wiki:UGT2012

R.Tayama

Data Compression of Huffman Encoding on GPU

The purpose of my study is how to speed-up Huffman encoding by parallel processing. For this purpose, I used a Graphics Processing Unit (GPU) to implement this compression in parallel and I have adopted the algorithm that called "Parallel Variable-Length Encoding" because an algorithmic modification was necessary to implement this compression on GPU by OpenCL, and I have compared the processing time on GPU and Central Processing Unit (CPU). As a result, the performance of data compression that using Huffman encoding on GPU was 10 times faster than that using only CPU.

file:///home/committee/aac/Thesis2012/s1160139

Last modified 10 years ago Last modified on Nov 5, 2014 9:38:58 AM