I leave boinc do its thing with most projects but it doesnt really work with Docking@home when my machines go off at night or something I loose so much time that deadlines are impossible to meet…
if this keeps up ill have to scale down the Docking@home offensive and pick up one of my other projects…
Anyone else get this issue ?
Strange, all my deadlines are about 5-7 days and this does not cause me an issue even with my little Atom which is only on during the day.
Is your buffer set too high maybe, so downloading too many to complete by the deadline?
My old 1.7 lappy only crunches a WU every 8 hours, but the deadlines are long enough to cope with it. :shrug:
What’s the Atom’s benchmarks like? I got myself a 230 on a board sat here but it’s not working yet as I’m waiting for a cable to arrive (molex to 4 pin ATX adapter)
Appears it’s at 891 in top CPU’s on BOINC Stats. (Not bad that I have the No.1 CPU as well mind - Q6600@2.4Ghz :D)
Not sure what the benchmarks are, just installed Boinc on it to see what it would do for a tiny netbook it seems to cope fairly well.
I’ve had my Q6600 off over the weekend and I’ve just checked, every unit it has is over deadline
I had the same issue when I came back from holiday. I just aborted all tasks and then allowed it to download new ones all then had 5-7 days before the deadline expired
ye ive just noticed the default thing for Docking@home is to get 10days worth of units…
I stopped that… changed it to 1
strangely, all those that were over deadline and have been reported whilst the machine has been on this morning have all got credit.
They have been overloading me… been really having to push it…
Interesting DT… think I’ll see if they are relaxed on deadlines some.
The answer is:
If you miss deadline the unit is resent, (that is norm) but if you still send the Work unit back before the resent one does… you get the credit and they get screwed.
so send them if you are not real late!
Kind of unfair for the other… Oh well that is how Docking does it.
so that’s what happens then, you basically stand the risk of wasting a few hours CPU time for no benefit. When will projects learn that this behaviour of deadlines will push users away, at least use the results to verify each other and reward.
I suppose that’s what you get for a single result=single credit system and not having the units validated like Seti.
They really do need to visit this.
[QUOTE=DoubleTop;438059]I suppose that’s what you get for a single result=single credit system and not having the units validated like Seti.
They really do need to visit this.[/QUOTE]
Wow, I would say so. I didn’t realize they weren’t validating results. That is more than a bit dodgy.
A lot of projects don’t now.
I know for a fact Riesel Sieve didn’t, Enigma doesn’t, Folding doesn’t, CPDN doesn’t etc
Mine seem to be ok now…
ye the way docking do it isnt very nice… hmm
im going to keep up with the output. But I am picking up some of my other lower projects just wish WCG would work…
[QUOTE=drezha;438076]A lot of projects don’t now.
I know for a fact Riesel Sieve didn’t, Enigma doesn’t, Folding doesn’t, CPDN doesn’t etc[/QUOTE]
It really brings into question the validity of the science. SETI had to learn their lesson because of rampant cheating.
Well not just cheating either. How about a mildly unstable overclock, or a dodgy memory module? Either could skew the calculations. That’s why redundancy is there, I know some projects don’t need to use it due to the nature of the calculations performed, I’m unsure where Docking is on this question.
Those too. Cheating was what really blew up in SETI’s face though.
The more I look… the more projects I find that do the same. This is not uncommon at all.
But agree validation should be done.
Writing them about it, not that it will do any good… but writing nevertheless.
Let you know if I get any response.
I believe Riesel Sieve got away with it because they offered two options, Sieving and Prime Hunting.
A mistake in sieving wasn’t a big deal. Sieving looked at numbers and if they could factor, they were removed from the prime hunt. It removed factors quite nicely/quickly. A missed factor would still be tested via the prime hunting.
However, I don’t know how they checked the primes… I guess with Riesel Sieve there was 100 numbers they were chasing and thus if they crunched all the numbers and found they were missing a few, then they could go over it all again (and it’s quicker checking what’s already been done)