As traditional compact MOSFET models have many unphysical fitting parameters, they cannot be used for electrical characteristics prediction with process condition change. Moreover, they cannot be used for physical variability extraction from electrical characteristics variation, either. Although TCAD has the same potential capability, they are too slow for circuit simulation. In this paper, a reverse-profiling based 1st principle compact MOSFET model HiSIM-RP is presented. HiSIM-RP does not have any unphysical fitting parameters and has 1,000~10,000 times faster calculation speed than TCAD. Good predictability of HiSIM-RP has been demonstrated in the case of channel profile change of 90nm transistor. Moreover, 90nm and 40nm CMOS electrical characteristics variability has been analyzed by HiSIM-RP. It has been clarified that random dopant fluctuation and external source/drain series resistance variation are the principal components for random variation of electrical characteristics. As for the systematic variation of the electrical characteristics, it has been clarified that gate length, dopant non-uniformity and external source/drain series resistance variation are the principal components.