I get about 3.65 days/24 hours from 4 CPUs (2 machines). When I work it out I'll be short by about 8 days. I'll increase my buffer towards the end to try to catch the shortfall that way, but I may be stuck needing a few cleanup units... It's part of the fun! I started crunching rice exclusively in October also, my plan has just missed though!

Damir1978, thanks for the positive message, but it is only a bit of fun

Hyperthreading by some efficiency analysis causes effective negative production [See Ingleside post testing with I7]. It's good for clocking loads of time. Mind you that might be on a science by science basis different as will it likely be varying when mixing different sciences. Someone posted few days ago to have switched off HT and saw dramatic improvement on run times for the DDDT2 A Type.

But I see where I screwed up. You were finding the number of WU's to get, whereas I was thinking in terms of how much time per day you were logging, and then extrapolating from that to how long it would take.

The specific reference Sekerob mentions above caused the DDDT2 A WU to complete 7 hours faster than the original projection. The WU had run for approximately 36 hours with another projected 20 hours to completion. I turned off HT to validate the WU faster and the WU was completed in 49 hours, and not the projected 56 hours. I do not believe that 6gigs of RAM with Windows 7 64bit Professional was the issue.

Based upon my experience for NRW, when you turn off HT the granted WU credit should improve sharply on a per WU basis. However, your daily aggregate granted WU credits decreases as fewer WU's are returned. This situation is true for NRW, excellent granted credits per WU, but lower overall granted credits over time.

Is March 25 (febr 23 +30 days) still about the end date of sending out the regular workunits?