Award Date

May 2017

Degree Type

Thesis

Degree Name

Master of Science in Computer Science

Department

Computer Science

First Committee Member

Ajoy Datta

Second Committee Member

John Minor

Third Committee Member

Yoohwan Kim

Fourth Committee Member

Venkatesan Muthukumar

Number of Pages

64

Abstract

Priority queues are data structures that store information in an orderly fashion. They are of tremendous importance because they are an integral part of many applications, like Dijkstra’s shortest path algorithm, MST algorithms, priority schedulers, and so on.

Since priority queues by nature have high contention on the delete_min operation, the design of an efficient priority queue should involve an intelligent choice of the data structure as well as relaxation bounds on the data structure. Lock-free data structures provide higher scalability as well as progress guarantee than a lock-based data structure. That is another factor to be considered in the priority queue design.

We present a relaxed non-blocking priority queue based on skiplists. We address all the design issues mentioned above in our priority queue. Use of skiplists allows multiple threads to concurrently access different parts of the skiplist quickly, whereas relaxing the priority queue delete_min operation distributes contention over the skiplist instead of just at the front. Furthermore, a non-blocking implementation guarantees that the system will make progress even when some process fails.

Our priority queue is internally composed of several priority queues, one for each thread and one shared priority queue common to all threads. Each thread selects the best value from its local priority queue and the shared priority queue and returns the value. In case a thread is unable to delete an item, it tries to spy items from other threads' local priority queues.

We experimentally and theoretically show the correctness of our data structure. We also compare the performance of our data structure with other variations like priority queues based on coarse-grained skiplists for both relaxed and non-relaxed semantics.

Keywords

lock-free; nonblocking; priority queue; relaxed; skiplist; work-spying

Disciplines

Computer Sciences

Language

English


Share

COinS