As technology nodes downscaling into sub-16 nm regime, the industry relies heavily on Optical Proximity
Correction (OPC) to improve mask pattern transfer fidelity and process window for optical micro-lithography. However, OPC
computational time becomes a crucial factor since mask data have to be prepared in manner of hours to cover
the huge industrial needs. Nevertheless, there is a trade-off between mask image accuracy and OPC simulation
time. In this paper, we propose a new grid based algorithm to improve the accuracy of simulated mask image with the least
number of convolutions during OPC process, which is proportional to OPC computational time. We analyze our algorithm and the observations we obtained on
public benchmark released by IBM for ICCAD 2013 CAD contest.