The latest S6Bucket resends need to be returned within 24 hours. It's hardly possible to finish those in time and with several wingmen already on the job who are days ahead I'm not going to waste time on this. No more S6Bucket for me, I hope this will be relaxed for the next run.
Copyright © 2024 Einstein@Home. All rights reserved.
Pushing too hard
)
No, it won't! Read leatest Bernds post here: http://einsteinathome.org/node/196604
Bernds latest post is about
)
Bernds latest post is about S6Bucket #3, not S6Bucket #2. I agree with the opinion of floyd. This pushing is a little bit excessive. To generate a further WU within 24 hours at a computing time of some 20 hours is to hurried. To wait the deadline of 3 days would be fair to be seen.
Of course, I do not know how much-needed the last tasks are, therefore, I compute further on S6Bucket #2.
By, Grubix.
RE: No, it won't! Read
)
Bernd mentioned two hours run times and two or three days deadlines. That's perfectly possible with a small cache. But the tasks I got yesterday came with one day deadlines and 20 to 25 hours run time. Chances are that after a day of work you just get told that your result is no longer needed and you're late anyway. I don't want to risk that.
RE: RE: No, it won't!
)
Run time or estimated time?
I think S6Bucket #2 needs to
)
I think S6Bucket #2 needs to be finished before the start of #3, that's why the short deadlines.
RE: Run time or estimated
)
That's extrapolated run times. Granted, an AMD FX may not be the best tool for the job. But it gets the job done. Usually.
RE: I think S6Bucket #2
)
Indeed, the results of S6Bucket FU2 are used to generate the workunits for S6Bucket FU3 . Because FU3 will be a rather short run, we want to generate all the FU3 work units in no more than two batches, and it makes things a lot simpler if all the FU2 results will be in by (say) a couple of days from now.
Sorry for any inconveniences. BOINC should actually be able to predict with some confidence whether your PC has a chance to make the reduced deadline and should decline work units that are too tight wrt. the deadline, but we know that BOINCC can't be perfect predicting this.
Cheers
HB