I also do what petrw1 mentions in post no. 1065. I carefully manage my DC assignments, and I've already lost some work in this latest move to organized poaching. So once again I respectfully ask everyone not to touch my assignments. Thank you.

[QUOTE=TheMawn;364929]Believe me, I would have gladly done it myself, but I don't know that any of the software I use can filter that. Unless excel has some comma-spacing option for columns.[/QUOTE]I import [url]http://www.mersenne.org/primenet/[/url] as txt file with fixed width columns on occasion to produce this awesome graphic. (Sorting on column A gets rid of all those peskie lines in the middle.)

[QUOTE=Uncwilly;364943]I import [url]http://www.mersenne.org/primenet/[/url] as txt file with fixed width columns on occasion to produce this awesome graphic.[/QUOTE]The graphic is positively beautiful. But it is also useless without knowing what the axes are!

[QUOTE=petrw1;364745]I'm interested in knowing when 10 Million digits are all DC'd.[/QUOTE]

This would not be a big project once M32582657 is double-checked, as there are fewer than 14000 larger exponents below the 10 Million digit limit that do not have known factors, and a good number of them will have been double-checked by time this milestone is reached. But maybe it would be fun to knock off the milestone of checking all exponents below M57885161 at least once first.

[QUOTE=retina;364945]The graphic is positively beautiful. But it is also useless without knowing what the axes are![/QUOTE]
Up down axis is %age of exponents factored out.
Left right is the exponents, the numbers are the millions digits. The purple arrow points to 332,000,000.

[QUOTE=Uncwilly;364951]the numbers are the millions digits[/QUOTE]
You wish! RDS will all be on your head for this remark :razz:
They are only million bits.

[QUOTE=Uncwilly;364951]Up down axis is %age of exponents factored out.
Left right is the exponents, the numbers are the millions digits. The purple arrow points to 332,000,000.[/QUOTE]

[QUOTE=LaurV;364955]You wish! RDS will all be on your head for this remark :razz:
They are only million bits.[/QUOTE]

xxx,000,000: would not the xxx represent the millions digits (like the tens or units or hundreds)?

Hehe. Bits is short for binary digits. So you are both correct (in a way). LaurV assumed decimal digits when he saw digits, and Uncwilly implied binary digits when he stated digits.

[QUOTE=retina;364958]Hehe. Bits is short for binary digits. So you are both correct (in a way). LaurV assumed decimal digits when he saw digits, and Uncwilly implied binary digits when he stated digits.[/QUOTE]

I am talking about those in the millions place. No binary in my thoughts. I am talking about the value of the exponents from 0,zzz,zzz to 999,zzz,zzz.:pancakebunny:

[QUOTE=Uncwilly;364959]I am talking about those in the millions place. No binary in my thoughts. I am talking about the value of the exponents from 0,zzz,zzz to 999,zzz,zzz.:pancakebunny:[/QUOTE]And the exponent describes the number of binary digits. Like I said above, you are both talking at crossed purposes.

[edit: my reply is for Uncwilly. The (uncalled) posts of retina were crossposts :razz:]
I understood exactly what you were talking about. Here is where "not being a native speaker" actually helped me. My post was a joke, as your post made me remember a [URL="http://www.mersenneforum.org/showthread.php?p=271281"]long argument[/URL] I had long ago with our resident expert.

[QUOTE=richs;364942]I also do what petrw1 mentions in post no. 1065. I carefully manage my DC assignments, and I've already lost some work in this latest move to organized poaching. So once again I respectfully ask everyone not to touch my assignments. Thank you.[/QUOTE]
And I, too, frequently still have an in-progress assignment which is older than 90 days. Two reasons:[INDENT]- Sometimes assignments actually take me longer than 90 days to carry out when working on them: my machine is only switched on for a fraction of the time

- I get my assignments from GPUto72 when they are sometimes already a few weeks old according to PrimeNet. For this reason I suggest also subtracting a fixed time, say 20 days, from the age of the assignment before calculating its expected finishing time.

[/INDENT]

[QUOTE=Brian-E;364965]And I, too, frequently still have an in-progress assignment which is older than 90 days.[/QUOTE]

I /think/ what is being discussed here is not to automatically re-assign candidates just because they're older than 90 days, but rather that after 90 days enough progress should have been done to be able to calculate with reasonable accuracy when the work on the candidate will be completed. This is instead of trusting the client's estimate, which is often very inaccurate (because the machine is not on as much as the configuration is set to -- by default 24 hours/day).

If no work has been done, or the completion date is past a year or so, then Primenet would reassign it to someone else's machine with a better history of productivity.

[QUOTE=Brian-E;364965]I get my assignments from GPUto72 when they are sometimes already a few weeks old according to PrimeNet. For this reason I suggest also subtracting a fixed time, say 20 days, from the age of the assignment before calculating its expected finishing time.[/QUOTE]

I just realized what you were saying here; excellent point.

I'll modify my Perl script to have it check if the assignment was made through GPU72, and if so use its assignment date rather than when the candidate was reserved from Primenet.

[QUOTE=richs;364942]I also do what petrw1 mentions in post no. 1065. I carefully manage my DC assignments, and I've already lost some work in this latest move to organized poaching. So once again I respectfully ask everyone not to touch my assignments. Thank you.[/QUOTE]

This is exactly what we wanted to prevent from happening.

In the years that the v5 server has been in operation the tail end of the DC front has not moved noticeable. Now, in just a few weeks it moves by about 5.000.000. Neither situation indicates a healthy condition.

So, during a walk in the forest today I came up with [B]Y[/B]et [B]A[/B]nother [B]M[/B]etric to let the server recycle assignments in time. I hope this one is the most simple to implement, easy to understand and fairest to all.

The treshold value for preferred DC exponents is saved by the server on a day to day basis. All the exponents lower than or equal to the threshold value and higher than the treshold value of the previous day get a year from that day on to be completed. If a client reports to the server an 'estimated completed by date' of more than 6 months prior to the final day the server sends an informative warning message to the client about the remaining time available. If the client reports a date beyond a year than the server takes the assignment away from the client.

So a client can lose an assignment if:

a: 60 days go by without a report.
b: it is estimated to take more than a year to complete since the treshold value passed the exponent value.

For first time LL test we may have to look for a similar value of the 'one year since the treshold value passes the exponent' limit.

Maybe it is even possible to assign exponents below the preferred treshold value that have less than a year to complete only to clients with an average rolling value high enough to complete the assignment still in time.

I'm finding it extremely difficult to sympathize with people hoarding their preferential numbers for over ninety days. There [I]is[/I] an option to tell your program how many days of work to stockpile, i.e. 5, not 180.

I don't [STRIKE]think[/STRIKE] know why people [STRIKE]should be[/STRIKE] are getting batches of work for their CPU from [B]GPU[/B]72. The Prime95 client is just so good at managing the exact cause of some of the problems we have. Typical GPU work gets done so fast that over-aged assignments are a complete non-issue. Prime95 can contact the server even on a 2-week basis to give it a reasonably accurate estimate of how long something is going to take.

Instead, people do their assignments manually and give absolutely no progress reports until the assignment gets finished, which leaves the percent complete at 0 for however many months, and flags the assignment for infinity days to completion.

I just want to further stress:

Report progress once in a while!

Don't take more assignments than you can [STRIKE]complete[/STRIKE] get started on in ninety days!


Do these two things and these two things only and you will never get "legitimately" poached upon. The Prime95 client can do this very, very well with about thirty seconds of internet once every two months.

[QUOTE=TheMawn;364928]"Anonymous" needs restrictions on their ability to get and extend assignments. Perhaps they too could be restricted from taking bottom-of-the-pile assignments entirely and be not allowed to extend them past 90 days for DC, 180 for LL?[/QUOTE]

[QUOTE=petrw1;364938]Except that ANONYMOUS is many different people who choose not to be named.

Some are very well behaved in this regard ... some ???[/QUOTE]

The anonymous users who are well behaved do not need their DC assignments extended past ninety days. I'm not saying we should force users to declare themselves. I'm saying they should accept [B]some[/B] restrictions in [B]highly preferential assignments[/B] and [B]some[/B] restrictions in assignment extension past a very generous 90 or 180 days.

The problem is with everyone coming in and telling us how they're an exception to every single proposal we give, no progress is going to get done. On the other hand, people could accept that their cherry-picked DC is going to get done by someone else and move on with their lives.

So far:
[LIST][*]We can't kill off all 90-day-old assignments with zero progress because of the GPU72 delay (this I can accept is fine because Chris is working on a fix)[*]Even after the fix, we can't kill all 90-day-old assignments with zero progress because some people like hoarding six months worth (or potentially years worth and constantly extend)[*]We probably can't kill any assignment with more than 0% progress because they're being worked on (possibly with a microwave oven).[/LIST]
[QUOTE=petrw1;364937]Safer but probably much more difficult might be to look for NO progress and NO reduction in Days to complete over time.[/QUOTE]

I'm glad petrw1 is actually proposing an alternate criteria. However, his 90-day-no-progress assignments do actually fall into the criteria he suggested; his assignments have made no progress in ninety days. 92, 93, and 97, as a matter of fact.

[QUOTE=tha;364980]So a client can lose an assignment if:

a: 60 days go by without a report.
b: it is estimated to take more than a year to complete since the treshold value passed the exponent value.[/QUOTE]

I agree with you, but would extend the criteria thusly:

1. Any candidate held for 120 days without reported [I]progress[/I] is automatically recycled.

2. Any candidate held for more than 365 days with less than 50% completion is recycled.

3. Any candidate held for more than 730 days is recycled regardless of the reported completion.

4. Manual assignments are valid for the same 60 days without updates as are regular assignments, and can't be extended except by putting them on a Prime95 / mprime instance and reporting them properly.

4.1. It continues to blow my mind that any random agent (human or bot) can reserve assignments from Primenet with no prior authentication, and they're then held up for 180 days.

5. Any work currently assigned which is no longer needed (e.g. TFing or P-1'ing) is released back into the pool. Several people are holding up candidates for LL assignment while they "claim" to being doing work.

In all honesty, and to be quite frank, it almost seems like George is tacitly approving this current "poaching" activity.

I personally, and I suspect many others in the community, would welcome his thoughts on this discussion.

[QUOTE=TheMawn;364985]The problem is with everyone coming in and telling us how they're an exception to every single proposal we give, no progress is going to get done.[/QUOTE]

I don't think this is actually the case. Some very valid points are being raised here.

"richs" is correct that he very carefully manages his assignments. And he did lose some work because of a mistake I made -- "petrw1" agreed to let me take back some "milestone" assignments to process on my own machines, but I made a mistake in my SQL which resulted in three of richs' assignments also being transferred. Even though his workers had not yet reported progress to Primenet, they had just started the work.

(For the record, when he brought this to my attention I immediately stopped work on them, and offered to simply hold them for him to complete.)

Surely we, as a community, can come to some reasonable compromise here. Ideally Primenet and GPU72 will only assign the low LL and DC candidates to those workers who are reliable -- I've already removed two "horders" from being able to reserve low candidates from GPU72.

I'm hopeful George will enter this discussion, and give his perspective and thoughts.

[QUOTE=TheMawn;364985]

I'm glad petrw1 is actually proposing an alternate criteria. However, his 90-day-no-progress assignments do actually fall into the criteria he suggested; his assignments have made no progress in ninety days. 92, 93, and 97, as a matter of fact.[/QUOTE]

Only 1 of the 2.

1. Reporting progress: No because they are in the queue
2. Reducing completion days: Yes because they are advancing in the queue.

Only if both are No can you reasonably conclude they are essentially abandoned.

Then for those that are progressing the question becomes: Is the total days to completion acceptable? I propose that is a forum decision. And my suggestion would be to allow 90 days of work (because and as long as that is the limit in the entry field) plus up to 60 days to complete. I further agree that if no report and no reduction in days for 60 days it is also fair game.

Finally why might I grab 90 days or more of low-end work?
I was doing this before GPU72 but for similar reasons:
- I could ensure it was going to get focus and get done; maybe in months but not years and not held and abandoned.
- And when there were more than I could handle I offered to share with others of good standing.
For example about 200 DC in the low 20s came available some years back. I took 80. I didn't get many takers for the rest and they went back to the general public...some of which got stuck again. In hindsight I wish I would have grabbed them all. Might have taken me more than a year but that would have been quicker than what happened.

[QUOTE=chalsall;364987]
1. Any candidate held for 120 days without reported [I]progress[/I] is automatically recycled.[/QUOTE]

In early January I left a computer in N.C. running until mid-May unattended and without an internet connection.

[QUOTE=Prime95;364995]In early January I left a computer in N.C. running until mid-May unattended and without an internet connection.[/QUOTE]

How much progress did said computer achieve (and I'm assuming you mean last year, since we're now in late January 2014)?

Could you please, sir, speak a little more deeply about the current debate?

[QUOTE=chalsall;364987]In all honesty, and to be quite frank, it almost seems like George is tacitly approving this current "poaching" activity.[/QUOTE]

I had no objections to tha's very conservative approach. I think your proposal and TheMawn's are too aggressive.

Some rambling thoughts of my own:

As best we can, we need to honor our prior commitment to not recycle for a year exponents that are being actively reported.

I don't think we can place much value on the %complete metric. I often queue up months of work. These exponents report no progress until they finally reach the top of the queue and then they finish quickly.

We can however use the %complete metric once it goes non-zero. If an exponent starts progressing at 1% a week for an extended period of time, it is likely to take a 100 weeks. The downside is the server does not keep a history of this data.

We need to come up with different strategies for DC and LL work at or near the trailing edge. LL work in the 100M area. Other DC and LL work. TF and P-1 work. ECM work. IIRC, this where my proposal 5 years ago failed.

Manual assignments do need to default to a longer time to expire. When I download exponents for a GPU, I get several months worth (and extend the expiration dates). I don't want to be bothered with doing this manually every 1, 2, or even 3 months.


How to proceed? Do you want me to start separate threads for each strategy with an initial proposal?

[QUOTE=chalsall;364996]How much progress did said computer achieve?[/QUOTE]

I left the computer in January 2014, so we don't know. I did several tests of having it reboot after a manufactured power failure. It passed all the tests, so I have hopes it will be able to run for 4+ months straight.

[QUOTE=Prime95;364998]How to proceed? Do you want me to start separate threads for each strategy with an initial proposal?[/QUOTE]

Yes. Please.

The current situation with Primenet is demonstrably sub-optimal.

[QUOTE=Prime95;364999]I left the computer in January 2014, so we don't know. I did several tests of having it reboot after a manufactured power failure. It passed all the tests, so I have hopes it will be able to run for 4+ months straight.[/QUOTE]
Similarly, I've left a few tests without supervision for what is now getting to be near the end of [URL="http://mersenneforum.org/showthread.php?p=335439#post335439"]the "free year" (on the EC2 cloud)[/URL]. All of them finished without an error, but were quite slow (and [I]so what[/I]). They did the job (even if only 3 or 4 DC tests per year).

I have retired them now. [SIZE=1]I've later used a free instance of so-ugly-it's-beautiful 32-bit linux "tiny"s which I used as a time machine to travel back to 2007 and build modified NewPGen binaries (one needs an ancient gcc-3 and accompanying static libs). [/SIZE]

[QUOTE=Batalov;365002]Similarly, I've left a few tests without supervision for what is now getting to be near the end of [URL="http://mersenneforum.org/showthread.php?p=335439#post335439"]the "free year" (on the EC2 cloud)[/URL]. All of them finished without an error, but were quite slow (and [I]so what[/I]). They did the job (even if only 3 or 4 DC tests per year).[/QUOTE]

From my perspective, we're trying to figure out and agree as to what is reasonable and fair.

Slow (or often offline) computers are more than welcome. But they should be assigned work which is appropriate for their ability. Ideally the curves would cross perfectly.

It's a difficult problem. But, then, we here often deal with difficult problems....

Agreed. Clearly defined rules and open discussion are fine things to have.

(As opposed to simply poaching "because the bureaucracy will be too slow to change the rules". Here, at GIMPS, the list of bureaucrats is very short ;-) and they actually demonstrate reasonable nimbleness.)

You wouldn't want to turn away any live users, no matter how slow they are. What seems to be the problem are zombies/unmanned droid type of things. You need a Turing like test to recognize their behaviors from the patterns in the existing database of accesses (which is hard because it is not very detailed) - and at the same time not hurt live users no matter how closely they might look like droids. When you set the new rules, you leave a grandfathered period for old rules, too - you are right in theory that "Slow computers ... should be assigned work which is appropriate for their ability." But they enter the pool all the time and you don't know if they are slow or not right away; and then fast forward to "now", they have the assignments and they are holding "the milestones" and if we will yank some assignments away, we will arguably have some immediate acceleration and then (depending on the fairness of the execution) piss more (or less) live users off and they will leave. Try to integrate this.

Conservative changes make sense.

[URL="http://mersenne.org/report_milestones/"]Countdown to proving M(30402457) is the 43rd Mersenne Prime: 15[/URL]

:max:

It appears that I have been absent for a great deal of interesting conversation. :coffee:

[QUOTE]some grow impatient with those who waste our time holding up milestones,.[/QUOTE]I've seen this and similar complaints many times over the years.

[U]Not one single person has EVER been able to explain to me what "waste" occurs in these case![/U]

In fact there is no "waste" involved, but what there _is_ is that some people want to justify taking poaching-type (or assignment-cancelling) action by projecting their own internal feelings of impatience onto the GIMPS project. By doing this, they can pretend that GIMPS is somehow "impatient" with the progress of milestones, but that's only their own self-deception, not reality.

Such people need to learn self-control, not propose rules that unjustifiably demean the contributions of "slow" systems.

[B]If anyone disagrees, then please publicly explain just how there is any [U]"waste"[/U] when milestones are not achieved as fast as impatient people want them to be completed.[/B] Do NOT confuse project "waste" with internal feelings of impatience because of poor self-control.

[QUOTE=cheesehead;365195][U]Not one single person has EVER been able to explain to me what "waste" occurs in these case![/U][/QUOTE]Please provide data showing how Capitan Entropy helps GIMPS. Specifically CE has taken out blocks of manual assignments and held on to them for [B]4 years[/B], with no work shown on the exponents. That seems to be a waste. Taking out a single year's worth of work on an isolated machine is one thing. But 4 years?

That seems like willful obstruction.

[QUOTE=Uncwilly;365198]Please provide data showing how Capitan Entropy helps GIMPS.[/quote]That's just a diversionary strawman request having nothing to do with what I posted.

[quote]Specifically CE has taken out blocks of manual assignments and held on to them for [B]4 years[/B], with no work shown on the exponents.[/QUOTE]

Lack of progress is not "waste"! It can be dealt with through assignment rules, not poaching.

[quote]That seems to be a waste.[/quote]... only because the endless highlighting of "milestones" misleads folks into thinking that they are a realistic measure of progress.

What was actually, objectively _wasted_ (other than Captain Entropy's waste of his own time)?

[quote]Taking out a single year's worth of work on an isolated machine is one thing. But 4 years?[/quote]... and where was the threshold of "waste" there, between 1 and 4 years?

[quote]That seems like willful obstruction.[/quote]Whose GIMPS work got wasted (besides Captain Entropy)? What meaningful measure of actual GIMPS project progress was obstructed? Why can't this be dealt with via assignment rules rather than by poaching?

[QUOTE=cheesehead;365200]Lack of progress is not "waste"! It can be dealt with through assignment rules, not poaching.[/QUOTE]
Would you be fine with me taking an exponent just below a known prime (via a manual assignment) and then reporting exactly 0.005% completion per week? That would hold up the proving a prime's position past the lifespan of everyone here today. Since it was a manual assignment, one could seek the number out and get it without the normal restrictions.

Cheesehead, all we're trying to do here is trim the fat and speed up the process of finishing the lower end of assignments. There was a bit of talk about just poaching some of the stragglers but we found that not very many of them were really being held up by users planning on taking several months.

We're trying to clean things up without poaching. We're changing the assignment rules. We're leaving wiggle room for the "slow" computers that take months for a single double-check so that we don't demean their work. We're in the process of determining what rules allow for swift progress while giving users a reasonable amount of time to do their work.

So far, it looks like far from the trailing edge, computers will still be given two years for an assignment as long as they update once in a while.

[QUOTE=TheMawn;365207]We're trying to clean things up without poaching. We're changing the assignment rules. We're leaving wiggle room for the "slow" computers that take months for a single double-check so that we don't demean their work. We're in the process of determining what rules allow for swift progress while giving users a reasonable amount of time to do their work.[/QUOTE]

Exactly!

If we get this right, there we never, ever be an incentive to "poach" because even computers which take a year to complete a single DC will be assigned work which is appropriate for their ability.

It's all about optimizing the curves....

[QUOTE=cheesehead;365200]


Lack of progress is not "waste"! It can be dealt with through assignment rules, not poaching.

[/QUOTE]

To my understanding recent conversation on this thread as transitioned from the 'morality' of poaching to a thoughtful discussion on changing assignment rules that will remove all temptation for such actions to be desired or a temptation in the future.

[QUOTE=cheesehead;365195]I've seen this and similar complaints many times over the years.

[B]If anyone disagrees, then please publicly explain just how there is any [U]"waste"[/U] when milestones are not achieved as fast as impatient people want them to be completed.[/B] Do NOT confuse project "waste" with internal feelings of impatience because of poor self-control.[/QUOTE]

The primary goal of GIMPS is to find new primes. But frankly, if there would be a Mersenne prime below the record prime as of today that we have missed, I would attribute more value to finding that prime, rather than a new record high prime. So, finding all primes below the record high and their assured rankings is a secondary goal.

Regularly DC's prove a LL residue to be incorrect. Being sure of each LL residue is the only method to be sure of a ranking.

Furthermore I care a little more about the findings during my lifetime than after it.

For statistical and other analysis a continuous range of results simplifies work.

[QUOTE=cheesehead;365195]Such people need to learn self-control, not propose rules that unjustifiably demean the contributions of "slow" systems.[/QUOTE]

Cheesehead... Will you agree that part of your position is based on you, personally, being targeted for poaching (and, thus, an emotion position)?

How many years ago did that take place?

Do you not think that it is reasonable that a slow system is given appropriate work from now on, and that the work will be recycled after a year or two if it isn't completed?

Must we wait for the rest of our lives for a slow machine to complete its work?

[QUOTE=Primeinator;365240]To my understanding recent conversation on this thread as transitioned from the 'morality' of poaching to a thoughtful discussion on changing assignment rules that will remove all temptation for such actions to be desired or a temptation in the future.[/QUOTE]
This is a good, pragmatic aim. I'd like to think that the new assignment and re-assignment rules currently being drawn up are geared towards precisely this. I suspect that this is indeed a primary aim.

[QUOTE=chalsall;365258]Cheesehead... Will you agree that part of your position is based on you, personally, being targeted for poaching (and, thus, an emotion position)?

How many years ago did that take place?[/QUOTE]
Not addressed to me, but sometimes I can't keep my nose out and this is one of those times. We [I]all[/I] have emotional reasons behind our positions. Sometimes they are altruistic (I suspect cheesehead's are indeed mainly altruistic) and sometimes they are purely selfish. But bringing them up in a discussion is rarely helpful and usually inflammatory. Its purpose is often simply to devalue someone's arguments on the basis of irrelevant personal situations, and whether that is your purpose here or not it is certainly likely to be perceived as such. Please let this one go.

[QUOTE]Do you not think that it is reasonable that a slow system is given appropriate work from now on, and that the work will be recycled after a year or two if it isn't completed?

Must we wait for the rest of our lives for a slow machine to complete its work?[/QUOTE]While agreeing with cheesehead's arguments, I also think that applying some new assignment standards is necessary in the interests of keeping as many contributors happy as possible. This needs to be done in such a way that users of slower systems remain happy as well. I think George Woltman's recent suggestions are striking the right balance.

:goodposting:
I think it summarizes most of the motivation and aim of this thread, and of people participating in this common effort.
+1!

[QUOTE=Brian-E;365272]This is a good, pragmatic aim. I'd like to think that the new assignment and re-assignment rules currently being drawn up are geared towards precisely this. I suspect that this is indeed a primary aim.


[/QUOTE]

The above (and what followed) was a great post. Reading it made me realize how poorly constructed my previous post was. One should not visit forums on two hours of sleep and make a posting. I sounded like a head injury victim.

Some very interesting points have been brought up in the other threads for determining DC and LL rules. I have not been a member of this forum for nearly as long as some members and also do not have the same level of appreciation for the intricacies of how the server integrates handling manual and automatically assigned tasks. Therefore, I do not feel like it is my privilege to throw out any suggestions myself. I do think something needs to be done to increase the lag of DC milestones. (Several) someones pointed out earlier that these milestones do add to a sense of community achievement. Achieving these milestones could become more predictable than finding the next prime- be it tonight or in three years. However, I also understand that users who possess slow machines need to be protected. Batalov and Cheesehead (and George) all made great points for my slow/remote machines and manual assignments are at risk with any aggressive type of change. Chalsall also makes great points. We should not have to wait five years for an assignment to be completed when the computer is over a year into the assignment and indicating virtually no progress. Someone else (I do not remember who) make a great point that many of the machines that are "delaying" milestones are likely users that only have their machines on a couple of hours/minutes per day or are users who have forgotten that they even have Prime95 installed. These users, especially the latter, likely care nothing for credit for assignments. It is the other group that needs to be protected- but how? How much protection is warranted? I'll leave that to wiser minds to solve!

[QUOTE]Countdown to proving M(30402457) is the 43rd Mersenne Prime: 14[/QUOTE]

Ever closer.

[QUOTE=Uncwilly;365206]Would you be fine with me taking an exponent just below a known prime (via a manual assignment) and then reporting exactly 0.005% completion per week? That would hold up the proving a prime's position past the lifespan of everyone here today.[/QUOTE]It wouldn't annoy me, but would probably get you poached by someone else _if_ our assignment rules didn't handle this case.

Annoyance is not "waste". You still can't demonstrate any "waste" (except of your own time) that damages the project in general, can you?

[quote]Since it was a manual assignment, one could seek the number out and get it without the normal restrictions.[/quote]If our assignment rules didn't handle that case, then we need better assignment rules, not poaching.

[QUOTE=TheMawn;365207]Cheesehead, all we're trying to do here is trim the fat and speed up the process of finishing the lower end of assignments.[/QUOTE]Fine with me.

Perhaps [I]you could have a word with the folks who bring up the same old rationales (e.g., "waste") for poaching[/I], so I wouldn't always have to be the one to do it.

[quote]We're trying to clean things up without poaching.[/quote]Of course we are. To echo Chris: "Exactly!"

[QUOTE=chalsall;365258]Cheesehead... Will you agree that part of your position is based on you, personally, being targeted for poaching (and, thus, an emotion position)?[/QUOTE]Whether it is has no bearing on the soundness of reasoning in the rest of my position, so why do you bring that up?

[quote]Do you not think that it is reasonable that a slow system is given appropriate work from now on, and that the work will be recycled after a year or two if it isn't completed?[/quote]Of course that's reasonable.

[quote]Must we wait for the rest of our lives for a slow machine to complete its work?[/quote]No.

Why would you pose such an absurd question to me? It was Uncwilly, not me, who proposed that scenario above. Do you need to read the posts more carefully?

:pancakebunny:

Here- a rabbit with a pancake to calm things down. See- cute, eh?

On a unrelated note:

[QUOTE]Countdown to proving M(30402457) is the 43rd Mersenne Prime: 13[/QUOTE]

If one of us just agrees with you, will you stop saying it?

A computer reporting hellishly slow progress isn't being directly wasteful itself, per se, no. Nobody is trying to make the opposite claim, that I know of. I certainly am not.

Now Cheesehead, what exactly is the point you're trying to make? If it doesn't involve something other than poaching, I'll be done wasting my time on this topic, because that is NOT what we're talking about here.

[QUOTE]Countdown to proving M(30402457) is the 43rd Mersenne Prime: 12[/QUOTE]

This is an impressive level of progress!

Can we get the next 2 done before year end?
(All LL to M48 and All DC to M44)

I think if George has a change to implement some form of assignment / reassignment rules for LL and DC it could be quite likely.

[QUOTE=petrw1;365489]Can we get the next 2 done before year end?
(All LL to M48 and All DC to M44)

I think if George has a change to implement some form of assignment / reassignment rules for LL and DC it could be quite likely.[/QUOTE]

Would be great to see but I find it unlikely any changes will be implemented immediately. Get back to me after my next glass of Belgian beer and I may have a different opinion...

[QUOTE=TheMawn;365367]A computer reporting hellishly slow progress isn't being directly wasteful itself, per se, no. Nobody is trying to make the opposite claim, that I know of.[/QUOTE]Yes, there are folks making the opposite claim. They quiet down after I refute it, but later they or others keep bringing it up again.

[quote]I certainly am not.[/quote]Of course not, and I never said you were.

[quote]Now Cheesehead, what exactly is the point you're trying to make?[/quote]I made it. See earlier posts.

GIMPS now boasts 120,000 users!

[QUOTE=Primeinator;365494]Would be great to see but I find it unlikely any changes will be implemented immediately. Get back to me after my next glass of Belgian beer and I may have a different opinion...[/QUOTE]

A change to the assignment rules and/or the same proactive approach to 30-33M that we have seen to 28-30M would easily do the trick for M44.

As for M48, most of these exponents will become "preferred" over the next few months, even under the current assignment rules, so in theory, this milestone has a good chance of falling in 2014 as well.

[QUOTE=Prime95;364998]How to proceed? Do you want me to start separate threads for each strategy with an initial proposal?[/QUOTE]

[URL="http://en.wikipedia.org/wiki/Analysis_paralysis"]Analysis paralysis[/URL]...

There has been a great deal of very valuable discussion on this issue. With many very good perspectives on how to move forward.

Perhaps it is now (or at least, soon) time to come to a communal decision, and then to actuate on same...?

Agreed. Do something now.

I'm in the middle of some 28.4 changes. After that, I'll start the web server work.

Analysis Paralysis... I can't believe there was a (group of) word(s) for that which I didn't know about.

I often lament at society's way of treating a solution with one small flaw as a flawed solution. For example, if all cars were driven by computers instead of people, the first crash would cause an uproar... despite having possibly reduced the number of accidents by 99.9%.

[QUOTE=tha;364686]These exponents you can look up in the [URL="http://mersenne.org/assignments/?exp_lo=26000000&exp_hi=30402457&execm=1&extf=1&B1=Get+Assignments"]active assignments[/URL] where they are listed in between some more assignments that have become superfluent but remain because of glitches or specific interest of some user.[/QUOTE]
This list now contains only the 12 expected exponents (instead of slightly less than 200 exponents yesterday).
Does anyone know what has changed?

I would say that George implemented phase one of the assignments reassignment unassignment process. The rest were superfluous.

[QUOTE=petrw1;365756]I would say that George implemented phase one of the assignments reassignment unassignment process. The rest were superfluous.[/QUOTE]

PrimeNet summary looks great now. It bugged me to see those counts in the LL-D column where there were none with status unproven.

[QUOTE=Chuck;365761]PrimeNet summary looks great now. It bugged me to see those counts in the LL-D column where there were none with status unproven.[/QUOTE]

Indeed.

And the good news is when the client checks in again it will be told "stop this work" (so it won't continue to waste its time and energy), and will be assigned new work.

Hopefully George also has in place the logic to assign the slow machines work which will become "trailing edge" appropriately ahead. (?)

[QUOTE=chalsall;365771]Hopefully George also has in place the logic to assign the slow machines work which will become "trailing edge" appropriately ahead. (?)[/QUOTE]

I don't know if we ever really clearly defined how far ahead to put them. It's a bit of a crapshoot. The fastest machine qualifying as a slow one will finish more than a year than the slowest machine still qualifying for any DC.

[QUOTE=TheMawn;365789]I don't know if we ever really clearly defined how far ahead to put them. It's a bit of a crapshoot. The fastest machine qualifying as a slow one will finish more than a year than the slowest machine still qualifying for any DC.[/QUOTE]

Defining the trailing edge 12 months out for DCing (assuming 41,000 candidates at year) is a very simple SQL statement.

[CODE]mysql> select * from Exponent where Status=4 order by Exponent limit 41000,10;
+----------+--------+-------+--------+--------+----------+----+----+-----+
| Exponent | Status | Facts | FactTo | P11 | P12 | LL | P1 | ECM |
+----------+--------+-------+--------+--------+----------+----+----+-----+
| 34338743 | 4 | 0 | 69 | 400000 | 10300000 | 0 | 1 | 0 |
| 34338809 | 4 | 0 | 69 | 385000 | 8181250 | 0 | 1 | 0 |
| 34338919 | 4 | 0 | 69 | 395000 | 8295000 | 0 | 1 | 0 |
| 34338949 | 4 | 0 | 69 | 395000 | 8295000 | 0 | 1 | 0 |
| 34338989 | 4 | 0 | 69 | 540000 | 540000 | 0 | 1 | 0 |
| 34339013 | 4 | 0 | 69 | 385000 | 7892500 | 0 | 1 | 0 |
| 34339037 | 4 | 0 | 69 | 375000 | 6468750 | 0 | 1 | 0 |
| 34339049 | 4 | 0 | 69 | 540000 | 540000 | 0 | 1 | 0 |
| 34339051 | 4 | 0 | 69 | 385000 | 7892500 | 0 | 1 | 0 |
| 34339087 | 4 | 0 | 69 | 405000 | 10833750 | 0 | 1 | 0 |
+----------+--------+-------+--------+--------+----------+----+----+-----+
[/CODE]

New minor milestone, all exponents below 29 million have been DC'ed. The lowest exponent is now 29,838,439 which is the only one below 30 million that has not been DC'ed yet. It will be about half way this month.

So the countdown from 10 down for the next major milestone starts now.

People are getting impatient
 

30402401 was registered to BigBrother (Minipax)

But the results page say:[code]Fredrik stella 30402401 C Feb 9 2014 9:07AM 0.0 31.6692 5EF4440792DFC1__[/code]If took only 0.0 days to run a 31.7 GHz-days test. That makes it an infinite-GHz CPU.

[QUOTE=retina;366487]30402401 was registered to BigBrother (Minipax)

But the results page say:[code]Fredrik stella 30402401 C Feb 9 2014 9:07AM 0.0 31.6692 5EF4440792DFC1__[/code]If took only 0.0 days to run a 31.7 GHz-days test. That makes it an infinite-GHz CPU.[/QUOTE]

Maybe it didn't report during the run? I don't know.. With a newer CPU, it can take only up to 24-48 hours or less.

If the user did not get the assignment through PrimeNet, there is no way of tracking its age. Hence the "0.0" age for such results. You see this with poached results (because you can't make PrimeNet assign you an already-assigned workunit) and with results where the user just selected a workunit and ran it without ever talking to PrimeNet.

I see that the exponent reserved by GIMPSChina is now showing 12.3% completion, while it wasn't even started a few days ago, so maybe it just took this long to get to the top of the queue.

Honestly, it seems to me that BigBrother should get the credit for this apparently poached result, especially with 77.8% reported completion. If the server could answer the next update with a message to the effect "result not needed, but full credit is awarded", maybe that would deter poachers to some extent.

I expect that most do not poach for the credit.

Honestly, if any result is poached the poach[B]er[/B] should be denied [B]all[/B] credit, which is to say their name should not appear anywhere and they should be given 0 GHz-Days of credit. Like the assignment never happened.

The one being poached upon, on the other hand, should be given full credit if they finish the assignment on time according to the rules. Their name should appear as the one who completed the test, and in the unlikely case that their exponent turned out to be prime, they should be credited as the discoverer as they would have been had poaching not occurred.


I agree that the credit thing is not a deterrent. If I ever did intentionally poach, it would be to move things forward rather than anything else. Luckily, that won't be an issue with the new rules in place.

I haven't followed all the new rules as of yet... if an exponent is already assigned to someone then will the new rules disallow reporting of the same work type while still assigned?

Well frankly I don't see why we should prevent reporting of the work.

Where poaching an assignment is wasting of resources, spending two months work to complete an already finished assignment is just further waste.

[QUOTE=flashjh;366535]if an exponent is already assigned to someone then will the new rules disallow reporting of the same work type while still assigned?[/QUOTE]

That will never happen. The primary goal of GIMPS is build an accurate mathematical database on Mersenne numbers. Disallowing data will not further that goal.

The secondary goal is to create a fair assignment/recycling/credit system to keep as many users as possible motivated to continue working on GIMPS.

[QUOTE=philmoore;366510]I see that the exponent reserved by GIMPSChina is now showing 12.3% completion, while it wasn't even started a few days ago, so maybe it just took this long to get to the top of the queue.[/QUOTE]

And now it's done - without poaching! Down to 7 left.

[QUOTE=TheMawn;366518]The one being poached upon, on the other hand, should be given full credit if they finish the assignment on time according to the rules.[/QUOTE]

Quite honestly, if it is a double-check which has already been verified by poaching, it seems a waste of time to require the one being poached upon to finish an unneeded triple-check. I would go ahead and just give them full credit if they have done any work at all on the exponent.

I agree, that credit is not often a reason for poaching. I just think that these kind of policies will send the message that poaching is frowned upon.

Over 900,000 computers have been assimilated into the Borg. :smile:

So now that 29838439 was poached is it time to suck out one more empty row from the [url=http://www.mersenne.org/report_classic/]classic report[/url]?

Looks like someone poached the exponent that I had poached (29838439) and beat me by two days. I will let it continue to roll so eventually four LL results may show up, but whatever.

I am also poaching 30399143 now, because the owner might need about a year more to complete it. I have no doubts that someone else will finish it long before I do, since I need about three weeks to do a single assignment.

I wonder if 30202603 was poached. I looked like the original owner was about to complete it, so I don't see why there would have been any incentive to do so.

[QUOTE=tha;366944]
I am also poaching 30399143 now, because the owner might need about a year more to complete it.
[B]I have no doubts that someone else will finish it long before I do, since I need about three weeks to do a single assignment.[/B]

[/QUOTE]

So why are you doing it? It´s a predicted waste of effort...

[QUOTE=lycorn;366946]So why are you doing it? It´s a predicted waste of effort...[/QUOTE]

Just to be sure at least one machine works on that assignment. Well anyway, with the new rules in place this will most likely be the last time we have to worry about manually overriding the assignment proces.

[QUOTE=tha;366955]Just to be sure at least one machine works on that assignment. Well anyway, with the new rules in place this will most likely be the last time we have to worry about manually overriding the assignment proces.[/QUOTE]
I don't get it. Why not let the new re-assignment rules do their work on this job as well?

[QUOTE=tha;366944]I wonder if 30202603 was poached. I looked like the original owner was about to complete it, so I don't see why there would have been any incentive to do so.[/QUOTE]

30202603 was done by the assigned owner, who has been posting regular updates and looks due to compete another one tomorrow. Two more should be done within a week-and-a-half, the last one by early March.

[QUOTE=tha;366944]I am also poaching 30399143 now, because the owner might need about a year more to complete it.[/QUOTE]

Let's see, 79% complete in a little over 4 months, I would expect this one to be done by mid-March at the latest. No need to poach.

By the way, I see that both recently poached exponents were by the same person (Fredrik). Maybe 29838439 was a legitimate concern, as the last updates had not shown any progress, but Big Brother certainly seems to have been a reliable tester.

I completely agree. None of this discussion has been about justifying poaching.

The rules will take care of the milestone bottlenecks soon enough. Assignments made before the switch WILL be grandfathered to give the promised year to finish the work.

Assignments which have been out for more than a year and classify as "preferred" will get recycled immediately as they have been given the promised time (and then some).

Assignments which have not been out for more than a year will be given that year to finish in case they are legitimately being tested. Give everything at most one year and all the bottlenecks will be cleared up.

:goodposting:

:smile:

Bravo! :cool:

I'll be excited to see the headers lowered another 10M and seeing some ECM done on 20M exponents.

[QUOTE=philmoore;366973]By the way, I see that both recently poached exponents were by the same person (Fredrik). Maybe 29838439 was a legitimate concern, as the last updates had not shown any progress, but Big Brother certainly seems to have been a reliable tester.[/QUOTE]30399143 was by Fredrik also. BigBrother got usurped again.

[QUOTE=retina;367014]30399143 was by Fredrik also.[/QUOTE]Add 30286357 to the list.

Grrrr... I will go and buy titans and do DC with them, like half a day for one expo, then invite you to poach those... :pirate: :yucky:

[QUOTE=LaurV;367023]... then invite you to poach those...[/QUOTE]I am not Fredrik.

[QUOTE=retina;367026]I am not Fredrik.[/QUOTE]
I know! Sorry for my English. That was general "you", i.e. the poachers. To make it clear, I also poach sometime, but not exponents which are progressing reasonable. And I see I also mismatched the pirate icon, the one with the scimitar was intended, and not the one with jumping hat.

The best way to deal with those guys is to have a hardware faster then them, and make them to lose time and resources. The bad thing is that you don't know (again, it is the "general you", not the particular one) when they start to poach one expo.

A nice surprise we could do to them, we can make a pool of "ready done" exponents, for example, I "legally" reserve and do my DC, report the residue, but if the residue matches, then it is not displayed immediately, but with some random delay (days). Then, if a poacher reports a result, my residue will be displayed instead of his. This would highly discourage them to poach progressing assignments.

[QUOTE=TheMawn;367011]I'll be excited to see the headers lowered another 10M and seeing some ECM done on 20M exponents.[/QUOTE]

There´s still plenty of ECM to do on exponents even lower than 1M, let alone 20M! We´d better concentrate our efforts on the really small exponents and make ECM progress at that level. ECM pays for small exponents.

[QUOTE=LaurV;367031]Sorry for my English. That was general "you", i.e. the poachers.[/QUOTE]Okay, got it.

If you don't mind me saying: I think that using "them" instead of "you" would have been better in that instance. Although my English is also not perfect either so perhaps there is an even better way?

[QUOTE=retina;367014]30399143 was by Fredrik also. [/QUOTE]

So I took it out of my list, I only wasted a day on it. The original owner had stopped working on it about 9 days ago. Might be because of a short holiday or so. I wonder when Fredrik started on it.

I switched to P-1 on the low ranges now, well below the current range. For one of my 4 cores that is. The other three will do server assigned DC's.

I think it would be best now if we let the new rules clean up the garbage that is left over from previous times. That will allow us to study all the effects of the new rules in detail.