During our courses/trainings we will teach you the best of what you can find here.
We try to keep the following information as complete as possible, so please contact us if something is missing.
- Hands on OpenCL, by Simon McIntosh-Smith and Tom Deakin from the University of Bristol in the UK. It currently is the most up-to-date tutorial on OpenCL, including code for lab-sessions.
- Bruno Jurkovski wrote a clear quickstart.
- AMD introduction to OpenCL.
- MacResearch playlist on Youtube. Code of episode 3 and 6. Zip of PDFs.
- CMSoft’s complete OpenCL tutorial.
- The Code Project has a series on OpenCL, episodes 1, 2, 3, 4, 5, 6, 7 and 8. By Rob Farber.
- Dr.Dobb’s has a series called “CUDA, Supercomputing for the Masses”. It is CUDA-oriented, but you can learn a lot about GPGPU in general and on NVIDIA specific optimisations. Login to their site and then you can access parts 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20 and 21. Registration is free.
- AMD’s university program. This is loads of information!
- NVIDIA’s OpenCL pages provide all you need to program on NVIDIA.
- Enjalot’s adventures in OpenCL giving the basics in OpenCL and pyOpenCL.
- StreamComputing’s basic concepts with various tips&tricks on OpenCL.
- KISTI Supercomputing Learning Centre has a beginners course for OpenCL. Material including PDFs and code is available on SF.net.
- OpenCL cookbook by Dhruba Bandopadhyay.
- Anteru’s introduction to OpenCL, part #1, #2 and #3.
OpenCL Optimisation guides
- Intel Xeon and XeonPhi
- NVidia (CUDA, but same applies to OpenCL)
- AMD GPUs and CPUs
- ARM MALI T600
- Altera FPGAs
Not available (yet):
- Imagination PowerVR
- Qualcomm Adreno
- Xilinx FPGAs
OpenCL-based GPU-programming courses
- Marcus Bannermann university course made for the university of Erlangen, Germany.
- Advanced Parallel Programming is a course on parallel programming by professor John Cavazos of University of Delaware.
- Programming for Performance is a course on parallel programming by Jonathan Eyolfson of University of Waterloo.
- Manchester OpenCL tutorial wiki. Materials from previous courses and more.
- University of Innsbruck GPU-programming using OpenCL by Juan J. Durillo PhD.
- University of Waterloo Programming for Performance. Lecture notes and assignments.
- AMD’s OpenCL introduction. Takes about an hour in total, slides are provided.
- Harvard Lectures on GPGPU. One hour each.
- AMD optimisation case study: Diagonal Sparse Matrix Vector Multiplication .
- AMD optimisation case study: Simple reductions.
WebCL is a new standard-to-be for OpenCL in the browser. Currently there are a few implementations, while Khronos is working on an official standard. WebCL is available on Firefox for Linux32, Windows32 and Windows64 by Nokia. Also available for Safari on OSX by Samsung. A Node.js-implementation is made by Motorola. Examples made for another implementation will probably not work.
Basic knowledge of C is needed to understand how to write kernels. Also many tutorials are in C++.
- A little C primer.
- C++ for Java-programmers.
- C for Java-programmers.
- http://www.stanford.edu/class/cs101/ for if you have never programmed – don’t think about GPU-programming yet.
Getting a grasp of OpenGL has advantages. Techniques for faster memory-operations in OpenGL have equivalents in OpenCL, giving reason to read on this subject.