Aggravating as this may sound and rather presumptuous on my part to speak of behalf of, but eventually choices had to be made over which volunteer devices to target and not compromise the object of this research, to also compute bigger molecules, real or synthesized. If 18 hours is not enough to get through even 1 of 8 jobs, the heaviest at start, than maybe consider your device was not made for this 'opt-in' science. If the validator were to be set to credit for even an incomplete job #0, it would invite others to go compute cep2 when clearly not fit for it. Read the instructions, then opt a device in or out. Just tested on linux with an ancient and one cpu thread and it has a very hard time with the first part, so decided it will not run this project routinely, opted-out.

What we're 'waiting' for is wcg developing skill to target heavy sets of tasks to heavy machines and lighter to, well less powerful. This would open up cep2 to a wider range of devices and consequently speed up the research. That ball is thoroughly in wcg/cleanenergy court, if they're interested to play that ball.

1) cleanenergy teaching wcg how to recognize heavy batches/molecules (pre-run a task for each batch for instance)
2) wcg classifying volunteer devices based on for instance their benchmark and active/on_frac hours.

Then we would not have lost time. But how big is the problem fraction, is it 0.01 percent, 0.1 percent 1.0 percent, 5 percent, 10 percent and is this worth the effort in a context of for instance doubling the cep2 participation, again if cleanenergy is even interested, or can even handle that much incoming data?