The slip-step method is widely used in simulating optical wave propagation in turbulent atmosphere, which treats propagation and phase perturbations caused by turbulence separately and in discrete steps along the propagation axis. The phase perturbations are represented by a series of phase screens, and hence, the precision of the phase screen concerns the accuracy of the simulation. In this paper, we first discuss the precision and computational performance of phase screens generated by the subharmonic complemented discrete Fourier transformation (DFT) (DFT-SH) method, three kinds of randomized spectral sampling techniques (sparse spectrum (SS) technique, sparse spectrum technique with uniform wave vectors (SU), randomized DFT technique), and optimization-based (OB) method; then, the simulations are implemented with the phase screens generated by these methods. Some statistical quantities of the received optical field are calculated, such as beam wander variance, long-term beam radius, short-term beam radius, and on-axis scintillation index. The statistical results show that the undersampling of phase screen in the low-frequency region causes underestimation of the values of beam wander variance, long-term beam radius, and focused beam on-axis scintillation index because these quantities are sensitive to the large-scale inhomogeneities. However, the undersampling does not affect the predicted values of the short-term beam radius and collimated beam on-axis scintillation index because these quantities are insensitive to the large-scale inhomogeneities.