Too fast for *you* :-( -- Need a quota of 20/day!
6 Jan 2006 20:18:12 UTC
Topic 190536
(moderation:
Excuse me guys, I will never again buy a fast CPU. But I did. And now this box has to sit around and wait for midnight to get new work.
Don't you like fast CPU's? Do you want to force people with fast CPU's to join a different project, causing a loss of the computing power we give to you?
Please adjust the quota a little bit. 20 would be fine for me, maybe 24 would satisfy almost everyone.
This here is the box I mean, and it will be out of work in exactly 18 minutes.
http://einsteinathome.org/host/504785/tasks
Same happened yesterday, same will happen every day until those shorties are gone.
Thanks!
Too fast for *you* :-( -- Need a quota of 20/day!
)
Forgot the first rule of BOINC?
"Attach to more than one project!"
Seriously, I am with you on this one.
Team Linux Users Everywhere
RE: Forgot the first rule
)
Did that already :-)
http://orbit.psi.edu/show_user.php?userid=246
Two are enough :-)
Fine!
Maybe I should go the boinc development forum and start a discussion about counting only downloaded/pending/trashed WU's for the quota, and not counting WU's which caused credits. For my special case this would solve the problem, sorry I am too lazy to do statistics for other boxes.
RE: Maybe I should go the
)
Good idea! Unfortunately, it is not nearly as easy as changing 16 to 20 (or 24).
What people NEED to remember,
)
What people NEED to remember, is this is a science project. They deem the terms, we do not. If they only want to feed us 1 a day, that's what they will do.
We just volunteer our time. We are not forced, paid, or any other way compensated for this, except for something called credits, which mean diddley squat for any of our purposes (except a few bragging rights).
They may have it toned down right now, because they are working on grant purposals, or something. Then when the grant comes due 6/12 months down the line they open up things a little more and show a growth. It's all in the name of the game.
I see no issues with running out of work, so your machine saves a little energy a day, you're electric bill will be smaller, etc.
I don't see why people make demands on a project that they just crunch for. We have nothing vested in this of any consequence. Allow the developers to do what they WANT/NEED to do for the project.
RE: I don't see why people
)
Okay, my electric bil will be cheeper :-)
However. For every project in the world you have to sides. The developers and the users. For a developer an elevator with a speed of 300 miles per h our would be optimal, because a maximum number of people could be lifted in the upper floors. But what is optimal for the developer is not optimal for the user.
And this is true for every project. You need a constructive, positive feedback from the end users. And I did not say 'Hey, the current quota is '. I said: (with other words) 'Count only the unsafe work units for the quota'.
However, happy crunching.
When I first read your post,
)
When I first read your post, I felt like it was coming across very strong and demanding. One does not have to use explicitives to be rude and arrogant.
Another thing to think about. If they are going to have a quota, then they must draw a line in the sand somewhere. No matter where they put it, someone is going to be on the outside. Maybe not today, but possibly tomorrow, or the next day someone will have a system that will fill the quota.
Pooh Bear offered a couple of possible explainations as to why the quota is set to its current setting. I'm sure, as a group, we could come up with many more legitimate possibilities. The only ones who know for sure why the quota is set at its current setting are those who make the decisions for Einstein.
Using your analogy, there are two sides to every story. You seem to be wanting the constraints set to benefit you, with no consideration for the ramifications the changes to those constraints will have on the dev team or any other cruncher.
Now, to your suggestion regarding the quota be limited only when bad w/u's are returned from the end user. This is already in use to some extent. If a user returns a bad result, then his quota will be decreased. This decrease has a lower limit of one result per day being sent to the user. The user can then increase his quota by returning good work units, upto the maximum quota set by the project. Why have a WU quota in the first place? There are many justifications for having a WU quota. One is to discourage WU hording. If a small number of users are allowed to horde a large number of WU's, the possibility of an errant computer propogating erroneous results through the system, thereby contaminating the science, is greatly increased.
I will leave it to the reader to consider other justifications for the enforcement of quotas.
To reiterate Pooh Bear's point. This project does not belong to us. It belongs to the Dev team, founders, and funders. Yes, we do provide a very important service and for that, I am sure the Project is very grateful. However, running this project is a privelge not a right. They allow us to run their software, they are not required to let us run their software.
Jim
**edit**
One other point I failed to mention. The change that you propose regarding the varible quota based upon errant returned WU's. Boinc determines a WU as valid based upon a quorum of results. Since these results do not get returned simulateously, there is a delay between the user returns the result and the quota is adjusted. I inferred from your statement that you expected it to be an instantaneous adjustment to the quota. Assuming that the quota is based solely upon erroneous returned results, the wayward computer could easily suck up volumes of data and return it corrupted before the validator gets a chance to compare them with other returned results. By that time, the damage could be tremendous. At least by having a hard cap, this possibility is throttled to the limit of the daily quota.
Jim
RE: When I first read your
)
Sorry, english is not my native language.
Agree. Currently the upper bounds of the qouta is static. As you say, he lower bound is somehow dynamic.
Yep! But their decisions are restricted by the software which is available. BOINC is good, but as everything it can be improved and it will be improved.
And I see no reason why a quota shall stop anyone from returning valid (or possible valid -- meaning pending -- results). I have no problem with the quota with any other of my boxes, it is just one which seems to be too fast. And in my team I have a couple of other people which get hurt by the quota and the missing logic ignoring valid results too.
[Do you want to force people
)
[Do you want to force people with fast CPU's to join a different project, causing a loss of the computing power we give to you?