The three-dimensional (3D) location optimization for reconfigurable intelligent surface (RIS) aided millimeter wave network is investigated. We first formulate the signal-to-noise ratio (SNR) maximization model by jointly optimizing the precoding vector, the RIS location and its parameter matrix in a multiple-input single-output downlink network. The optimal maximum ratio transmission precoding is applied, and the alternating direction method of multipliers is proposed for the highly nonlinear combinatorial problem. The subproblem with discrete variables has closed form solution, and the nonlinear square subproblem is solved via the Levenberg-Marquardt method. In simulations, the proposed model achieves the best SNR among all the compared models. And the RIS is suggested to be deployed near the base station, similar to the lower frequency band case.