler: http://developer.nvidia.com/object/visual-profiler.html
Comments