Real-Time Application Performance Steering and Adaptive Control
Abstract
High-performance computing is rapidly expanding from single parallel systems to distributed collections of heterogeneous sequential and parallel systems. The emerging applications are irregular, with complex, data dependent execution behavior, and dynamic, with time varying resource demands. The objective of the Real-time Application Performance Steering and Adaptive Control project is to replace ad hoc, post-mortem performance optimization with an extensible, portable, and distributed software infrastructure for real-time adaptive control that dynamically optimizes the performance of distributed applications. By integrating dynamic performance instrumentation and on-the-fly performance data reduction with configurable, malleable resource management algorithms and a real-time adaptive control mechanism, flexible runtime systems could automatically choose and configure resource management algorithms based on application request patterns and observed system performance. Such an adaptive resource management infrastructure can increase portability by allowing application and runtime libraries to adapt to disparate hardware and software platforms and increases achieved performance by choosing and configuring those resource management algorithms best matched to temporally varying application behavior. The Autopilot real-time adaptive control infrastructure is based on this thesis. Autopilot provides a flexible set of performance sensors, decision procedures, and policy actuators to realize adaptive control of applications and resource management policies on both parallel and wide area distributed systems.
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 01, 2002
- Accession Number
- ADA406840
Entities
People
- Daniel A. Reed
Organizations
- University of Illinois Urbana–Champaign