[en] A comparative analysis has been made to describe the potentialities of hardware and software tools of two most widely used modern architectures of graphic processors (AMD and NVIDIA). Special features and differences of GPU architectures are exemplified by fragments of GPGPU programs. Time consumption for the program development has been estimated. Some pieces of advice are given as to the optimum choice of the GPU type for speeding up the processing of scientific research results. Recommendations are formulated for the use of software tools that reduce the time of GPGPU application programming for the given types of graphic processors