high performance computing on graphics processing units: hgpu.org

Neural networks have been part of an attempt to emulate the learning curve of the human nervous system. Graphics Processing Units (GPUs) that come with a Graphics card have hundreds of processing cores, and have highly parallel architecture. Because of the highly parallel architecture of GPUs, it suits very well for parallel architecture such as Neural Network. In fact, GPUs have become a General Purpose Processor, and is a good option for implementation of many Parallel Algorithms, including ANN. Further recent advancements in GPU computing have made it easier to utilize the resources of a GPU. Specifically the programming model has been made much simpler. NVIDIA OR AMD OpenCL for example can be extremely helpful in accelerating ANN algorithms on GPUs. Thus OpenCL accelerated ANN algorithms can be used in many real-time applications, including image processing, object classifications, voice recognition and in a number of systems which require intelligence and auto control. This research thus aims at implementation of a Neural Network on a GPU in order to improve the performance as compared to CPU implementation in a particular application. Particularly, we have chosen NVIDIA OR AMD’s GPU and OpenCL platform for this implementation.