Why do I get so little task o FGRP4?
29 May 2015 14:19:22 UTC
Topic 198109
(moderation:
Hallo!
For prolonged time I observ, that I get less than 1% of FGRP4 tasks than S6BucketFuB.
From the server stautus page I learn, that the overall mean value is about 20%.
Why is this so?
Kind regards and happy crunching
Martin
Why do I get so little task o FGRP4?
)
You will have noticed from the thread in Technical News that there was a pause in work production while they prepared for, and switched to, a revised application?
Hallo Richard! This I took in
)
Hallo Richard!
This I took in mind :
from BM
But my effect last several weeks long now.
But nevertheless, I´ll observe this for another week.
Kind regards and happy crunching
Martin
When you have your
)
When you have your preferences set to accept both, I don't suppose the project makes any undertaking as to which you are more likely to get.
If you'd like more GRP work, you could just disable the other in your preferences. That happens to be what I do, partly because I don't like the way BOINC handles work mixes.
RE: When you have your
)
Yes when things like that happens it is a good idea to tell Boinc what to do since otherwise it can run around like a kid doing what it wants and telling you what the priorities are
RE: When you have your
)
Actually we do.
Currently we aim for GW:FGRP of 90:10 on average.
BM
BM
Hallo! For the last 2 1/2
)
Hallo!
For the last 2 1/2 month I carried out a test.
First I uncheked S6BucketFuB in my preferences and promptly got 100% of FGRP4. After about a week again I checked S6BucketFuB and the rate of tasks from FGRP4 dropped abruptly to about 10% at the beginning of the new serie as expected. But within the next about 4 weeks it dropped slowly - exponentially (?) - to less than 1% now.
At my computer tasks of FGRP4 get about 30% higher earnage [Cobbleston/hour] than S6BucketFuB. So partizipants, who are more interesting in high earnage, will prefer FGRP4. If 10% of participants have unchecked S6BucketFuB because of this less earnage and there is a limitation to 10% for FGRP4, other participants have very little chance to get a task.
As we have just 10% of remaing tasks of FGRP4 to become crunched, this is of little practical value, but it might be of interest for the adminstrators.
Kind regards and happy crunching
Martin
A few facts: - We're
)
A few facts:
- We're currently trying to send GW tasks to everyone who is able to receive these. Most of these tasks get rejected for disk space limitations, not because of people opted out of the S6BucketFU application. The GW:FGRP ratio we get is monitored per day, it varies quite a bit, but is pretty close to 90:10 on average.
- Only less than 5% of the active users did ever make an application selection in their preferences.
- Credit has been carefully manually tuned to be the same credit/h on average for all our (CPU) applications. If you observe a large benefit from running FGRP tasks over GW tasks, there's probably at least one other participant who sees it the other way 'round.
BM
BM
Hallo BM! Thankyou for this
)
Hallo BM!
Thankyou for this prompt and early reply.
These are interesting facts, indeed.
Rejection of tasks due to spacelimitations at me will not happen, as E@H uses just 15% of the space booked for BOINC and I´m crunching E@H only.
So, it´s a bit late for further tests with FGRP4, but I´ll have an eye on that effect with new applications comming soon and report here about my observations.
Kind regards and happy crunching.
Martin