recently, after some project down time, i noticed that my completion times on WUs had jumped up from 1:30-1:20s on my 5870s (running different CPUs and different GPU overclocks), to 2:07-2:10 across the board. updating the project, rebooting, and detaching and reattaching to the project in BOINC didnt set it back down to 1:25. did something recently get changed in the work load per WU that makes them take longer to complete? since the down time was for so long and it took my systems some time to get work again im still working my way back up the recent average credit scales (current #23 :P), but i'd like to see myself back up around 800k again and im concerned that my WU completion time will have an effect on that. any comments from the staff, or advice on how to fix this? the TeAm over at anandtech is usually not very in the know on these type of issues since most of them spend their time working for F@H, and im doing my best to keep our milkyway team alive and kicking :)

It could be longer wus. I have almost only recieved longer wus lately. They run about 40-50% longer on my cpu.
____________
Doesn't expecting the unexpected make the unexpected the expected?
If it makes sense, DON'T do it.

The 17_3s tasks currently being issued are much longer than the tasks that were issued in the past. The credit granted for 17_3s tasks has been increased according to the increase in task length. The time your 5870s @ 900 Mhz are taking to complete a 17_3s task is correct, it scales almost exactly to the time each core of my 5970 @ 810 MHz takes to complete a task.

Your 4 5870s would normally yield between 800-900K per day when there is no server downtime. Occasionally it may be more than that due to a random server glitch that assigns extra credit.

Recent Average Credit values take a long time to increase to match the daily credit you are currently being granted.