Expiration time

[NormanRKN]

hello,

how long are the expiration times of assignments from mersenne.org and gpu72.com really . ???

i believe in mersenne.org manual testing page they are 180 days and in gpu72.org they are forever ? :D

i´m a little bit confused ;)


Norman

[chalsall]

Originally Posted by NormanRKN

hello,

how long are the expiration times of assignments from mersenne.org and gpu72.com really . ???

i believe in mersenne.org manual testing page they are 180 days and in gpu72.org they are forever ? :D

i´m a little bit confused ;)


Norman

Correct.

[NormanRKN]

oouuh, OK.

what happens if someone has a huge huge workload.
p.e. 1000 assignment with huge ghz-d and this person will die (trafficcrash, cancer etc. ) or some idiots which assign unnecessary a lot of work.
what will happens with the assignments ?
are they lost for ever or is there a "secure time lock" for such situations ?

Norman

[chalsall]

Originally Posted by NormanRKN

what will happens with the assignments ?
are they lost for ever or is there a "secure time lock" for such situations ?

Fear not!

There are many AIs in the loop advising the humans... And currently (although this may change soon) many humans advising the AIs....

[NormanRKN]

OK, thank you Chris for the explanation.

[chalsall]

Originally Posted by NormanRKN

OK, thank you Chris for the explanation.

Think about it!

[Prime95]

If Norman is going on a 180-day trip and wants to keep his GPUs busy, is there a way to get TF assignments that are ahead of where we expect the LL wavefront to be 180 days from now? If not, does it make sense for him to "hold up" assignments below the wavefront for 180 days? Or are the number of assignments "held up" in the noise category?

Should Norman let you know of his 180-day trip to make the "AI" a bit smarter?

[chalsall]

Originally Posted by Prime95

If Norman is going on a 180-day trip and wants to keep his GPUs busy, is there a way to get TF assignments that are ahead of where we expect the LL wavefront to be 180 days from now?

Sure... The hypothetical Norman could reference this report, and conclude that queuing up 180 days worth of work above about 64M would "make sense". Of course, the system by default wouldn't allow him to do so -- most people are constrained to holding at most 30 days worth of work -- so he'd have to contact us to have the automatic restriction waived.

Originally Posted by Prime95

If not, does it make sense for him to "hold up" assignments below the wavefront for 180 days? Or are the number of assignments "held up" in the noise category?

Definitely in the noise category. Those few assignments which are "old" from GPU72 are to trusted workers -- I have a report which shows when assignments are aging, and recycle those who are assigned to "unreliable" or new workers (after trying to contact the "owners"; and I generally take on the abandoned work myself to avoid any risk that someone else gets the dreaded "not needed").

As an aside, there's some noise in the Primenet TF and P-1 assignments between 50M and 60M as well. All these assignments are now useless -- the candidates might as well be released for LLing.

Originally Posted by Prime95

Should Norman let you know of his 180-day trip to make the "AI" a bit smarter?

It wouldn't hurt. But the AIs are pretty good -- there's no reason he can't leverage on MISFIT for example to automate his systems while he's away. At the end of the day "Norman" won't be able to hold up the show, even if he tried.