Working on the 28,000,000-30,000,000 range

[flashjh]

Originally Posted by chalsall

I've recently been thinking of removing DC and LL assignments from availability from the Manual Assignment pages (on GPU72), and only allowing such assignments via the Proxy to ensure the candidate will be immediately "claimed" by the user, and said user's machine, so we all know who we're dealing with.

Thoughts?

I don't think that the overall system impact will be huge, for me. I can use a slaved Prime95 to get assignments. Now that GPU72 hands out assignments for DC/LL, I rarely use the manual assignments anyway (though I've been using it lately to test the newest CUDALucas). I guess you'll need to let us know how many people are using the manual assignments now.

However, if I use Prime95 to claim them and switch to CUDALucas, they're still not going to get updated in the system and others may think they're abandoned. That is if I let them go too long or until CUDALucas can communicate with PrimeNet on it's own.

With the technology of today, I think that it's time to set some reasonable guidelines and expire any exponent that is over 180 days old OR without any activity within 60 days. Can anyone say where the tail end of the DCs would be if these really old exponents had been expired and reassigned? Either way, there has to be some common-sense here. I can run all my old systems and 'contribute' to the effort as 'any' work done can't be replaced. I'm not saying it's a bad thing. But, how much 'energy and time' is spent running these old systems when newer ones can do the same assignment in a few days with much less energy usage? I know someone with the analytic ability can do the calculus on which systems can still contribute and which ones should be shut down. All I'm saying is that people who care will follow the new guidelines and finish assignments sooner.

[richs]

Originally Posted by chalsall

I've recently been thinking of removing DC and LL assignments from availability from the Manual Assignment pages (on GPU72), and only allowing such assignments via the Proxy to ensure the candidate will be immediately "claimed" by the user, and said user's machine, so we all know who we're dealing with.

Thoughts?

My computer at work has a spam filter that I cannot get to communicate with the Prime95 server no matter what I try. So I manually reserve DC assignments from GPU92, and they are never claimed since my computer cannot communicate with Prime95. But I am very careful not to over-reserve so they are not held for very long periods. So, Chris, please do not remove the ability to manually reserve DC assignments that are not immediately claimed. Thanks!

[Miszka]

Originally Posted by chalsall

I don't entirely disagree.

For the record, 28550909 was "captured" by GPU72, and then made available to our authorized workers. A certain someone then reserved it via the GPU72 manual assignment page, but didn't "claim it".

A bit commotion came into being.
I done 15% test at present LL-D for 28550909
Do I not know if to execute farther?

[LaurV]

Originally Posted by richs

...from GPU92...

Cool! When do we start?

[tha]

Originally Posted by Miszka

A bit commotion came into being.
I done 15% test at present LL-D for 28550909
Do I not know if to execute farther?

All I can say is that I continue doing the exponents I started working on, unless the one who made the reservations communicates progress made on his assignments. It still looks unlikely to me as only 117 slow cpu cores could explain these reservations.

The good effect of this commotion is that the whole process of reserving assignments through all kinds of software packages that have been developed by many people is reviewed for unintended effects.

[Miszka]

I decided to test one exponent yet - 28900997 ( P-1 and LL)

[chalsall]

Originally Posted by flashjh

I don't think that the overall system impact will be huge, for me. I can use a slaved Prime95 to get assignments. Now that GPU72 hands out assignments for DC/LL, I rarely use the manual assignments anyway (though I've been using it lately to test the newest CUDALucas). I guess you'll need to let us know how many people are using the manual assignments now.

Well, clearly there's at least one (I copy you Rich).

Perhaps what I'll do is increase the privilege level for those two forms, with a notice to use the Proxy or ask for permission for the manual access.

Originally Posted by flashjh

However, if I use Prime95 to claim them and switch to CUDALucas, they're still not going to get updated in the system and others may think they're abandoned. That is if I let them go too long or until CUDALucas can communicate with PrimeNet on it's own.

I think an important part of tha's proposal is he's only going to process candidates which have clearly truly been abandoned, by someone who didn't bother to claim the assignments. I'm sure he'll take into consideration the "weight" of someone's production history -- thus "claiming" the assignment is important in this.

Originally Posted by flashjh

I can run all my old systems and 'contribute' to the effort as 'any' work done can't be replaced. I'm not saying it's a bad thing. But, how much 'energy and time' is spent running these old systems when newer ones can do the same assignment in a few days with much less energy usage?

I wholeheartedly agree!

I have access to several older machines upon which I could produce a small amount of work. But I've decided not to because of the power usage. We still mostly burn oil here in Barbados...

Originally Posted by flashjh

I know someone with the analytic ability can do the calculus on which systems can still contribute and which ones should be shut down. All I'm saying is that people who care will follow the new guidelines and finish assignments sooner.

James???

[richs]

Originally Posted by LaurV

Cool! When do we start?

Lol, a Freudian slip!

[petrw1]

Is anyone officially tracking the ownership here...other than checking all the posts above. If required I could probably complete about 5 a week.

[Brian-E]

Not really wanting to argue with anyone, and this will likely be my only post in this thread, but I'd like it on record that I feel quite threatened in my continuing tiny contribution to GIMPS by this sort of poaching. By the criteria indicated by a few in this thread, my slow but reliable DC work could easily be pinched from me before it reports.

[ET_]

Originally Posted by Brian-E

Not really wanting to argue with anyone, and this will likely be my only post in this thread, but I'd like it on record that I feel quite threatened in my continuing tiny contribution to GIMPS by this sort of poaching. By the criteria indicated by a few in this thread, my slow but reliable DC work could easily be pinched from me before it reports.