International Association
for Cryptologic Research

Zero-Knowledge proof systems are widely used as building blocks of different protocols, e.g., such as those supporting blockchains. A core element in Zero-Knowledge proof systems is the underlying PRF, usually modeled as a hash function that needs to be efficient over finite fields of prime order. Such hash functions are part of a newly developed paradigm known as Arithmetization-Oriented designs.

In this paper, we propose two new AO hash functions, XHash8 and XHash12 which are designed based on improving the bottlenecks in RPO [ePrint 2022/1577]. Based on our experiments, XHash8 performs $\approx2.75$ times faster than RPO, and XHash12 performs $\approx2$ times faster than RPO, while at the same time inheriting the security and robustness of the battle-tested Marvellous design strategy.