Plenty of numerical algebra libraries have been developed in recent years. These libraries are tuned for the given CPU and its memory architecture, fully utilize its memory hierarchy and inner pipelines and achieve impressive computation power. There is a new trend in the highperformance computing: GPU computing. This trend is caused by the surprising fact that the most powerful part of modern Intel PCs is not the CPU, but the GPU. Modern graphic cards (shortly GCs) overcome modern CPUs in the memory bandwidth, the number of computational units and possibilities of the vector execution. It results in their surprising floating point performance. In this paper, we have compared advantages of CPU and GPU computation. We will discussed possibilities of the GPU computation and demonstrate them on some program from linear algebra package.