Performance introspection in multi-threaded processors

01 October 2009 → 30 September 2019
Regional and community funding: Special Research Fund
Research disciplines
  • Natural sciences
    • Computer system architecture
    • Performance modelling
    • System software and middleware
    • Computer architecture and networks not elsewhere classified
    • Language processors
    • Programming languages and technologies
  • Engineering and technology
    • Computer architecture and organisation
    • Memory structures
    • Performance evaluation, testing and simulation of reliability
    • Processor architectures
    • Computer hardware not elsewhere classified
system software computer architecture performance modeling
Project description

Modern processors execute multiple software threads concurrently to improve hardware utilization. This leads to non-dependable performance: threads that execute concurrently affect each other's performance in unpredictable ways. This project investigates techniques to estimate per-thread progress and leverage these predictions in system software and hardware to better manage hardware resources and improve per-thread performance as well as overall system performance.