Development of an effective model of parallel processing of multimedia data on the CPU and GPU in the cloud system

The paper deals with the technology of paralleling processing multimedia data process on the CPU and GPU of the user's personal computer. To do this, we have developed an algorithm for planning the sharing of video accelerators, which controls and distributes the load between the processor and the GPU. The main difference of the proposed algorithm is that the integral transformations include the stabilization of three-dimensional objects and textures. Defining a unique sequence number allows you to apply optimal texture filtering when scaling for each object. These actions create optimal video acceleration when processing objects.