Computing Folding@home efficiently
After thinking about efficiently computing for Folding@home, we decided to change the setup of our GPU from it's default state.
As our GPU is the exact model that was used for efficiency experiments conducted by Gorgon [1], we just replicated setting our 'clock speed' to the values found there. This was done via the following Bash command, where the GPU_ID was found using nvidia-smi -q.
GPU_ID = GPU-5a404...
upper = 1250
nvidia-smi -i $GPU_ID -lgc 0,$upper
If you have a differing GPU model, running that experimental procedure by Gorgon, to determine if your GPU's efficiency curve is significantly different, is advised. If you choose not to do that experiment, set your GPU's clock to approximately 90% of it's normal (base) value.
[1] https://linustechtips.com/topic/1433841-howto-profiling-folding-gpus/