Software for Printing Using a GPU

Project Name GPUed: A GPU-based Error Diffusion Library in C. Project Objective To adapt an error diffusion algorithm to function on a GPU, leveraging the parallel architecture of the GPU to achieve a significant performance improvement over a comparable error diffusion algorithm running on a multi-core CPU. Deliverable A C/C++ source code project, to be compiled to a static library for use with other projects. Our preference would be to work in Microsoft Visual Studio, but this is not a requirement. The project is ultimately intended for application across multiple platforms. Work Performed to Date We have already developed a multi-threaded error diffusion algorithm that executes one color channel per core, and a functioning framework for testing new error diffusion routines. The existing code already leverages OpenMP and OpenGL, and performs image scaling and color transforms on the GPU prior to error diffusion on the CPU. Our goal now is to adapt our error diffusion algorithm to a GPU. Post-Development Objectives Academic publication of the work, including pseudo-code; conference presentation. If applicable, joint patent application and joint licensing for commercial purposes, and release of source code under GNU GPL for non-commercial and academic development. Additional Comments Additional documents are available under Resources. Cassinian developers are available for questions, additional information, access to past research projects, and additional project specifications to facilitate development, by email or video conference.