The problem of maximizing weighted sum rate in the MIMO multiple access channel with individual power constraints is considered. The optimum is achieved by successive interference cancellation, where the covariances are found by iterative water-filling. As successive interference cancellation implies long decoding delays, we consider linear approaches with zero-forcing constraints. To avoid the associated non-convex and combinatorial optimization, we allocate successively data streams to users, while keeping transmit filters and user allocations of previous steps fixed. The transmit filters are determined based on two lower bounds for the weighted sum rate. The algorithms converge to the optimum linear solution for infinite transmit powers in many scenarios at low computational complexity.