50 likes | 140 Views
Real-Time Multi-core Scheduling With Power, Thermal And Reliability Awareness. Advanced Real-time and Computing Systems (ARCS) Lab Electrical and Computer Engineering Florida International University. Ming Fan. Problem Formulation.
E N D
Real-Time Multi-core SchedulingWith Power, Thermal And Reliability Awareness Advanced Real-time and Computing Systems (ARCS) Lab Electrical and Computer Engineering Florida International University Ming Fan
Problem Formulation • We study the problem of scheduling real-time applications on multi-core systems with consideration of power/energy, thermal and/or reliability awareness.
Methodology • Real-Time Multiprocessor Scheduling. • Proposed an innovative metric to measure the harmonic relationship among different real-time periodic tasks; • Developed novel partitioned/semi-partitioned real-time scheduling algorithms for independent real-time tasks ; • Analytically proved the feasibility and utilization efficiency of the proposed algorithm; • Conducted extensive simulation studies to investigate the performance of the proposed algorithms. • Thermal/Power Aware Embedded System Design 01/2012 - present • Developed new dynamic thermal management strategies for on-line multi-core scheduling; • Analytically solved the fundamental problems of multi-core systems on both temperature and energy calculations; • Designed and developed new techniques to minimize the energy consumption for real-time applications. • Reliable and Fault-Tolerant System Design • Studied the problem on how to maintain high reliability for real-time computing systems; studied the real-time scheduling with fault-tolerance constraint; • Developed a real-time scheduling algorithm for real-time tasks with data dependency to minimize the energy consumption under guaranteed reliability constraints.
Application Tools • Software simulation • Hotspot, Wattch • Matlab, MS Visual Studio. • Hardware verification • Intel i5 quard core • Other tools • SimpleScalar • TGFF • PSPISE
Research results • Partitioned/Semi-Partitioned Real-Time Scheduling on Homogeneous Multiprocessor Systems by Exploring Harmonic Relationship Among Real-Time Tasks • Partitioned scheduling. (EUC’2011) • Semi-partitioned scheduling. (DATE’2012, TPDS’2013) • Multiprocessor Real-Time Scheduling with Thermal/Power Awareness • Throughput maximization. (DATE’2012) • Feasibility analysis. (DAC-Wip’2013) • Analytical temperature calculation. (DAC-Wip’2013) • Analytical energy calculation. (RTSS-Wip’2013) • Real-Time Scheduling with Reliability and Fault-Tolerance Constraints • Reliable scheduling for DAG-based real-time applications. (under preparing) • Real-time scheduling with fault tolerance. (ISLPED’2013, submitted)