Dynamic Concurrency Reduction Methods for Power Management using Petri-nets

Power consumption and heat radiation form the main limiting factors to high-performance microprocessors with multi-core architectures. Power management policies are the control procedures to adjust the per-core performances according to a power budget. When there is a power limitation, the cores will slow down or even shut off, reflecting the reduction of a system’s concurrency relations, which are defined in this paper. Furthermore, different static and dynamic control mechanisms are proposed for concurrency reduction, with the analysis of their influences on the system performance.

[1]  James L. Peterson,et al.  Petri net theory and the modeling of systems , 1981 .

[2]  Alexandre Yakovlev,et al.  A look at concurrency semantics through "lattice glasses" , 1989, Bull. EATCS.

[3]  Albert Koelmans,et al.  Petri Nets and Digital Hardware Design , 1998 .

[4]  John Sartori,et al.  Distributed peak power management for many-core architectures , 2009, 2009 Design, Automation & Test in Europe Conference & Exhibition.

[5]  Margaret Martonosi,et al.  An Analysis of Efficient Multi-Core Global Power Management Policies: Maximizing Performance for a Given Power Budget , 2006, 2006 39th Annual IEEE/ACM International Symposium on Microarchitecture (MICRO'06).

[6]  Luciano Lavagno,et al.  On the models for asynchronous circuit behaviour with OR causality , 1996, Formal Methods Syst. Des..