Accelerating the MMD algorithm using multi-core environments

This paper presents two approaches to accelerate the MMD algorithm in multi-core environments. The MMD algorithm is a transformation-based algorithm based in the field of reversible logic synthesis. It is used to synthesize and optimize reversible circuits which are an integral part of future technologies like quantum computers. However, the MMD algorithm is computationally intensive and the acceleration of the algorithm might not only produce faster but also better results. This paper focuses on two parallel hardware environments, the Cell Broadband Engine and the NVIDIA Tesla architecture. In the course of this project two different parallel algorithmic approaches have been implemented on both hardware architectures. These implementations have been compared in order to find the best combination of algorithmic approach and matching architecture. Additionally, the answer to the question if parallel hardware architectures are a means to improve algorithms in the field of reversible logic synthesis has been examined.