Analysis of Concurrent Lock-Free Hash Tries with Constant-Time Operations

TL;DR

This paper introduces an extension to concurrent hash tries by adding a cache, which reduces operation time complexity from logarithmic to constant, enhancing scalability and efficiency.

Contribution

The paper proposes a novel cache-based extension to concurrent hash tries, achieving constant-time operations and improving performance over traditional logarithmic-time methods.

Findings

Operations run in O(1) time with the cache extension

The cache improves scalability and efficiency

Theoretical analysis supports performance claims

Abstract

Ctrie is a scalable concurrent non-blocking dictionary data structure, with good cache locality, and non-blocking linearizable iterators. However, operations on most existing concurrent hash tries run in O(log n) time. In this technical report, we extend the standard concurrent hash-tries with an auxiliary data structure called a cache. The cache is essentially an array that stores pointers to a specific level of the hash trie. We analyze the performance implications of adding a cache, and prove that the running time of the basic operations becomes O(1).