New WU length, super short?

In S4 we have recognized two clusters of WU length in compute time, one roughly 3.5x the other. On my Gallatin 3.2 GHz host running HT with SETI on the other side, the "longs" have been running about 75 minutes, and the shorts about 24 minutes, with variation about the typical of maybe 10 or 20% keeping the two clusters cleanly separate.

Now I've got 5 new results reporting at about 9 minutes, or less than half the previous shorts. All the IDs starts with z1_0297.5. Stderr does not contain an obvious indication of early exit, and at least some other users on the same WU also seem to have run fast also.

Here is an example super short result

I run tx36 and am calibrating, so the real_cpu_time in stderr is much closer than the posted CPU time.

I never even knew what the difference between the longs and shorts was, so have less than now idea what is different about these super shorts.