Complete Computer System Simulation: The SimOS Approach
IEEE Parallel & Distributed Technology: Systems & Technology
SMARTS: accelerating microarchitecture simulation via rigorous statistical sampling
Proceedings of the 30th annual international symposium on Computer architecture
An Evaluation of Two-Stage Systematic Sampling in Micro-Architecture Simulation
CHINAGRID '08 Proceedings of the The Third ChinaGrid Annual Conference (chinagrid 2008)
| Hi-index | 0.00 |
Due to the prohibitively long time when detailedly simulating a realistic benchmark to its completion, sampling is frequently used to reduce the simulation time. However, it may often require profiling or iterative simulations to determine the sampling parameters. This paper employs the generation procedure of trisection Cantor set, one classic fractal, to select instructions simulated in detail as an approach to enable a simple and fast micro-architecture simulation. Randomly selected six benchmarks from SPEC CPU2000 are tested on the simulator, CantorSim, which implements the trisection Cantor fractal approach. The results show that it is very easy to use this approach and it can achieve actual average acceleration of 23.4% over SMARTS [3] while the accuracy only reduces marginally.CantorSims accuracy is validated against the sim-outorder and is accurate in a 3.2% error margin. Similar CPI relative errors with the same parameter values of experiments on simulators of different processor models indicate that this approach is micro-architecture independent and can be applied to well predict the performance of new micro-architecture design.