Implementation of high speed rainbow table generation using Keccak hashing algorithm on GPU

This paper proposes the implementation of high speed rainbow table generation using Keccak hashing algorithm with the integrated development environment CUDA for GPU in the heterogeneous GPU+CPU system. Utilizing the GPU's powerful capacity, the algorithm greatly improves the performance of rainbow chain generation by dispatching the pre-computation of rainbow chain to each GPU thread. The table generation speed on GPU+CPU system and CPU was compared by the configuration of chain length and number of chains in this paper. In addition, the password coverage rate of table generated by the proposed reduction function was evaluated.The speed-up of pre-computation on GPU GeForce GTX 1080 outperforms that on CPU Xeon E5-1620 v4 3.5GHz by 239 times when the chain length is 200.