Thesis
Ctunes: a framework for self-adaptive, high performance parallel programming in distributed systems
Washington State University
Master of Science (MS), Washington State University
05/2016
Handle:
https://hdl.handle.net/2376/105681
Abstract
The growing ubiquity of distributed system requires programmers to write communication programs that can be executed in parallel on multiple computers connected by a network. In addition, multicore architectures require programmers to write concurrent programs that can be executed in parallel on multiple cores in one computer. The combination of communication programs and concurrent programs is an emerging programming architecture, which is known as hybrid programming, or two-level parallel programming. The traditional way of writing hybrid programs mixes network code, concurrency code, and functionality code. As a result, in order to exploit the parallelism of the underlying hardware, programmers must address critical challenges in their codes, such as transferring data, managing resources, and collecting the results. To simplify the task of writing efficient hybrid programs, we propose CTunes, a new software architecture for hybrid programming, which addresses these challenges by separating programs' concurrency and networking code from their functionality code, and providing APIs for both communication and concurrency control. Specifically, there are two main components of the proposed new software architecture of hybrid programming: a communication model and a programming paradigm. The new communication model handles distribution of computational tasks to nodes (or computers), and the new programming paradigm optimizes the level of concurrency of a computation at run-time based on predefined tuning policies in each node. Experimental results show that CTunes is effective in achieving high performance on high-end computing architectures, yet it does not introduce extra overhead on the hardware which does not provide high level of parallelism.
Metrics
2 File views/ downloads
12 Record Views
Details
- Title
- Ctunes
- Creators
- Tai The Nguyen
- Contributors
- Xinghui Zhao (Chair)Scott Wallace (Committee Member) - Washington State University, Engineering and Computer Science (VANC), School ofSarah Mocas (Committee Member)
- Awarding Institution
- Washington State University
- Academic Unit
- Electrical Engineering and Computer Science, School of
- Theses and Dissertations
- Master of Science (MS), Washington State University
- Publisher
- Washington State University; [Pullman, Washington] :
- Number of pages
- 72
- Identifiers
- 99900525114801842
- Language
- English
- Resource Type
- Thesis