> but without knowing all of the numbers from a single snapshot, it is > impossible to determine exactly why the project was in EDF. For example, > the CPU efficiency might have been 0.2 at that point
On that system, the CPU efficiency hovers around 0.97 - that's why I did not bother to mention that specific detail. My point here is that some details "bloat" the description without significantly affecting the "big picture". > The duration correction factor could > have been higher than reality, increasing the estimated wall times. What are you saying here ? My reporting of "four hour" tasks was based upon the 'time to completion' shown for those tasks by boincmgr. [And when they had finished, the sum of their runtimes did not appear to exceed the sum of the initial 'time to completion' estimates for those same tasks.] I had always thought the 'time to completion' shown by boincmgr __did__ include the effect of the duration correction factor. Are you saying here that when the duration correction factor is higher than reality, then (and only then) the estimated wall times could be higher than the 'time to completion' shown by boincmgr ? mikus _______________________________________________ boinc_dev mailing list [email protected] http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev To unsubscribe, visit the above URL and (near bottom of page) enter your email address.
