In the past week of Crunching SN2S WUs, I've had them run anywhere from 8.5 to nearly 14 hours on my quardcore laptop. It seems this project adds some balancing issues given the wide variety of crunching results. Some projects are pretty consistant from WU to WU, but it looks like SN2S is not one of them. I can see how doing Performance work in this environment can be a real challenge at times given all the input variables the tech team has to deal with.

Sep.21, one day average runtime: 6.30 hours
Sep.21, seven day sliding average runtime: 6:36 hours

Crunch On.

How about you mosey over to the Dashboard where all the forum frequenters go, which has had these numbers printed for years... no need to ask anyone!

edit: as for you status webpage... take a pick from this selection

Could you hoover around a wee bit and make me a sandwich and a coffee too Rob please ? ;-)

All the science apps that run on the non-deterministic search principle have outliers and if they hit on a slower device you get those numbers. The dashboard also prints the max and minimum average that has occured in the last 6 months, so SN2S has had averages ranging from 1.97 hours to 7.92 hours. My Linux quad is getting through them in 3:50 to 7:35 hours [WCGDAWS history sorted on run time]. Vina is very difficult to cut, so the techs do at least two things:

1) Filter out certain "suspect" tasks [not 100% waterproof]
2) Cut the number of jobs in a task so that a certain average is achieved.

Then it is the luck of the draw. The future, the long away future has as vision that jobs are cut to device capability. "We had a dream...;>)" and that it will be for years to come, so is the nature of the crunching beast. (A few experts surely will arrive soon to opine how WCG *should* do it, just like project X/Y/Z)

Whl., wholewheat with dairy butter or vegetable margarine, filtered with milk or Italian curto style coffee with a glass of mineral water on the side? Any other particulars... we do go out a long way for our best customers ;O)