The computational schemes that are supposed to be used in the processes of processing operational information in real-time systems are considered. Two algorithms for scheduling jobs that do not allow interruption and switching are proposed in real-time multiprocessor computing systems. The first algorithm, based on the branch and bound method, solves the speed problem. Processors may differ in their performance. The main feature of this algorithm is that the lower and upper estimates of the length of the optimal schedule are calculated using recurrent relations, which increases its efficiency. The second algorithm solves the problem of finding an admissible schedule in the case when all processors are identical and the directive intervals of all jobs are the same. The algorithm has pseudopolynomial complexity when the number of processors is fixed. A description of the parallel implementation of the algorithm is given