Efficient Implementations of Hash Functions in AArch64 Assembly

In the field of big data processing, universal and strongly universal hash functions are essential for handling large amounts of data. This thesis focuses on implementing a set of hash functions as efficiently as possible. After implementing each function in Rust, a modern high-level language, an inline AArch64 assembly implementation is developed. The results show that, even compared with modern compilers, hand-written assembly can still outperform compiler-generated code. Through benchmarks, this thesis also analyzes how specific design choices affect the functions' performance. Finally, it determines whether the complexity of writing AArch64 assembly is worth the resulting performance gain.