Acceleration of MP3 encoder by using the GPU

Abstract

At the present time, CUDA is a useful and common computing engine used by many people. It accelerates the computing speed by running simply same tasks at the same time which is called the parallel processing. Many scientists use it as a coprocessor to decrease the computing time in various areas such as image processing, weather forecast or heart wall tracking. Lame is a high quality and an efficient MP3 encoder. Also, it is an open source software which makes it the most popular MP3 encoder in the world. In 2008, NVIDIA had held a contest for accelerating Lame by CUDA. Only 2 out of 181 teams handed the program. The champion team got 2.4x speed up in GPU compared to CPU. It seems like there are still a lot of room to improve. In this project, I am going to use the CUDA to accelerate the lame MP3 encoder and figure out which parts are worth to do.