FEATURES OF USING PARALLEL CALCULATIONS ON GPU IN ADOBE AFTER EFFECTS ENVIRONMENT

Abstract

The robots have some peculiarities of parallel calculation on the basis of the computer graphics subsystem in the Adobe After Effects medium. The fact that the number of parallel charges on the GPU can significantly change the hour of video processing is accepted by the majority of the processors and plug-ins, as it is without any problems, however, in practice, the efficiency of the GPU in parallel with the charges on the CPU becomes far less important.

The analysis showed that Adobe After Effects supports a number of ways to use GPGPU (General-Purpose computing on Graphics Processing Units), especially CUDA, OpenCL and Metal [1]. I will confirm the assessment of the capacity of the GPGPU in this case to ensure the improvement of two warehouses:

1) good understanding of the problem of parallelization of programs, which is related to the security of the correct sequence in interdependence between different processes and the coordination of resources that are distributed between them;

2) the dichotomy of programs that hack CUDA, OpenCL or Metal, is represented by a standard code for the CPU and a code for the GPU (kernel).

In this article, an additional factor has been added, which contribute to the efficiency of parallel computing on the GPU and to the impact of these factors on the results of the work of various GPGPU technologies in the case of typical operations in the middle of Adobe After Effects. The results of the follow-up can be reviewed at the stage of configuring the graphics subsystem of the computer for work with the medium.