Abstract: In this article the author considers the development of a parallel algorithm to speed up calculations in an applied scientific problem. Particular attention is paid to software code profiling tools. The use of such tools allows you to quickly select portions of code for parallelization. The libraries and technologies of parallel code implementation for multiprocessor systems with shared memory are considered. Discusses approaches to writing programs based on threads and task-based. The stages of developing a parallel code by changing the sequential code are considered. An analysis of the execution time of application functions using the VTune tool is given. A method for parallelization of tasks based on the TBB library is proposed. Shows the implementation of the code allows you to perform parallel calculations. As a result of this work, a new parallel algorithm was developed to solve the problem of finding the characteristics of internal waves in the ocean within the framework of the weakly nonlinear theory. The obtained algorithm was analyzed, metrics of efficiency were obtained. Acceleration has been achieved, which allows making calculations of maps of two-dimensional distributions 39 times faster than a sequential algorithm on a 56 nuclear computing server. The results obtained will be applied further in the study of internal waves in the ocean, as well as will increase the effectiveness of research in the calculation of maps of parameters of a larger scale.