The search will now focus on the real subfields. For our chosen bound, there are 49 such subfields. The search starts with the one having discriminant 161.

I forgot to mention-- You may notice some WUs taking a very long time. I ran about 500 WUs on my offline server; the longest took about 90 hours and there were about a dozen others between 40 and 60 hours; but the vast majority were 10 hours or less. These were run on a pretty fast corei7, so there is potential for some users to notice run times of 100+ hours. Just to be safe, I will increase the grace period to give users more time to return results.
I am also looking into a better way of dividing up the search space, so that the run times are more uniform in the future.

I'm noticing work units taking the opposite of a very long, with some finishing in one second, and apparently validating. Is this expected behavior for some WUs?

I'm noticing work units taking the opposite of a very long, with some finishing in one second, and apparently validating. Is this expected behavior for some WUs?

Yes, this is expected. I am working on making the run times more uniform. Originally, I divided up the search space into pieces of equal size; but as one can see this is not working very well.