At the start of the pandemic in 2020, I was the director of the software team at Local Projects, where we had over 15 high-powered workstations with Nvidia GTX1080's or better.
As the office shut down, and we all started to work from home on our laptops or personal desktops, I set the idle machines to run Folding@Home targeting the SARS-COV-2 virus. I had two workstations at home, one for my CFD work and one for my real work, which I added to the cluster at evening time.
I also reached out to past clients whose museums and immersive video walls were not in use.
I received permission from some of them to remotely connect and configuring the folding client, using their electricity and cooling, yes, but contributing to possibly finding a vaccine.
This cluster was a small part (still, a decent amount of cluster power :)) of the project going exascale and finding 50 pockets for vaccine vectors.
At one point, we had 55 computers (572 CPU cores and 55 GPU's) in our cluster. Since I shut down the initiative when I stepped down from my position at LP, our team rank was pretty well- 27,714 of 2,879,363
50 pockets of change across the proteome were identified as targeting vectors for vaccines. Vaccine makers used that data to help develop their antiviral candidates.
The data went exascale! 0.1 seconds of simulation data were generated. At its peak, Folding@Home had 280,000 GPU's and 4,800,000 CPU Cores.