|
发表于 2007-10-12 19:16:50
|
显示全部楼层
感谢万能的google~~
From BOINC WIKI (see the BOINC home page)
The Result Duration Correction Factor was introduced in Version 4.70 of the BOINC Client Software.
The Result Duration Correction Factor is kept on a per Project basis. It starts at 1.0 and will quickly rise if the Project's Initial Duration Estimate is low. If the Initial Duration Estimate is too high the Result Duration Correction Factor will reduce at a more moderate rate.
The Result Duration Correction Factor will increase to match the observed duration if the Initial Duration Estimate is too low.
If the observed duration is less than 10% of the Initial Duration Estimate, the Result Duration Correction Factor will decrease by 1%.
If the observed duration is more than 10% of, but less than the Initial Duration Estimate, the Result Duration Correction Factor will decrease by 10% or enough to match the observed duration whichever is less of a change.
A Result Duration Correction Factor of 1.00 means that the computer is crunching at the rate predicted by the Benchmark. Since most computers work faster than this when working for the SETI@Home Project, and I'd consider a figure of 0.6 or so to be about normal when running the standard SETI@Home Science Application. If the computer is running an optimized one, then I'd expect 0.4 would be reasonable. On an efficient system with tweaked memory timings and a CPUCPU with 1 MByte, or more of L2 cache then figures below this are possible. A figure greater than 1 can probably be seen on older computers with small L2 cache especially if is badly configured or sat on a cheap motherboard with a poor quality support chips.
貌似是评估任务包实际完成时间和预期完成时间的关系,可能会影响服务器的调度策略吧,应该关系不大。 |
|