Many software implementations and algorithms can be improved with using of OpenCL and on Intel's platform we have strong support of such GPU-based computations and in addition to traditional CPU path. During this presentation I will describe and cover several topics of OpenCL details on IA, some plans for the next platforms and OS, possible optimization and performance tuning with such tools like Vtune, as well as used power correlations.
En detail:
Intel's focus on OpenCL as computational vehicle while on IA:
* OpenCL standard and new feature upcoming, details of OpenCL 2.0 and upcoming one
* What specifics do we have implemented and developers need to know while on IA HW , major differences to other vendors.
* How do we deliver and handle OpenCL related questions and provide technical support
* Tools from Intel which can help to develop and optimized OpenCL kernels
- Using Code Builder and Vtune with highlights explained
- Interoperation of OpenCL kernels with video HW processing, like MediaSDK/QSV
* BKM/Best Known Methods about optimization of particular kernels and outcomes that each developer should know
Skills
Basic knowledge of OpenCL would be good to have.
Lernziele
It would be good to see more applications using OpenCL while on Intel Architecture, and hopefully after this presentation there will be no questions about position of Intel company and OpenCL, as well as software tools available for developers during implementation and tuning on Intel Architecture.
// Referent
//
Maxym Dmytrychenko
is working for last 8 years at Intel, Europe, Developer Relation Division with main responsibility in engaging with ISV across Europe in domains of video codecs, raytracing and professional image processing. His main focus is the early adopting and optimization SW implementations toward Intel Achitecture, including CPU, GPU etc.