high performance computing on graphics processing units: hgpu.org

This project was done with the Mobile Microprocessor Group at Intel Corporation as a part of a six month internship. The primay objective of this project was to study the performance of GPGPUs (General purpose computation on Graphics Processing Units) for various benchmark applications. GPGPUs have gained wide spread importance in recent years because of their extensive computation capabilities. A GPGPU uses the GPU, which traditionally handles computation only for graphics, to perform computation for applications traditionally handled by the central processing unit (CPU). My role was to do detailed performance analysis on silicon to understand various aspects of the micro-architecture, usefulness for future generations; as well as gain some experience and expertise in developing ,programming and optimizing GPGPU workloads. During the life of the project of six months, initial phase was learning GPGPU programming models, Open Computing Language (OpenCL) architecture, thread management in OpenCL, and a new Intel propriety language used to for media programming called C for Media. Another topic of widely accepted importance was media performance studies (ongoing). The high level goal of this exercise was to understand and analyze performance bottlenecks in a generic media playback pipeline scenario, optimize it and thus increase efficiency of a media playback pipeline.