In this paper we consider the problem of computing the stationary distribution of nearly completely decomposable Markov processes, a well-established area in the classical theory of Markov processes with broad applications in the design, modeling, analysis and optimization of computer systems. We design general classes of algorithmic solution approaches that exploit forms of mixed-precision computation to significantly reduce computation times and that exploit forms of iterative approximate methods to mitigate the impact of inaccurate computations, further reduce computation times, and ensure convergence. Then we derive a mathematical analysis of our general algorithmic approaches that establishes theoretical results on approximation errors, convergence behaviors, and other algorithmic properties. Numerical experiments demonstrate that our general algorithmic approaches provide significant improvements in computation times over the most efficient existing numerical methods.