Inverse lithography technology can improve printed pattern fidelity, but computation time has limited its use across entire chips. A memory-manufacturing evaluation explores how GPU acceleration, stitchless optimization and multi-beam mask writing can make full-chip curvilinear ILT practical.