CUDA Compatible GPU as an Efficient Hardware Accelerator for AES Cryptography

Abstract

This paper presents a study of the efficiency in applying modern graphics processing units in symmetric key cryptographic solutions. It describes both traditional style approaches based on the OpenGL graphics API and new ones based on the recent technology trends of major hardware vendors. It presents an efficient implementation of the advanced encryption standard (AES) algorithm in the novel CUDA platform by Nvidia. AES is currently the most widely adopted modern symmetric key encryption standard. The performance of the new fastest GPU solution is compared with those of the reference sequential implementations running on an Intel Pentium IV 3.0 GHz CPU. Unlike previous research in this field, the results of this effort show for the first time the GPU can perform as an efficient cryptographic accelerator. The developed solutions run up to 20 times faster than OpenSSL and in the same range of performance of existing hardware based implementations.