The course presents OpenCL within the framework of other relevant parallel computing graphics adapter card standards and as a part of the Khronos standards.
The course will also outline parallel processing capabilities when coding video.
Who should attend?Developers and product managers in Start-ups providing parallel computing and video services and who want to understand OpenCL.
Basic understanding of video compression and transfer. Knowledge of C / C++.
1. Introduction to parallel computing
- Processing methods: SISD, SIMD, MIMD, Parallel
- Efficiency of parallel processing
- Parallel processing and video coding
- Review of past forum and standards (OpenMP, GPGPU, DirectCompute etc)
2. Khronos Standards
- Interoperability issues
- OpenGL, OpenGL ES, OpenMAX
- OpenSL, OpenCL
- Industry adaptation and usage in mobile OS/Devices
3. OpenMAX and video coding
- OpenMAX standard
- OpenMAX layers
- Industry adaptation
4. OpenCL – 4 hours
- Processing Architecture
- Memory Resource (Shared, private)
- Events & Synchronization
5. CUDA and OpenCL
- CUDA pros cons in comparison to OpenCL
- Main differences review
6. Host coding sample
7. GPU Coding samples
- C99 and kernel programing
- Review of Kernel program
- DCT Kernel
- Video processing transform