Hello.
I have multiple systems participating in GPUGRID. One if them has a GTX 1070, so long runs aren't an issue, the other is only running a GT 640, and its current estimate for one task is over 1.65 days. Not really good.

When i ticked the long run box after getting my GTX 1070 i made the assumption your servers (or BOINCs?) would note the ridiculousness of getting a GT640 to do these tasks and only send them to the GTX 1070 machine.

Is there any way to set the settings so that only the GTX 1070 machine gets long runs, and all others remain at short runs only?

We are in a work shortage, and since the work gets assigned first come first serve, I wouldn't expect any miracles. For now I would take the gt 640 off GPUgrid and abort all the tasks. Try to refresh the 1070 rig soon after you abort them. It is most efficient if we get only fast cards doing WUs during shortages.

Switch slow machine to different location such as "Work" and set short WUs only for this location.

I've just switched the location of one computer to 'work' and the other to 'home' so hopefully that'll resolve it. cheers.

Quick heads up though, the settings page is full of warnings & errors?
See this image.

Yes we know. Has been like that since moving servers a while ago. Not to worry minor problem.

I'll be putting a GTX 570 into the other (the current GT640), which is what the GTX 1070 replaced. Any idea if the 570 will be worth running long GPUGRID tasks compared to the 640? I previously only ran short tasks on the 570.

Will be faster than the 640 but will still take 2 days to run a normal long WU

Ah. I'd say that isn't worth the wattage compared to output then, the GTX 570 is a high W card, the 1070 uses about 65W less than it lol. I'll keep it to short runs and let the long runs be given to the faster computers in your network.