There are some pretty good FAQ at the Stanford website ... and I like https://en.wikipedia.org/wiki/Folding@home
The folding process depends on the forces of attraction/repulsion between atoms and probably would follow Newton's Laws at absolute zero, but proteins don't spend much of their life at absolute zero. At "reasonable" temperatures, atoms move somewhat randomly. (See https://en.wikipedia.org/wiki/Brownian_motion
). FAH takes both of those effects into account so that the mathematical results resemble the physical results.
There is a constant exchange between potential energy and kinetic energy as the protein seeks to settle from a high energy state to a minimum energy state. There are lots of bumps and hollows along that road and a protein spends a lot of time sort of stuck in an energy hole until the random energies all happen to add up to enough to get it over one of the surrounding hills, at which point it moves rapidly to the next energy hole. (Look carefully at the Markov state model image on the Wikipedia article and imagine you're looking down on a rough surface with those bumps and hollows are pointing up at you.) There are many paths from the unfolded state to the folded state, and along each path, the protein will spend some time in the colored states before following an arrow toward the folded state.
Along each path, most of the time is spend stuck ... i.e.- not folding -- before being kicked out of that hole. Asking about the overall folding speed has very little meaning.
Generally, FAH starts anumber of Run,Clone locations on that bumpy surface and maps out the popular hangouts denoted by the colored shapes. Once there's a sense of where the protein dwells, a new project can be started from each of those energy holes to study all of the possible surrounding arrows, finding new ones, and mapping how strong each path is. i.e.- The randomness of the process is better studied if one can filter out the important jumps between local energy minimums from the unimportant ones.