Frequently downloading work units with expirations too short

[MajorCaliber]

I'm not sure what to do in your case. You are on the right track with deleting the slots and re-creating them. I use one CPU slot and one GPU slot, so I only have to delete and re-create the GPU slot which is the only one that ever has the problem. I don't know how it would work with two CPU slots. That may be the reason for your problem. Why do you have two CPU slots running? I'm hardly an expert on the internals of the program, but I would expect that your available CPU cycles are being split between the two slots so each takes twice as long to complete. This all would come out in the wash if it were not for expiration, but it is a problem with it. I would try deleting one slot and letting the remaining one go at full speed and I would expect it to finish twice as fast. I have a single CPU with 4 cores and 8 threads. They can all work on one slot and one work unit at the same time. If you truly have two CPU's perhaps they cannot both work on one work unit at the same time? Can you not just dump the one slot with the doomed work unit? On the other hand If I misunderstand the workings of the program, would somebody with more knowledge please set me straight.

[foldy]

You can install linux on mac mini to use GPU if supported

[PantherX]

Welcome to the F@H Forum sejtam,

Would it be possible for you to report what project that was? Reason is that we might be able to do something on the Project level to prevent those Projects from being assigned but there's no guarantee given the constraints.

[PantherX]

MajorCaliber has the correct understanding, 1 CPU Slot per system where you can allocate as much of the CPUs as you want to fold.

While there's no way to select WUs with short deadlines, sharing the Project details might allow us to talk to the researchers and potentially come up with a solution.

[bruce]

I'm not sure what to do in your case. You are on the right track with deleting the slots and re-creating them. I use one CPU slot and one GPU slot, so I only have to delete and re-create the GPU slot which is the only one that ever has the problem. I don't know how it would work with two CPU slots. That may be the reason for your problem. Why do you have two CPU slots running? I'm hardly an expert on the internals of the program, but I would expect that your available CPU cycles are being split between the two slots so each takes twice as long to complete. This all would come out in the wash if it were not for expiration, but it is a problem with it. I would try deleting one slot and letting the remaining one go at full speed and I would expect it to finish twice as fast. I have a single CPU with 4 cores and 8 threads. They can all work on one slot and one work unit at the same time. If you truly have two CPU's perhaps they cannot both work on one work unit at the same time? Can you not just dump the one slot with the doomed work unit? On the other hand If I misunderstand the workings of the program, would somebody with more knowledge please set me straight.

If you are configuring two CPU slots on he same system, that's the wrong thing to be doing. A single CPU with 8 threads can be configured so that all 8 threads work on a single assignment or two slots can be allocated with 4 threads each. The latter is wrong thing to do because you'll get two different WUs assigned and they'll both run at half speed.

I don't see where you posted the top ~100 lines of your log so we can only guess how your system is configured. See below if you need help finding that information.

Perhaps there is a problem with terminology. I haven't seen a Mac-Mini in many years, but my memory is that they contain a dual CPU in a single socket. FAH calls them CPUs but they're really not in separate sockets so they're treated like separate threads even though each has a dedicated FPU rather than being hyperthreaded devices sharing a single FPU. If memory serves, they all can be allocated to work on a single assignment in a single slot.