Aes algoritmasinin cuda & MPI kullanilarak parallellestirilmesi

According to today's standards, life goes online. People do their shopping and buying electronics through online stores. They date online and banks are transferring money online. Even, bachelor courses are online. Also, governments may keep their sensitive data such as tactical information for troops or messages for embassies on cloud computing systems which might be located on other countries. Because of its sensitivity, these type of data must be protected from unauthorized access and its integrity should be guaranteed. Cryptography is based on mathematical techniques which concentrated on data con fidentiality, integrity and origin authentication. Encryption Standard (AES) is the national standard of U.S. which is accepted by U.S. government on October 2000. Encryption is a good way to protect data integrity and confi dentiality. Still, encryption requires time and computation power. Today, computers have reached high clocking speed measured by Gigahertz. If one tries to encrypt a data over 1GB it could take more 10 than minutes to nish the operation. Upon thinking of computers, they come with multiple processors. Also, today we have very expensive GPUs installed in our computer cases. These GPUs are almost powerful than CPUs. There are several libraries to get full advantage of CPUs and GPUs. Two examples for these libraries are OpenMPI and CUDA. While OpenMPI allows developer to use all CPUs parallelly, CUDA allows developer to submit his/her code to run on GPU. The application running on GPU might be a serial application or parallel application divided to GPU cores. This study aims to paralleling AES algorithm using both OpenMPI and CUDA libraries and comparing time di erences between these two methods and classical serial method on a CPU.