SMARTS Simulation Sampling

SMARTS is a framework that applies statistical sampling to accelerate microarchitecture simulation. It employs systematic sampling to measure only a very small portion of the entire application being simulated. The accelerated simulation rate allows complete benchmarks to be used in performance studies. SMARTS is a part of SimFlex. SMARTS/TurboSMARTS and SimFlex tools are available for download.

• Students
  • Roland Wunderlich (PhD Thesis)
  • Thomas Wenisch (advised by Babak Falsafi)
• Publications
  • SimFlex: Statistical Sampling of Computer System Simulation. Thomas F. Wenisch, Roland E. Wunderlich, Michael Ferdman, Anastassia Ailamaki, Babak Falsafi, and James C. Hoe. IEEE Micro, Volume 26, Number 4, July/August 2006. (pdf) (note: simulation sampling for multiprocessors.)
  • Statistical Sampling of Microarchitecture Simulation. Roland E. Wunderlich, Thomas F. Wenisch, Babak Falsafi, and James C. Hoe. ACM Transactions on Modeling and Computer Simulation, Volume 16, Number 3, June 2006. (pdf) (note: this is the much extended journal version of ISCA 2003.)
  • Simulation Sampling with Livepoints. Thomas F. Wenisch, Roland E. Wunderlich, Babak Falsafi and James C. Hoe. International Symposium on Performance Analysis of Systems and Software (ISPASS), March 2006. (pdf) (note: simulation sampling using checkpoints.)
  • SMARTS: Accelerating Microarchitecture Simulation via Rigorous Statistical Sampling. Roland E. Wunderlich, Thomas F. Wenisch, Babak Falsafi, and James C. Hoe. International Symposium on Computer Architecture (ISCA), June 2003. (pdf)
  • Applying SMARTS to SPEC CPU2000. Thomas F. Wenisch, Roland E. Wunderlich, Babak Falsafi, and James C. Hoe. CALCM Technical Report 2003-1, June 2003. (pdf) (note: this is the tech report version of 2003 ISCA paper with the complete SPEC data set.)

• Thesis
  • Statistical Sampling of Microarchitecture Performance Simulation. Roland Wunderlich, PhD, ECE/CMU, May 2010. (pdf)