Summary
We implemented one of the stages of the WebP image encoding pipeline on NVIDIA GPUs, to take advantage of the parallelism offered by CUDA. To do so, we investigated the algorithms used by the reference implementation of WebP encoding and rewrote several versions of portions of the encoding pipeline to best take advantage of CUDA. We then compared the modified pipeline’s performance on both GHC Cluster machines and the Pittsburgh Supercomputing Cluster Bridges machines, versus the sequential C implementation.
December 10: Final Project Writeup, Poster, and Presentation
Our final project writeup is located here, and our poster is located here.
Preview our poster below:
November 18: Project Checkpoint
Our checkpoint report is located here.
Octobr 30: Project Proposal
Our project proposal is located here.