Low GPU usage at rendering animation on Google Colab

[vaimd]

This script is really impressive!
I'm new at VR. I was just wondering.
I used Google Colab to render animation. But, it seems that eeVR doesn't use GPU so much. At "nvidia-smi" command, GPU use-rates are between 0 - 6%.

Am I doing something wrong?
or Is it expected behavior?

To Reproduce
Steps to reproduce the behavior:

1. Install blender-2.83.3-linux64 to Ubuntu 18.04.3 in Google Colab through VNC viewer
2. "vglrun ./blender"
3. start rendering animation with eeVR ( output resolution setting is 2048x2048 for VR180 )

Expected behavior
I hope it uses GPU near 100%.
( Normal eevee used over 60% )

System information (please complete the following information):

• Platform: Google Colab
• OS: Ubuntu 18.04.3
• Processor: Intel(R) Xeon(R) CPU @ 2.30GHz
• Graphics card: Tesla P100-PCIE-16GB
• Blender version: blender-2.83.3-linux64

[EternalTrail]

this might be due to the fact that this script can't be parallelised, as it renders one image at a time. I could look into how this could be solved in the future. For now I'm marking this as an improvement for the future.

[kice]

@EternalTrail Could you write a guide on how to run multiple blender instances to make it "parallelized"? Like splitting the video into two parts with two instances running; or running two instances with each instances render one eye (, and then let users stitch together them afterward).

I know there will be a lot of limitations, like requiring more GPU memory, and maybe blender will be more subject to crashes. But I think it might easily cut a lot of rendering time by utilizing the hardware more.