A dynamic multiparameter method for analyzing the performance of a heterogeneous computing system designed for executing various resource-intensive tasks is proposed. The concepts of the single-task mode and normative performance indicators of the system are introduced. The set of various tasks executed simultaneously in one operational window is considered. To determine the maximum amount of work, a multiple criteria optimization model of efficient resource allocation of various types is used. The resulting Pareto optimal solution vectors provide unimprovable upper bounds on the functional capabilities of the system. At certain check points in time, the current maximum values of the performance indicators are compared to the normative ones. A diagram of relative deviations is constructed, which makes it possible to track the dynamics of the changes in the performance indicators. Processing and analysis of data over a long period of time provides multicomponent guaranteed estimates of the system's workability in the case of failure of its elements.