A Tutorial on High Performance Computing Applied to Cryptanalysis

Cryptology and computers have a long common history; in fact, some of the early computers were created as cryptanalytic tools. The development of faster and widely deployed computers also had a great impact on cryptology, allowing modern cryptography to become a practical tool. Today, both computers and cryptology are not only practical, but they have became ubiquitous tools. In truth, computing devices incorporating cryptography features range from very small low-end devices to supercomputer, going through all possible intermediate sizes; these devices include both general purpose computing devices and specific, often embedded, processors which enable computing and security features in hundreds of technological objects.

In this invited talk, we mostly consider the cryptanalytic side of things, where it is fair to use very large amounts of computing power to break cryptographic primitives or protocols. As a consequence, demonstrating the feasibility of new cryptanalytic methods often requires large scale computations. Most articles describing such cryptanalyses usually focus on the mathematical or algorithmic advances and gloss over the implementation details, giving only sufficient data to show that the computations are feasible. The goal of the present abstract is to give an idea of the difficulty facing implementers of large scale cryptanalytic attacks.