The article demonstrates, by the example of solving a problem of a high degree of static indefinability that the efficiency of parallelizing calculations drops drastically as the number of cores increases. This limits considerably a possibility to design computer systems capable to simulate reliably enough normal and abnormal operation of real technical objects by means of widely used multiprocess hardware. To solve a wide range of problems of deformed solids mechanics, a variant is offered to modernize the method of dividing an object into design units, suggested by the authors earlier, and to divide it into continuums, that provides practically unlimited possibilities for parallelizing a calculation process.