[en] The opportunities of technology CUDA (Compute Unified Device Architecture - the unified hardware-software decision for parallel calculations on GPU)of the company NVIDIA were described. The basic differences of the programming language 'C' for GPU from 'usual' language 'C' were selected. The examples of CUDA usage for acceleration of development of applications and realization of algorithms of scientific and technical calculations were given which are carried out by the means of graphic processors (GPGPU) of accelerators GeForce of the eighth generation. The recommendations on optimization of the programs using GPU were resulted.