In this paper, we extend the service-speed competition game with boundedly rational customers considered in Li et al. (2016) to the case with general reward function and multiple servers. The N servers make strategic decisions on their service rates sequentially and repeatedly. Since the competing servers' payoff functions can only be expressed by an implicit-function set, we propose a matrix method to derive the uniqueness of the equilibrium service rates, and we establish the stability of the equilibrium through a tatônnement scheme. By conducting a sensitivity analysis regarding the number of competing servers and the demand density, we find that the server competition benefits the customers by improving their utilities as well as getting more customers to be served. Furthermore, for a fixed demand density, the equilibrium service rate increases in the market size and converges to a certain level when the market size is large enough.