We investigate a reconfigurable intelligent surface (RIS)-aided multi-user massive multiple-input multi-output (MIMO) system where low-resolution digital-analog converters (DACs) are configured at the base station (BS) in order to reduce the cost and power consumption. An approximate analytical expression for the downlink achievable rate is derived based on maximum ratio transmission (MRT) and additive quantization noise model (AQNM), and the rate maximization problem is solved by particle swarm optimization (PSO) method under both continuous phase shifts (CPSs) and discrete phase shifts (DPSs) at the RIS. Simulation results show that the downlink sum achievable rate tends to a constant with the increase of the number of quantization bits of DACs, and three quantization bits are enough to capture a large portion of the performance of the ideal perfect DACs case.