There are many architectures for RSA hardware implementation which improve its performance. Two main methods for this purpose are Montgomery and RNS. These are fast methods to convert plaintext to ciphertext in RSA algorithm with hardware implementation. RNS is faster than Montgomery but it uses more area. The goal of this paper is to compare these two methods based on the speed and on the used area. For this purpose the architecture that has a better performance for each method is selected, and some modification is done to enhance their performance. This comparison can be used to select the proper method for hardware implementation in both FPGA and ASIC design.