@@ -10,7 +10,7 @@ \maketitle -\section{Introduction} +\section*{Introduction} Hash functions are a critical tool in computer security. A hash function allows for fingerprinting and verification of a piece of data by running a deterministic algorithm over a set of data with the property that if any slight changes are made to the data a very different hash will result. @@ -28,23 +28,23 @@ Thus there is a requirement, as there similarly was in the AES competition, for NIST will analyze ``each submission’s optimized implementations on a variety of platforms as specified in Section 6.B, and for a variety of input message lengths.'' \cite{Sha3Request}. Since computational efficiency will be an important part of the decision in choosing SHA-3 some authors of proposed algorithms have taken great care to design their algorithms in such a way to be exploit parallelism where ever possible. -\section{Approach} - -We are planning to choose a number of hash functions. -Including at least one NIST competitor, implement them in OpenCL and see what kind of speedup we van get by using general purpose graphics hardware programming. -We will be using a Mac Pro with an ATI Radeon HD 4870 for testing, as well as other machines. - - -\section{Plan} +\section*{Approach} +Because we are not skilled cryptographers and others are already analyzing these algorithms \cite{Sha3Zoo}, we will instead explore the runtime speed of a subset of the round 2 SHA-3 candidates. +We plan to evaluate the computational efficiency of the following SHA-3 candidates: Skein, CubeHash and Echo. +We choose Skein and CubeHash primarily based on the popularity of the authors in the cryptographic community, and we choose Echo because it is based upon AES. +While the each of these algorithms already have general purpose implementations we plan to implement them for execution on a GPU using the OpenCL and/or CUDA frameworks. +There is currently a wealth of information for FPGA and ASIC implementations of these algorithms, however we are unaware of OpenCL/CUDA implementations of any of the SHA-3 candidates \cite{Sha3Hardware}. +Our hypothesis is that some of these algorithms will benefit from parallel execution while others may not. +Thus, we will compare out runtime results on the GPU with runtime results on a CPU. +Our reference machine will be a Mac Pro with an ATI Radeon HD 4870. + +\section*{Plan} \begin{description} - \item[October 25th] Learn and experiment with OpenCL APIs. Select number and which specific hash functions to evaluate. - \item[November 13th] Have results for at least two algorithms. Explore either attempting to find collisions for one of those algorithms, or using more algorithms. - \item[November 23th] Project due. + \item[Week 1] Learn and experiment with OpenCL/CUDA APIs. + \item[Week 2-3] Implement algorithms such that they comply with the test vectors as specified by the authors. + \item[Week 4] Evaluate the runtime of our implementations as compared to the reference implementations on a CPU. Prepare project presentation and report. \end{description} - -\section{Literature} - \begin{thebibliography}{9} \bibitem{Sha3Request} ``Announcing Request for Candidate Algorithm Nominations for a New Cryptographic Hash Algorithm (SHA–3) Family.'' Federal Registrar, Vol. 72, No. 212 (November 2007), pp. 62212-62220. Available at \url{http://csrc.nist.gov/groups/ST/hash/documents/FR_Notice_Nov07.pdf}. \bibitem{Nostradamus} Marc Stevens, Arjen K. Lenstra, Benne de Weger. ``Predicting the winner of the 2008 US Presidential Elections using a Sony PlayStation 3.'' November 30, 2007. Available at \url{http://www.win.tue.nl/hashclash/Nostradamus/} @@ -53,5 +53,4 @@ We will be using a Mac Pro with an ATI Radeon HD 4870 for testing, as well as ot \bibitem{Sha3Hardware} ``SHA-3 Hardware Implementations.'' Available at \url{http://ehash.iaik.tugraz.at/wiki/SHA-3_Hardware_Implementations} \end{thebibliography} - \end{document} ProjectProposal.tex 040ae565f68a90584a138793c7ea244e3f61a771 Cory T. Cornelius Cory.T.Cornelius@Dartmouth.edu http://github.com/dxoigmn/ENGS116Proj/commit/8523e8912ce5535d34b9072dcfee80d965d30f08 8523e8912ce5535d34b9072dcfee80d965d30f08 2009-10-13T19:31:32-07:00 2009-10-13T19:31:32-07:00 Fill in approach. Be date agnostic in the plan. 48d827624022cc187d80f0b61eeb170b0b0b0b9b Cory T. Cornelius Cory.T.Cornelius@Dartmouth.edu