Not sure how wide spread this issue is, but I've noticed on the N-Body Simulation units running with 2 cores and a description ending with "(nt)" (I think that's what it was) that I'll get to about 0.37% before my remaining time estimate takes off. Before aborting, one task was estimated at over 700 days! My computer is a custom build and is a few years old, but up until very recently, it's been able to support all but these new work units without issue. I've had it running several other projects (WCG, Rosetta, LHC, MLC, etc) and they all seem to run fine. I think there may be a bug in these work units. When I see ridiculous estimates such as 700 days, it leaves me to believe a process got stuck in a loop. I've since adjusted my compute preferences the Separation 1.46 tasks since they've been able to run without issue.

The N-body tasks are (MT) multi-thread and normally would run with all your cpu threads available. So limiting them to only two cpu threads will slow them down a bit.

But more to the point . . . . until you have completed 11 valid N-body tasks, the estimated times to completion will be completely invalid.

Only after the 11 valid tasks will the APR for the app on that host has been calculated will the estimated times to completion will be accurate.

Since your computers are hidden, I can't tell if you have reached that point yet.

---

The computers should be visible now. As far as the limiting - I had forgot that I reduced the number of cores that BOINC was allowed to use, so that makes sense. I'll have to adjust those settings to allow the sims to run quicker.