Abstract
The optimization algorithms based on global techniques like evolutionary algorithms tend to be time consuming due to a very high number of objective function evaluations. The computational effort can be reduced when the parallel approach is used. The paper is devoted to parallel versions of optimization algorithm developed for a multicore computer, manycore coprocessor and a supercomputer. The idea of dividing tasks between available cores is described. The optimization of carbon based flat structures problem is used as a test problem in the paper. The paper is focused mainly on the efficiency and scalability of the proposed algorithms.