I have a Alienware Windows computer running High Fidelity. For academic research, we’re using this VR software to allow multiple people to occupy the same virtual space and make design decisions. We had our first pilot today, which is the first time all three of our rendering computers were running for >1h. In addition to handling screen-recording software and VR rendering, one computer is also the server for the other two computers. That computer gave us a BSOD today mid-experiment.
Normally that wouldn’t be enough to go on, but a few minutes before that point, all three participants noted the items they had drawn where “flashing” in and out of existence. The drawings were made of small 3D cylinders, and I’m intuiting that HF won’t draw smaller / farther-away objects when the GPU load is really high. So for some reason the GPU load was cycling between low and high, and then after a few minutes of that, the computer acting as the server BSOD’d. I don’t know for sure if they’re related, but I think it’s likely.
So for the next session, I kept tabs on computer usage with Windows Task Manager, and saw a really weird effect. GPU usage was consistently 66-67% according to Task Manager, but when I sorted all processes by GPU usage, the HF process oscillated between using either 67% or 0% of the GPU - but no other process was high on GPU when HF was low. When HF showed 0%, all other processes read 0% GPU usage.
How do I stop the ‘flashing’ effect and have HF render more consistently?
Is the crash related to the GPU usage?
How do I prevent future crashes?