Using GPUs for runtime monitoring and trace analysis

使用 GPU 进行运行时监控和跟踪分析

• 批准号: 479865-2015
• 负责人: Fischmeister, Sebastian
• 金额: $ 6.37万
• 依托单位: University of Waterloo
• 依托单位国家: 加拿大
• 项目类别: Collaborative Research and Development Grants
• 财政年份: 2015
• 资助国家: 加拿大
• 起止时间: 2015-01-01 至 2016-12-31
• 项目状态: 已结题
• 来源: https://www.nserc-crsng.gc.ca/ase-oro/Details-Detailles_eng.asp?id=580332
• 关键词: Using; GPUs; runtime; monitoring; trace

Computer software is growing exponentially in size and thus complexity. Despite this growth in complexity, software still must be correct, which means that the software must comply with a given specification and is thus free of software defects. Runtime monitoring is a technique that checks the correctness of the software at run time on the deployed system. With the advent of general purpose computation on graphics processing units (GPUs), the question becomes how this new technology can be leveraged for runtime monitoring systems, especially in combination with trace analysis a combined approach. This project will investigate the use of GPU for runtime monitoring and accelerating algorithms for trace analysis, and develop novel recovery techniques using the independence of GPU systems from applications running on the main processor. These results will raise Canada's visibility in systems research and strength Canada's role in research. Besides the expected scientific publications, the industrial sponsor AMD agreed to help promote the developed library and tools if they meet AMD's requirements, that will further raise the impact of the project.

计算机软件的规模和复杂性呈指数级增长。尽管复杂性增加，软件仍然必须是正确的，这意味着软件必须符合给定的规范，因此没有软件缺陷。分布式监视是一种在部署的系统上运行时检查软件正确性的技术。随着图形处理单元（GPU）上通用计算的出现，问题变成了如何将这种新技术用于运行时监控系统，特别是与跟踪分析相结合。 该项目将研究使用GPU进行运行时监控和加速跟踪分析算法，并利用GPU系统与主处理器上运行的应用程序的独立性开发新的恢复技术。这些成果将提高加拿大在系统研究方面的知名度，并加强加拿大在研究方面的作用。除了预期的科学出版物，工业赞助商AMD同意帮助促进开发的库和工具，如果他们满足AMD的要求，这将进一步提高该项目的影响。