Event-based Program Analysis and Debugging with DeWiz

Title: Event-based Program Analysis and Debugging with DeWiz

Research Question: How can program analysis and debugging tools be made more scalable and user-friendly for parallel and distributed programs?

Methodology: The authors developed DeWiz, a novel debugging tool, to address the scalability and user-friendliness issues in program analysis and debugging. DeWiz uses an event graph model to represent the program's execution, which allows for modular and scalable analysis and visualization. It consists of several modules that can be combined to process event graph data, extract useful information, and present it to the user at an appropriate level of abstraction. Additionally, DeWiz serves as a framework for implementing user-defined modules.

Results: DeWiz demonstrated improved scalability and user-friendliness compared to existing tools. It successfully analyzed and visualized event graph data from parallel and distributed programs, providing valuable insights to the users. The tool's modular architecture and scalability make it suitable for handling large-scale programs that utilize many processors over extended periods.

Implications: The development of DeWiz has significant implications for the field of program analysis and debugging. It provides a new approach to handling the complexity of parallel and distributed programs, making the process more efficient and accessible to a wider audience. The tool's scalability and user-friendliness can lead to improved software development practices, resulting in more reliable and efficient software systems. Furthermore, DeWiz's modular architecture and framework functionality encourage further research and development in the field, facilitating the creation of customized analysis and debugging tools tailored to specific user needs.

Link to Article: https://arxiv.org/abs/0310007v1 Authors: arXiv ID: 0310007v1