Implementasi Algoritme Enkripsi Advanced Encryption Standard secara Paralel dengan GPU NVIDIA CUDA

Abstract

Processor streaming architecture in GPU has provide general purpose computation which have its own potential to help CPU with its parallel computation capability. One example of supported processor streaming architecture is Fermi which is developed by NVIDIA. Furthermore, NVIDIA has developed its own software development kit namely CUDA, which can integrate many popular existing programming languanges such as C++ for GPU-based parallel computation. This research focus on the implementation of parallel AES algorithm with GPU and analyze the performance result of the implementation in two different NVIDIA devices, namely NVIDIA GeForce GT220 and NVIDIA Quadro 600. In general, the performance of parallel GPU computation is better than sequential computation for large data input. It is obtained that the maximum speedup of NVIDIA GeForce GT220 in 8 MB input is 89.3 with 180% efficiency. Higher speedup of 101.6 can be obtained by NVIDIA Quadro 600 with the same input size with around 100% efficiency.