Instruction level analytic prediction of parallel CPU architecture performance
暂无分享,去创建一个
We present a first version of a performance prediction method for compiler driven superscalar CPUs, founded on analytic calculations on the basis of an unlimited resource simulation of the benchmark program. We illustrate the accuracy of the results by comparison with real execution simulation for three benchmark programs.
[1] Ko-Yang Wang. Precise compile-time performance prediction for superscalar-based computers , 1994, PLDI '94.
[2] David R. Stiles,et al. Better than one operation per clock (panel): vectors, VLIW, and superscalar , 1990, ISCA '90.