Performance Technology for Productive, High-End Parallel Computing

Colloq: Speaker: 
Prof. Allen D. Malony
Colloq: Speaker Institution: 
Performance Research Laboratory, Department of Computer and Information Science, University of Oregon
Colloq: Date and Time: 
Tue, 2005-03-08 10:00
Colloq: Location: 
ORNL, Bldg. 5700, Room L202
Colloq: Host: 
Jeffrey S. Vetter
Colloq: Host Email:
Colloq: Abstract: 
High-End applications and systems are evolving towards more sophisticated modes of operation, higher levels of abstraction, and larger scales of execution. These trends challenge the state of technology for performance evaluation and optimization. While scalable toolsets for parallel and distributed system performance analysis exist, offering a range of capabilities and sophistication in performance observation and analysis, the tools are not as widely used by HPC application groups nor as broadly integrated in the application development, engineering, and tuning processes as they could and should be. Moreover, as applications grow to tens of thousands of threads, the human-centric, data-intensive approach of most tools will become cumbersome. This talk will discuss our current and future work on improving the utility of parallel performance tools and evolving their capabilities to meet high-end application requirements. New developments in the TAU parallel performance system will be presented, primarily in relation to how they are better directed to application-specific concerns. The talk will also outline our new efforts towards autonomic performance assessment and optimization that involve work in performance data mining and knowledge discovery to better integrate and automate the monitoring/data gathering -> performance analysis/problem diagnosis -> optimization/adaptation process.
Colloq: Speaker Bio: