Newer milestone thread

[tha]

Originally Posted by Qubit

Is there a list of those 106 numbers somewhere?

The list of exponents is in the first column here, the other columns contain data of the first time LL test of these exponents.

These exponents you can look up in the active assignments where they are listed in between some more assignments that have become superfluent but remain because of glitches or specific interest of some user.

In the past weeks a number of contributors have banded together and looked for ways to speed up progress without harassing other users too much.

[Uncwilly]

Originally Posted by Qubit

Is there a list of those 106 numbers somewhere?

The current number is 77.
Countdown to proving M(30402457) is the 43rd Mersenne Prime: 77

[flashjh]

77 As of 19:29 EST:

28305107	30075313	30255503	30354253
28748821	30088549	30257027	30356087
28764713	30112079	30261811	30356647
28873849	30138149	30267511	30356653
28879159	30170867	30267751	30358501
28961683	30184789	30286357	30358921
29719853	30187747	30286493	30364979
29806061	30195889	30293587	30374779
29838439	30201211	30299561	30375701
29862601	30202373	30299749	30375739
29871139	30202603	30299833	30376009
29878237	30204901	30300539	30376441
29898031	30207883	30302123	30392501
29898079	30232183	30302539	30392707
29921741	30241259	30302761	30392777
29928341	30243641	30333203	30392833
29931263	30244273	30334763	30398911
29937487	30250669	30339389	30399143
29939321	30254773	30341747	30402401
29940049

[LaurV]

74 right now. Two of the 30M2 range are mine, they will be finished in 4-5 days. [edit: Legally assigned some time ago! No poaching please! :P]

[Brian-E]

Originally Posted by LaurV

74 right now. Two of the 30M2 range are mine, they will be finished in 4-5 days. [edit: Legally assigned some time ago! No poaching please! :P]

Right!
And a lot of the others will be being worked on by machines which take weeks or a few months and whose human operators don't post here. Let's watch and wait.

[tha]

I started the tread in the LMH subsection a month ago to clean the assignments list from some apparently 'not being worked on DC assignments'. At that time I hoped we might be able to do some 40 to 80 exponents from a list of 117 within two months time. We are trying to find a balance between the risk of poaching and the risk of an exponent left behind.

As it says on the assignment rules page: 'The server may reassign exponents where the assignment is more than one year old.' I don't know if the server ever did do that. However you can find a guideline here. Don't embark on a single assignment that takes over a year to complete. (or do that well out of the range of server assignments or in the 332M range.) Maybe it is time to update the minimum system requirements for types of assignment as well, although that is tricky since PC's have different rolling averages.

It would be 'nice' if the server could transfer an assignment to a group of volunteers to complete a DC. That would then involve transfering the intermediate save file. Like when at least 25% is done and the outstanding time needed to complete is significant more than most other assignments in that range, a transfer would be instigated and proportional credit would be given in the case of a succesful DC. Of course the risk of a bad residue would remain, hence only volunteers to take such assignments.
Another option would be that the client can receive a message from the server in an early stage, like before 10% has been completed to abandon the assignment and switch to another type of work such as additional factoring.

In the mean time I think it would be good to 'protect' assignments in the LMH subsection by mentioning their shepherding in the relevant thread.

[kracker]

65 now
I have a few exponents as well.

[petrw1]

My last in 2 hours....and I am one willing to volunteer for cleanup duty...but I prefer to take the cautious road to declaring assignments abandoned.

Oh and unlike some of you I can't do a DC in under a day but I can do several in less than a week.

[petrw1]

I'm interested in knowing when 10Million digits are all DC'd.

[tha]

Originally Posted by petrw1

I'm interested in knowing when 10Million digits are all DC'd.

Well, that wil be about when M44 will be confirmed in it's rank. The preferred DC exponents treshold will reach M44 around the first of June. So, all exponents will be assigned to a trusted client 60 days later, the first of August. With some luck and an end of year drive we might be able to accomplish that by Dec, 31.

[chalsall]

Originally Posted by tha

So, all exponents will be assigned to a trusted client 60 days later, the first of August.

Not exactly correct...

Those candidates which have been assigned, but then immediately abandoned, will be thusly assigned. Those which are "sat on" by very slow machines will linger; potentially for years (until, of course, they're "poached").

I would like to publicly support tha's comment about the "one-year limit" idea. This has been talked about (at length) here, and actually partially implemented on Primenet. Both the warning on the manual assignment page, plus via a message sent back to the client if the reported estimated completion date is more than a year away.

However, George / Scott never actually implemented the removal process. The client will drop an assignment (even if has had work done on it already) if it receives the appropriate message from the server, and then get a new assignment which is actually useful.

In addition to this, at least for DCing, I would like to suggest that the "preferred" status of assignments be those below the next "Milestone" (read: proofing the rank of an MP), OR 3000 candidates above the lowest DC candidate; whichever is greatest.

I know that some consider "poaching" to be abhorrent. And the "100.1 km/h is better than 100." argument.

But at some point a decision has to be made as to just when is a contributor not really contributing to the goals of this community. Is it really reasonable to allow someone to sit on a milestone blocking candidate for literally years, completing only a few tenths of a percent of work on a candidate per month (or, none at all)?

As an example, I plan to "poach" 45456781. Assigned to one of "curtisc's" machines for 1033 days; clearly stuck on P-1 Stage Two. I can complete it on one-half of one of my machines in about four days.

Does anyone consider this immoral? (And, if it turns out to be Prime, I'd be happy to let him add it to his collection.)