The switch to displaying elapsed time instead of CPU time gives a massive distortion for long duration CPU tasks.
48 days after switching from BOINC 6.2.19 to 6.6.31 a system I use for CPU intensive testing has a CPDN task with a 6 day discrepancy between the elapsed time reported in BOINC Manager (136.5 days) and the CPU time reported on the task's graphics and webpage (130.5 days). This actually understates the distortion because the task has only been scheduled for 75% of the time. In reality the elapsed time overstates the CPU time by 20% in this case. I can see this being a major support headache as BOINC 6.6.* becomes more widely adopted by CPDN users. Ian _______________________________________________ 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.
