OK, they seemed to have fixed the problem with these units being sized too big, but they have not fixed the credit issue with them. I had 5 of them complete in 8 to about 13 hours. The two shorter ones actually got credited with what I think is the proper amount of points (500+), but the three longer ones all exceeded the "maximum point level" and got 64.7 points. The maximum point level must be set too low for the longer running units. Once they exceed this level they are penalized for potential cheating. This is on a stock Xeon e5410 with Linux Mint 16. Once again I don't care that much about the points, but I do care about the accuracy of the system. Perhaps the techs could raise the maximum allowed as some of these units would require. I am pretty sure I am not the only affected by this.
Cheers

The work units that are currently being sent out were built before the sizing was fixed. Since the problem only affected some work units, many will be normal sized, but others will still be large. We're working on getting an estimate for when we'll get to ones built with sizing fix.

Seippel

As to the credits of these long WUs. My 245 hour WU claimed 64.8 which is the default anti cheating value. My wingman claimed 227.7 for almost 50 hours of work. The result is we both got 146.2 credits for the WU. So this anti cheating credit amount is affecting everyone. It suggest WCG create a script to look for these underscoring long WUs and fix the allocated credit for them.

Just sent this back after 7:45 hours at 6.000 percent, with code 203, user abort.

7.16 fahv FAHV_x3VQA_IN_LEDGFa_rig_0222584_0008_10 07:45:12 (07:13:36) 93.21 6.000 00:26:14 03d,04:14:23 9/18/2014 3:55:31 AM [0] 00:08:18 Running

Copy _10, i.e. the 11th. 7.75 / 0.06 = 129.66 hours, never going to make the deadline, if it would not crash with signal 11 before that. Copy _11 and a 'too late' conversion pre-programmed to happen.

For signal 11 on android, think now that it's something coming away from the gui manager doing it's communications and interfering with the tasks or client_state. The jargon something like race condition? Been running the past 48 hours without the gui loaded, using the disconnect option and none happened since. Only occasionally reconnected to have a quick peek, and then discovering one of these long x3vq's running headlessly into that 3.5 days deadline wall.

Still a few floating around..
FAHV_ x3VQD_ IN_ LEDGFa_ rig_ 0223664_ 0025_ 0-- 720 Pending Validation 9/15/14 10:50:01 9/19/14 00:28:37 65.87 227.5 / 0.0

Seesh 65 hours for 227.5 points. Seems like a waste of Hydro running these. I have 3 more of the same still running with 42 hours for 53%. Think i should abort these.