Data Cache Techniques to Save Power and Deliver High Performance in Embedded Systems

  • Authors:
  • Major Bhadauria;Sally A. Mckee;Karan Singh;Gary S. Tyson

  • Affiliations:
  • Computer Systems Lab School of Electrical and Computer Engineering, Cornell University,;Computer Systems Lab School of Electrical and Computer Engineering, Cornell University,;Computer Systems Lab School of Electrical and Computer Engineering, Cornell University,;Department of Computer Science, Florida State University,

  • Venue:
  • Transactions on High-Performance Embedded Architectures and Compilers II
  • Year:
  • 2009

Quantified Score

Hi-index 0.00

Visualization

Abstract

Minimizing power consumption continues to grow as a critical design issue for many platforms, from embedded systems to CMPs to ultrascale parallel systems. As growing cache sizes consume larger portions of the die, reducing their power consumption becomes increasingly important. Voltage scaling reduces leakage power for cache lines unlikely to be referenced soon. Partitioning reduces dynamic power via smaller, specialized structures. We introduce a reuse distance (RD) drowsy caching mechanism that exploits temporal locality, delivers equivalent or better energy savings than the best policies from the literature, suffers little performance overhead, is simple to implement, and scales with cache size and hierarchy depth.