Different CPU\GPU usage was recorded when rendering the project on the same computers

[DLepeshko]

Good afternoon,

When using Adobe After Effects software in conjunction with Adobe Media Encoder with the same settings on 3 hardware configurations (differences at the chipset and video card level), different capacity usage was recorded when rendering the project on different computers.

In particular, the difference was expressed in the following:

On PC1 (i7-8700, 32B DDR4, GTX 1070 Ti 8GB), both when the "Mercury Playback Engine GPU Acceleration" is turned on, and when the "Mercury Playback Engine Software Only" is selected, the load during the whole process occurs only on all processor threads, and the video card is used at a maximum of 1-2%.

On PC2 (i7-8700, 32B DDR4, RTX 2070 8GB), when you turn on the "Mercury Playback Engine GPU Acceleration" or when you select "Mercury Playback Engine Software Only", the load during the entire process only affects all processor threads, and the video card 1-2% maximum is used.

On PC3 (Ryzen 7 3800X, 32B DDR4, RTX 2070 8GB), when you turn on the "Mercury Playback Engine GPU Acceleration" or when you select "Mercury Playback Engine Software Only", the load during the entire process occurs only on 1 processor core, and the video card is used in both modes up to 40%.

On PC4 (i7-8700, 32B DDR4, RTX 2070 8GB), when you turn on the "Mercury Playback Engine GPU Acceleration" or when you select "Mercury Playback Engine Software Only", the load during the entire process occurs only on 1 processor core, and the video card is used in both modes up to 40%.

In accordance with the equipment load, rendering was faster in the cases where the video card was used. The built-in Windows 10 utility and the latest version of HWMonitor were used for data monitoring (from CPUID).

The video card drivers were updated to the latest (as of the test date) version, the chipset drivers were taken from manufacturers' websites (as well as the BIOS firmware versions), and the latest operating system updates were installed.

Thus, we would like to understand:

1) Why is there such a difference in working with the same project on 4 computers?

2) Why the equipment load varies during the rendering process?

3) What is the correct working mode for it?

4) How many kernels / threads should the software use, and in what modes?

5) If it depends on the settings, how to configure the proposed configurations?

[Test Screen Name]

That's interesting study. I hope you'll be able to use it to optimize your own workflow. I doubt anyone anywhere has answers to the kind of question you are asking, though.