Title: Performance Analysis of Parallel Sha-3 On Gpu
Author : Mohamed Mohamed Abd El-Fatah Mohamed
Collection : M.Sc. Computer
Abstract:
Cryptographic hash function is one of the most important tools in the field of cryptography. It is widely used in many cryptographic applications such as digital signatures and message authentication.
Cryptographic hash functions are typically designed based on an iterative manner and that causes a performance bottleneck. Any speedup of the hash function performance can directly enhance the performance of its applications.
Because a stream of attacks on the standard hash function SHA-1 were proposed [1, 2] and its successor SHA-2 follows the same design, NIST (National Institute for Standards and Technology) announced a competition to develop a new powerful hash function design for a new standard SHA-3. Three of the final candidates are working in iterative manner.
The subject of this thesis is the study of cryptographic hash functions performance on GPU. GPU is a powerful architecture because it contains a high number of processing units and it is a common device on most of the modern PCs.
A secure parallel model is used to extend the ability to enhance the hash function performance, while preserving the original security claim of it.
The three iterative candidates of SHA-3 are implemented on GPU to discuss their performance on such architecture.
The performance analysis shows that a performance speedup is gained for all candidates with different speedup values, their design and the factors affecting speedup values are discussed in details. Also, it is shown that the implementation of all candidates on GPU can gain more performance speedup, if they are executed on GPU with
more processing units.
ليست هناك تعليقات:
إرسال تعليق