The methods of independent tasks allocations which are coming at the same time for running on the parallel processors of computing cluster are developed. The methods of job allocation for homogeneous and heterogeneous cluster architectures, different methods of pre-sorting of the input stream of tasks are developed. It is shown that to achieve the desired level of load balancing for the cluster inter-processor with a homogenous architecture advisable distribution of tasks with a preliminary sorting of the input stream is descending their duration, for a cluster with heterogeneous architecture - using sorting tasks in descending order of density, used to construct a feasible schedule assignments that optimizes power consumption of the processor. The results of computational experiments confirming the provision of a high level of load balancing through the use of the proposed methods for different amounts and types of processors of computing cluster.
task, allocation, frequency scaling, sorting, duration, task density, computing cluster, the average utilization rate, processor