>> Term Paper
Research Thesis Review: Thesis: Parallel Lossless Data Compression on the GPU Author: Ritesh A Patel
Institution: Electrical and Computer Eng., University of California, Davis
In this homework assignments, students are asked to review and write a report on an academic research thesis related to parallel and GPU programming and Applications. This paper is a Master's thesis published at the University of California, Davis. The thesis may be downloaded from the link above.
Please read the thesis carefully and critically. After reading the thesis, please prepare a short report on the research - specifically its approach, performance, and future extensions. To guide you in developing your report, below, you may find questions which would be helpful to ask yourself while reading the thesis, and answer these questions based on what you learned from the paper.
a. what is the background of the research, and problem the thesis is addressing?
b. Explain the pipeline (the stepwise processes involved with the problem the researchers are trying to solve) of the technique in question?
c. What challenges the researchers are attempting to address?
d. What's the contribution of this research?
2- Pipeline: a. Explain in your own words, your understanding of the whole process and its implementation in this thesis. b. Explain in your own words, the traditional approach for the said pipeline.
3- Results: a. Explain your understanding of the results obtained by the researchers in applying their implementation of the pipeline. b. Explain in your own words, the comparisons between the results of this work and those of the traditional techniques, as presented in the thesis. c. Draw your own conclusions on how this thesis compares with the traditional technique(s) from the literature; in particular, what's the strengths of this work and what may be its weaknesses compared with the traditional approaches.
4- Extensions: a. Draw your own conclusions on what needs to be done to improve and extend this work.
Preview Container content
Back Ground and Overview:
Data compression is obviously one of the foremost concerns of people dealing with the data transmission. Data compression plays a vital role in the instances where large amount of data will be transferred. The objective of the process is to increase the effective data density by reducing the redundant data and by minimizing the data communication.
In the current instance GPU based data compression techniques are considered for data compression. Actually GPU based techniques are well suited for large-scale file processing requirements. The GPU benefits the performance of parallel compression through its memory hierarchy.
Anyhow the two main issues the thesis is trying to address are two, firstly the cost of computation is tried to get minimized and secondly through compression the cost of the data transmission is tried to get reduced. As well the time and subsequent cost of computation are considered to reduce during the process by compressed data transmission.