I run BoincTasks to monitor and control my three systems. The task display gives prominence to the High Priority mode, so I noticed this morning that two of my systems had received 1.20 Gamma-ray pulsar binary search #1 on GPUs (FGRPopencl1K-nvidia) tasks from older workunits for which multiple previous tasks had timed out or otherwise failed so that the tasks I received carried _3 or _4 suffixes. I
These generally were from older groupings than the currently distributed 36L work, including 32L and some 35L. None of which would matter for High Priority, save that these were sent with deadlines five days from sending date, rather than the currently standard fourteen days.
While my queues are shorter than five days, I run dissimilar GPUs, so the estimated completion times rise and fall, and near the high end these units are deemed a deadline risk and run in HP mode.
Is this 5-day deadline stuff a routine practice I did not notice in the past, or something else? On closer review of a few of the affected WUs, it appears that "something else" is afoot, as I see a single successfully returned task, one timed-out task, one still in progress task sent with a 14-day deadline on July 13, and two new 5-day deadline tasks sent out today.
Copyright © 2024 Einstein@Home. All rights reserved.
There are some unfinished
)
There are some unfinished workunits that use an old downloadserver that we want to retire. Thus I have created additional tasks for those workunits with a shorter deadline in order to get those workunits finished as fast as possible so we can retire the server. It is a routine practice that is used from time to time to speed up things.
Thanks, Christian. So not
)
Thanks, Christian.
So not exactly the daily routine, but something similar to what I've seen before when you are sweeping up the last stragglers from an early calibration run of an application or such, but for a different reason.