The article describes the experience of setting and solving the inverse problem of determining the velocity of shear waves under the Caucasus from surface wave dispersion curves using a high-performance distributed grid cluster. It is about options for adapting programs written for personal computers for modern high-performance clusters. Various models for assessing the velocity structure of the Caucasus and their suitability for the programs used are considered. It is concluded that the three-layer model of the crust does not fit the data well enough, which requires changes to the source program code. Also shows the fundamental possibility of using distributed computing systems such as GRID to solve inverse problems of Earth Physics.