This article discusses problem of improving performance of large dimension random number generators. With current trends , as an accelerator for CPU considered field-programmable gate array ( FPGAs ) . The analysis of hardware implementations of algorithms of random number generators offered by the Altera company is carried out. It is offered to use a method of calculation of a gain of productivity Reconfigurable Amenability Test (RAT) offered by B. Holland, K. Nagarajan and A. D. George. Assessment of several random number generators is made by this method. Received results it is analyzed, and on the basis of the drawn conclusions, simulation of the algorithms showing a high rate of a gain of productivity is made. Results received theoretically and during simulation are compared.

Keywords: Reconfigurable amenability test, field-programmable gate array, random number generator, high-performance calculating performance prediction