Heterogeneous Computing at AMD Developer Summit 2013
Today we live in a world where the divide between types of processors such as CPUs and GPUs have disappeared. Heterogeneous computing brings together processors of different types in one unit such as AMD’s A-Series Accelerated Processing Units (APU), giving developers the chance to tap into vast amounts of compute power to increase application performance and enable new user experiences.
The AMD Developer Summit, taking place at 11-13 November in San Jose, will have a track that provides a high-level view of GPU compute strategies and the Heterogeneous System Architecture (HSA), showcasing the possibilities of heterogeneous computing and what it means to developers.
In 2011 AMD released the world’s first APU that packed a discrete-class GPU with a powerful CPU. In the past, GPUs were used to render pixels on screen, but with an AMD A-Series APU, the GPU can be used to provide parallel compute capabilities that until recently were limited primarily to supercomputers.
The AMD Developer Summit’s heterogeneous computing track offers 12 tracks providing a wide range of information such as the use of OpenCL and OpenMP on server APUs, optimizing applications to make use of HSA to libraries for neural networks, tips and techniques to extract parallelism and optimize for GPU addressable memory.
AMD has been a proponent of OpenCL, a programming language that allows developers to tap into the compute power in APUs and discrete GPUs. A number of software developers have seen the potential of OpenCL, including industry leading imaging software firms Adobe and Corel, with AMD providing support to enable key features by tapping the GPU in AMD APUs.
While AMD has been a vociferous supporter of OpenCL, HSA will offer developers the chance take GPU acceleration to the next level. At the AMD Developer Summit, AMD will provide an overview of HSA and show how it will make working with the GPU a seamless experience. But it is not just theory, there will be a session showing how OpenCL performance in Corel’s popular multi-platform photo management software AfterShot Pro can be improved by making use of HSA.
GPU compute is becoming ubiquitous, and AMD is working with the developer community to make it easier for them to access the power of the GPU. Some examples of AMD’s work in this area include the open source ClMath library and Aparapi, which enables Java developers to convert Java bytecode into OpenCL. This track will highlight some of the work being done in the areas of libraries and code-profiling in order to extract the most compute performance out of a GPU.
The heterogeneous computing track at the AMD Developer Summit will allow developers the chance to see what is possible through GPU compute and provide information on how to extract the most out of the compute power within APUs and GPUs. With seven of the talks being given by AMD engineers, there is no better place to learn the current state-of-the-art on GPU compute and HSA.
George Kyriazis is a Principal Member of Technical Staff at AMD. His postings are his own opinions and may not represent AMD’s positions, strategies or opinions. Links to third party sites, and references to third party trademarks, are provided for convenience and illustrative purposes only. Unless explicitly stated, AMD is not responsible for the contents of such links, and no third party endorsement of AMD or any of its products is implied.