If we were looking for a silver lining for this problem - your statement is it. We had a few days where the results returned was above 1.25 million and (after the changes made last week) the system handled the load well. This gives us the confidence that we will be able to handle the GPU load once we move that into production. Additionally, we are going to start resizing new work that we create for GFAM, HPF2 and FAAH to drop their average runtime from their current 9 hours/workunit down to 6 hours. It will take a week or two before that work starts being distributed, but it is nice to be able to do that step.

Seems new longer units are trickling through, I got a 5.8 hour task just recently (completed for 108 wcg points-compared to the 2 from the shorter ones)

Altho this has messed up with the client, as not estimated completion times for c4cw tasks are estimated at 113 hours (when they where at 1:40 prior to completing this task)

I forced it to redo CPU benchmarks with no success. I'm guessing the client has it's own idea of how fast the computer computes flops regardless of what the benchmarks tell it.

This has caused many of my tasks to go into "high priority" mode. Nothing critical, but hopefully after completing a few more c4cw tasks it will realize that my computer is much more powerful then that =D

I just verified, the "to completion" number is inflated for all tasks, not just the c4cw ones.

I'm guessing that since I took 5 hours to complete the HCMD2 task when it estimated 5 minutes that it thinks my computer is alot slower then it actually is, and adjusted the "to completion" times accordingly.
(I'm basing this because the c4cw tasks are now down to 99 hours)

Hopefully this will clear itself within the next few days. Will keep you guys up to date.

To put it simply, with the way BOINC adjusts the Duration Correction Factor (DCF), it will take a given number of correctly estimated WUs to bring the DCF back to its normal value (maybe something like 80 WUs in your case).
Therefore the actual time it will take depends upon how many WUs your machine computes each day.
If you are impatient to recover, choose a "quick" project like HCC until the DCF is back to its usual value for this machine. If you prefer to stay with longer WUs it will only take longer but it's OK too if you can live with it.

If you want to avoid being hit again by such events better unselect HCMD2 because the duration of its WUs is so much variable and difficult to predict that the probability is rather high that there will be other such WUs.

dskagcommunity and Speedy51, I have the same questions, since I also would like to raise a silver badge to a golden one in this project. There's another thread that seems to answer this question: Help Cure Muscular Dystrophy Phase 2 - 18 days left of new work.