Future wireless multimedia systems will support a variety of services with diverse range of capabilities and bit rates. For these systems, it is highly desired for real-time conversational and non-real-time services to efficiently share the available channels and bandwidth in an optimized way. The partitioned resource shaping with either fixed or a slow changing dynamic, proposed for conventional packet scheduling techniques, proves difficult and inefficient under fast-changing dynamics of radio channel and traffic. By taking into account almost all the aspects (dimensions) of quality-of-service (QoS) provisioning, the proposed unified fast dynamic multidimensional QoS-based packet scheduler (MQPS) in this paper elegantly and efficiently encapsulates features of many possible packet scheduling strategies. MQPS applies an optimization and tuning mechanism to packet scheduling weights to adopt the most appropriate packet scheduling and channel assignment strategy in response to the varying traffic and radio channel conditions. As an example, the technique is applied to a high-speed downlink packet access (HSDPA) system. It is shown that MQPS provides significantly better performance than existing techniques by satisfying all the requirements of a successful QoS provisioning to maximum possible level simultaneously.