increase 'Maximum daily WU quota per CPU'
28 Dec 2005 21:14:36 UTC
Topic 190463
(moderation:
Since the release of the Albert WUs I have noticed several posts about systems going idle because they exceeded "Maximum daily WU quota per CPU". One of my own looks likely to do the same. I hope the quota can be increased soon to avoid wasting all those cpu cycles.
ps I am aware of the option of running other projects, I am simply not interested enough in any of them to spend the time managing multiple projects.
increase 'Maximum daily WU quota per CPU'
)
I second that. I wouldn't like to see cpu cycles wasted.
Spend what time, it is set
)
Spend what time, it is set and forget for me, am running 6 projects....
BOINC Wiki
Thank you
)
Thank you
I run several projects, BOINC
)
I run several projects, BOINC is meant to set and forget. Add a secondary project for backup. Einstein may not always be up and running. What happens if a main component goes down for a few days? If they get something like a DOS attack and virtually shuts them down, or (forgive what I am about to say) they get a virus and it locks up the project for days/weeks? What if a large influx of users come on and the data all gets analyzed that they have in their hands?
All apart of life. Your machines could be productive in other ways. You can make Einstein at 100 and the second project as 1, just to have as a backup in case of hiccups.
I see the comfort in a single project, but do not understand the logic. Allow yourself to grow a little.
If not, that is your choice. Either way, good luck on all future endevors.
Pooh Bear 27: Thank you for
)
Pooh Bear 27:
Thank you for your input. I don't like making half hearted commitments. I am sure that you do not see it my way, but for me that is what it would be.
ps The quota has been doubled and that was what my thank you was about.
RE: Pooh Bear 27: Thank you
)
I understand your position - but please think of the following points.
You could choose a backup, give it a resource share of 0.1% while leaving Einstein with the standard 100.
This means there is a one in a thousand share to the backup project. It will immediately download one cache of wu for the backup project, and will most likely leave it and then rush to finish it just before its deadline. It will then wait a long time before downloading another wu for that project - over a month.
There is nothing half hearted about giving a one part per thousand 'insurance premium' to prevent wasted cycles. The Einstein team ask you to do that. It means that something useful will happen even if they do something silly and knock out their project for a few days -- takes the pressure off them to know other useful work is being done.
What goes around comes around. Einstein also gets backup cpu cycles from other projects - so far Einstein has benefitted more than it has given, as it has never gone down itself and these Albert wu are the first time it has refused work to people.
Or, if that is still too part-hearted for you,
If you give the backup project a resource share of 0.01% you'll probably wait more than a year to get that second wu from the backup project, unless Einstein goes down for a while inthe meantime. Would you consider doing that?
If you give the backup porject a share of 0.001 you will most likely upgrade your machine before you ever see the second batch from the backup project.
Would you really rather risk wasting machine cycles than lose one cacheful of crunching for Einstein?
Of course, it is *your* box and *your* donation to the project, and *your* choice: but I do hope you will take a second look at your decision.
If you don't want to spend time choosing, maybe go for Rosetta as they have (currently) the friendliest project team of any. That's why I support them and me - I don't even like biochem!
Happy New year,
River~~
~~gravywavy
gravywavy: Thank you as well
)
gravywavy:
Thank you as well for your input. I don't expect everyone to agree with me and I can accept the possibility that no one shares my opion. I never had anything to do with BOINC before E@H and I doubt that I will have anything further to do with BOINC if/when E@H goes away. The exception to that would be another project somehow connected to GR.
Mark, I agree with you
)
Mark,
I agree with you completely and I share your opinion about E@H and BOINC. If there is no E@H I wouldn't use BOINC at all. So, E@H is the reason I start using BOINC.
I just hope there will be enough work for us crunchers in the field of Gravitational Wave detection, and in the future in the field of Gravitational Wave Astronomy.
Edo