Folding Forum

I've started using F@H, and it works fine, except when I go to any setting above 'Light', the GPU starts working (as I expected it to) but even at medium, the the two programs I monitored it with show that the GPU was running at just under 100%, obviously causing the temperatures to skyrocket, and everything became sluggish.

Is this supposed to happen? If so, what's the difference between medium and full? I assumed this was a bug.

CPU: i7 4790
GPU: GTX 970 Asus Turbo
Programs used to monitor: TThrottle and 'Asus GPU Tweak' (the default program that comes with the gpu)

The Medium setting applies to the CPU slot, by its nature GPU's are either processing or not processing. Attempts have been made in the past to modulate GPU usage, they were not particularly successful and were dropped.

Starting in the 1950s, intensive work to prioritize programs on CPUs and to prevent one user process from bothering another use's data began. Now in the 2010's we are pretty good at it.

Sadly, we have not begun to prioritize or separate GPU processes. The assumption is that you want it all, and for gamers, that is the correct assumption. Folders do not make a significant percentage of sales to have GPU drivers that can treat process A as higher priority than process B.

Hello Tom_Bombadil, welcome to the Folding Support Forum.

Yes, this is intentional, but not as you might think. Unfortunately, the design of the GPUs does not allow for more granularity when using the device to fold. GPUs are either full on, or full off, so that is what you are seeing on the slider. CPUs, unlike GPUs, do allow for using All, Part, or Little (or none) of the CPU cores.

Full, for the CPU, uses all the CPU cores (minus 1 to feed the GPU, if present). Medium uses 80% of the CPU cores, and Lite uses 50%. GPU is off at the Lite setting, GPU is on for the other 2.

If you are seeing lag while use the GPU, right click on the GPU slot, and select the option to Fold while Idle. Then the GPU only runs FAH while you are not using the computer.

The top-of-the-line GPUs from both nV and AMD are touting features that may eventually be able to moderate GPU usage and to prioritieze screen updates over compute tasks like FAH. Those features are in their infancy and only on the most recent GPU designs. Also, their stated goal is improving performance for games and for VR.

Like JimboPalmer said, CPUs have had 60+ years to develop the kind of features you're looking for and GPUs may develop them eventually[/u] but not soon, and not retroactively to older GPUs. The <idle v='true'/> feature mentioned by 7im is the best that FAH can offer with today's technology.