| Subject: Re: Upload of work units |
| From: Bill Jillians <"\"Bill Jillians\""@blueyonder.SPAM_NO_THANKS.co.SPAM_OFF.uk> |
| Date: 07/12/2005, 18:11 |
Quietly we read
<rpzlf.24268$uR.7904@newsfe7-gui.ntli.net>, unable to
contain our credulity we realized that Martin 53N 1W
<ml_news@ddnospamddml1dd.co.uk.dd> said this:
Bill Jillians wrote:
[...]
The question I have ... which isn't answered here ... is will my WU
which finished on Sunday and which I have been trying to upload for
three days now still get me credit if I have to wait till after 6pm
GMT tonight when it expires?
The answer to that is "maybe".
If your result gets back before the quorum of results is assessed, then
yes. Now, you're in a race to get your result back before any other
third result for your WU. In this case, you'll still get credit even
way past the deadline.
Look in:
http://boinc-doc.net/boinc-wiki/index.php?title=Main_Page
for further detail.
And this is a good example of how too large a WU cache is bad. Sounds
like you have a 10 day cache. That leaves you only 4 days to process
the WU and get the result back... It also delays granting credit to
others to a minimum of your 10 day delay...
Thanks Martin.
I'm not so sure a big cache is bad. I have a few machines and for all
but one of them I uploaded and got new WUs on Friday or Saturday. For
just one very slow machine that takes 30-40 hours to process a WU the WU
was finished on Sunday when the outage started. I had units in reserve
so that was not a problem. Just the unit that finished on Sunday
expires just about now and there is no way I can submit it.
One beauty of Boinc is that you can attach to a second (or more)
project and so keep your idle time busy throughout outages on any one
project. The accumulated "Long Term Debt" will give that interrupted
project preference when back online so as to 'catch up' on the resource
shares. All very neat.
I'll stick with s@h until there in a download crisis ... at which point
I shall probably turn off my machines and save some energy. The only
useful thing they are doing is s@h and with the dearth of signals found
I'm not so sure how useful even that is. Still it has to be done.
My preferences are s@h and einstein, with a good chunk of CPDN. Your
interests may vary ;-)
Happy crunchin',
Martin
--
Bill Jillians