Does this mean the result is ignored? it'd be a shame if tasks that just miss the deadline don't go into the 'useful results' pile assuming they pass the validation test.

---

Does this mean the result is ignored? it'd be a shame if tasks that just miss the deadline don't go into the 'useful results' pile assuming they pass the validation test.

They are classed as not needed since the task has already been done.
You also get no credit for it :-(


So I would guess that they are just removed and never used.

---

Team mauisun.org

seems like a waste - the credit thing is fair enough (although if they're still useful then maybe they should be credited), but they're still likely to be useful if they're only a day or so late...

---

to bad they are just tossed. they should use the result if it not very late (under 24 hours or something) or RAH scheduler should know when the result is due and if no processing has started on the WU that is due within X hours, then send a abort command and have that unit removed from the list.

its silly to crunch a WU to completion and then have it tossed and no credit given if it misses the deadline by whatever minimal period of time. waste of cpu time and energy.

i saw an earlier post somewhere asking why not have a grace period of X hours after the deadline to report results. what if you successfully crunched the WU but then your ISP or modem gives out on you and you can not report in time? then what? that's just a what if.

---

to bad they are just tossed. they should use the result if it not very late (under 24 hours or something) or RAH scheduler should know when the result is due and if no processing has started on the WU that is due within X hours, then send a abort command and have that unit removed from the list.

its silly to crunch a WU to completion and then have it tossed and no credit given if it misses the deadline by whatever minimal period of time. waste of cpu time and energy.

i saw an earlier post somewhere asking why not have a grace period of X hours after the deadline to report results. what if you successfully crunched the WU but then your ISP or modem gives out on you and you can not report in time? then what? that's just a what if.

Well a deadline is err a deadline, if you had a grace period then your're in effect extending the deadline.
I don't know if Rosetta needs the short deadline it has, but it's up to them.
They don't actually need the result so why use it... One thing I still don't understand is this
a) random number is set by server
but
b) you can do more than one model, surely it cannot use the same random number each time.
c) are the results the same or not?

for c) if not the same because b) does actually generate different sets of random number (say I use 24hr instead of 3hrs) then why bother resending. Just count it as lost if not sent in after some fortnight or month after the due date (your grace period I guess here, but for credit and tracking only)

---

Team mauisun.org

It's actually a random number "seed". Each seed generates a stream of random numbers which are used to start each of your models. There is a tiny tiny chance that my seed and yours eventually generate an identical number and we crunch the same exact model, but this is very rare. By pulling a new WU and a new seed, you are back to the randomness of your models.

If two machines were given identical seeds, they would crunch all of the same models, in the same order... and if one crunches faster or longer then the other, then it would do more models along that stream of numbers. But the first models would be identical to another machine. This is how Rosetta is better of not having redunancy.

---

Rosetta Moderator: Mod.Sense

It's actually a random number "seed". Each seed generates a stream of random numbers which are used to start each of your models. There is a tiny tiny chance that my seed and yours eventually generate an identical number and we crunch the same exact model, but this is very rare. By pulling a new WU and a new seed, you are back to the randomness of your models.

If two machines were given identical seeds, they would crunch all of the same models, in the same order... and if one crunches faster or longer then the other, then it would do more models along that stream of numbers. But the first models would be identical to another machine. This is how Rosetta is better of not having redunancy.

I half mentioned that in my post '... does actually generate different sets of random number ...' It's a set of numbers.

But what I mean it they send out a second task, though that person could be doing the default 3hrs, where I could be doing the 24hr task.
So '3' not equal to '24' Hence, Rosetta should stop resending work out as there is no point to it.
The reason there is no point.
1) the total set of returned results is not necesseraly going to be the same.
2) they don't cover ever permutation anyway (i.e. every starting number), so anything missed is not going to harm them. It's all just statistical analysis.
3) this allows them to use my result if returned late, which may or may not include more results than the other persons (see 1) ;)

---

Team mauisun.org