| Home Page |
Author: Sudarshan Rao
Committee
Chair: Prof. Alan Kaminsky
Reader: Prof. Stanislaw Radziszowski
Observer: Prof. Edith Hemaspaandra
Title: Parallel Cube Testing on GPUs
Defense Date: June 11, 2010
Abstract: As computers become more involved in our day to day lives, computer security becomes more of a necessity. Cryptographic primitives are the building blocks of protocols that guarantee safety of our data. Given their frequent use, they must be studied for vulnerabilities extensively. Considering improvements in construction of these primitives, tools that study them must change too.
In this project, we develop a framework for performing property testing on graphics processing units (GPUs). The testing technique used is a relatively new technique known as cube testing. We study two new primitives, Keccak and Threefish, and an older primitive, AES. The framework developed makes use of NVIDIA's CUDA framework to get a significant performance boost by parallelization. We present the results of a balance test on the three primitives and show that the GPU does indeed provide considerable speedup.
Proposal: proposal.pdf
Report: report.pdf
Defense Presentation: presentation.pdf
| Home Page |