Some observations on the optimization of a parallel SHAKE function using Sakura

TL;DR

This paper explores parallel SHAKE hash function constructions using Sakura coding, analyzing their efficiency in terms of parallel time and processor requirements based on tree-based algorithms.

Contribution

It introduces new parallel SHAKE constructions with analysis of their performance metrics, based on Sakura coding and Keccak permutation.

Findings

Parallel algorithms achieve reduced depth with more processors.

Tree-based structures influence the trade-off between speed and resource use.

Preliminary results suggest optimal configurations depend on chaining value length.

Abstract

Some parallel constructions of a SHAKE hash function using Sakura coding are introduced, whose basic operation is the Keccak's permutation. For each proposed tree-based algorithm, observations are made on both its parallel running time (depth) and the required number of processors to reach it. This preliminary work makes the assumption that the tree-level chaining value length is equal to the capacity of the underlying sponge construction, as recommended in the Sakura paper.