Cause this don't belong in the milestone thread

[bcp19]

Originally Posted by davieddy

Yes. Probability of being prime is inversely proportional to the exponent.
Time taken to LL is proportional to exponent².
So probability/year of finding a prime is proportional to 1/exponent³.

Humble toilers such as myself like these plums, and resent them being reserved for Chalsall's Chums.

Look after the waveFRONT speed, and let the tail look after itself.

If Primenet can't exploit GPUs properly yet, it should retire all the hoarded TF assignments >53M.

D

1) You just resent Chalsall period. With his new proxy setup, you could use his system to get lower exponents than you current get through Primenet(and thus being BETTER plums) and have no further complaints, but I have a feeling you will be like rcv and decide to continue bleating.

2)
1/43,000,000 = .0000000232558
1/47,000,000 = .0000000212766
difference = .0000000019792

Wow, that is SUCH a huge increase in the chance of finding a prime, I can so easily see WHY you would feel the need to want such plums. (in case your sarcasm detector is broken, yes, that was sarcasm)


I may not have devoted as much time as some, but I have devoted a lot of effort towards helping people like yourself to get lower than normal exponents that have been thoroughly checked out to make sure we are wasting as little time as possible doing LL's that are not needed. 68,486 TFs, 11,563 P-1s, 57 LLs, 58 DCs. Maybe I should take your attitude and only work for myself. I could ignore the TF work(or pawn it off on another GPUer like you do) and let people suffer through poorly worked over exponents while completing an average of 1.4 LL's per day. Then again, Xyzzy has already cut back operations because of the heat wave induced expenses, and after complaints lilke yours and rcv's, I have seriously considered doing the same.

Then there comes the other side of the coin... what if that perceived 'Plum' is indeed a prune(very likely), but, because I was so intent on getting that 'Plum' to work on, the assignment I would have gotten turns out to be the real plum and I have now lost out by chasing the perception instead of going with the flow?

The point still remains that these "plums" are nothing more than PERCEPTION. There is no proof that the next Mp will come from them, it's just the Human Nature to irrationally quantify things that cannot be put into set bounds and to resent people when they 'take these perceived things'. There is no mathematical way of predicting the area the next Mp will fall in, but that doesn't stop people from trying anyway. RDS has blasted many many people who have brought forth theories based on what has gone before.

It's sad that people see others in such negative light when they are working towards a common goal. If chalsall were the person rcv and you paint him out to be, wouldn't he just keep those plums to himself instead of passing them out to be worked on? After all, if he gives me a number to work on and it turns out to be a prime, will he get anything from it?

The ball's in your court, step up and be part of the team, or continue to nurse your 'perceived' greivances.

[Brian-E]

I think you make valid points, but I also think those who are uneasy with, or positively dislike, the claim which GPU72 makes on low exponents have made important points. I see these points being made by various people, not just davieddy and rcv. So I think there is more to it than just attitude or anti-chalsall feeling by one or two people.

Further to what you just wrote above:

Originally Posted by bcp19

The point still remains that these "plums" are nothing more than PERCEPTION.

I think the whole project is geared very much towards the perception of those who take part. Different participants are motivated in different ways. A justification of the type you give as to why the "plums" are not particularly important will not meet the emotional needs of every participant. Maybe some sort of compromise is needed, such as only allowing GPU72 to take exponents which are 1 mod 4 (suggested by George) or 1 mod 4n (Garo) to help keep as many people as possible happy.

[davieddy]

Originally Posted by bcp19

1) You just resent Chalsall period. With his new proxy setup, you could use his system to get lower exponents than you current get through Primenet(and thus being BETTER plums) and have no further complaints, but I have a feeling you will be like rcv and decide to continue bleating.

2)
1/43,000,000 = .0000000232558
1/47,000,000 = .0000000212766
difference = .0000000019792

Wow, that is SUCH a huge increase in the chance of finding a prime, I can so easily see WHY you would feel the need to want such plums. (in case your sarcasm detector is broken, yes, that was sarcasm)

No it was stupidity.
The "difference" is 10%, and allied to the time taken, it amounts to 30%.
I was simply answering your naive question.

[chalsall]

Personally, I really think that some of us take this whole thing way too seriously... At the end of the day, finding the next Mersenne Prime has no real direct benefit to anyone -- this is all just for fun!

I am involved with the GIMPS project because for over ten years now I have used the GIMPS client to monitor the machines I'm responsible for. I believe in giving back to any community which has assisted me, so when the need for a system to coordinate those with GPUs to do trial factoring became apparent, I volunteered. I already had a lot of the back-end code (spiders, the proxy which has just been made available, etc) written, and the servers available to host it.

Now, admittedly, what is now GPU72 is far beyond what I had ever imagined would be built. Extreme feature creep... But I take joy in providing the services, reports and graphs which others find useful, and I use the system as a test-bench for experimenting with ideas and technologies which I often use in other projects. It's great having a sophisticated user-base who are able to report problems in a useful manner, instead of the usual users' reports which are often along the lines of simply "it doesn't work"...

But, please... Let's never forget that this is simply a geeky hobby which should never be taken too seriously....

[chalsall]

Originally Posted by Brian-E

Maybe some sort of compromise is needed, such as only allowing GPU72 to take exponents which are 1 mod 4 (suggested by George) or 1 mod 4n (Garo) to help keep as many people as possible happy.

Which I am perfectly happy doing. The system only keeps a cache of low candidates as a function of how many are likely to be claimed by workers within a reasonable time. The rest are returned to Primenet for distribution. These numbers are adjusted from time to time, and will continue to be.

I have to say as well that I am very pleased that George has adjusted upwards the "Preferred" cut-off -- as I've said before it always annoyed me when low candidates were processed by the GPU72 system (TFing, sometimes P-1ing) only to be released back to Primenet and then grabbed by some Anonymous account that with 80% likelihood would never finish the assignment.

[bcp19]

Originally Posted by davieddy

No it was stupidity.
The "difference" is 10%, and allied to the time taken, it amounts to 30%.
I was simply answering your naive question.

God, you are like the spin doctors on TV... "There was a 50% increase in violent crime in the last year!!!!" was 2 last year, now 3... Might be a bigger deal if it was 1000 last year and is now 1500. Number do not lie, but how you present them does.

10% of infinitesimal is still infinitesimal.

Using your logic then, say the chances of winning a lottery were 1/47,000,000 and was then changed to 1/43,000,000, or 10% greater, would you then go put down your life savings because you have a 10% better chance to win?

[davieddy]

Originally Posted by bcp19

God, you are like the spin doctors on TV... "There was a 50% increase in violent crime in the last year!!!!" was 2 last year, now 3... Might be a bigger deal if it was 1000 last year and is now 1500. Number do not lie, but how you present them does.

10% of infinitesimal is still infinitesimal.

Using your logic then, say the chances of winning a lottery were 1/47,000,000 and was then changed to 1/43,000,000, or 10% greater, would you then go put down your life savings because you have a 10% better chance to win?

30%
If I were a core2 processor in a syndicate of 100,000, I might well be tempted to chip in the electricity required.
AS I DO.

And note I said "proportional to" NOT "equal to".
I'm beginning to suspect such "subtleties" may be over your tiny head.

[davieddy]

Originally Posted by chalsall

Personally, I really think that some of us take this whole thing way too seriously... At the end of the day, finding the next Mersenne Prime has no real direct benefit to anyone -- this is all just for fun!

Indeed.
Rather like a roller-coaster ride.
But the holidaymakers would hope that the designers and constuctors took their job seriously.

@Garo: Surely TF assignments between 40M and 70M should be reclaimed after a month at most. Just look at the Primenet summary to see how this is impacting the take up of LL tests.

With multiple cores, and AVX processors on stream, one could reasonably have hoped for a dramatic increase in the rate of LL completions.
This has not occurred.

David

[bcp19]

Originally Posted by rcv

There is much to be discussed openly, honestly, and constructively...

The main goal, I think, is to have GPU72 users do useful work, while treating all GIMPS users fairly and equitably. In my mind, advancing the tail is no more important now than it has been for the last 15 years. [Debate is welcome. George should have the final say. IMHO, one group, no matter how well-intentioned, should not unilaterally make a policy decision such as this.]

While my personal feelings don't quite rise to the level of "resentment", as someone else suggested, I do feel that permitting a bot to obtain the plum assignments is "unfair". In fairness, and in my opinion, those who have done the majority of the work to get GIMPS where it is today should be allowed the majority of the plum assignments, even if that means the tail won't catch up for years.

I seriously believe GPU72 should only claim a percentage of the newly released plum assignments in proportion to the amount of work GPU72 contributes. So, for example, if GPU72 clients put in 600 GHz Days of LL work, and Prime95 clients put in 14400 GHz Days of LL work, the GPU72 clients should be permitted to claim 4% of the next day's plum LL assignments. [A similar principle might apply to DC work, but I don't feel as strongly about that.] There is an unlimited amount of leading edge work (now standing at about the mid- to high-50M range). The more leading-edge work GPU72 clients perform, the more trailing edge (plum) work GPU72 clients will be permitted. I freely admit this proposal will limit the ability of the tail to "catch up" until and unless GPU72 recruits an much bigger army of volunteers. [A thousand more GPU72ers as dedicated as the first hundred would thrill everybody!]

Separately, we can discuss whether or not there is a problem with the Prime95 assignment rules.

In your second assignment, above (99 days old at 0%), I see that as relatively normal. With Prime95, you can ask for a queue of up to 90 days of work. Being 10% overdue on starting an assignment isn't a problem for me. And it's well within both the implicit time limits supported by the Prime95 program and the explicit time limits given on this Web site. It's also normal for the amount of CPU time donated to GIMPS to vary over the life of a computer. Perhaps the user was giving a high number of cycles to GIMPS over the winter, but has cut back during the hot summer. [Even some GPU72 users have stated they cut back when the temps warmed up.] So, 90 days of work, when assigned in early May might be 180 days of work in the dog days of August. Some leeway should be allowed.

The first assignment (small DC at 440 days) is a concern for me. If it has gone from 0 to 11% linearly over the last year, the assignment should be revoked. If it has gone from 0 to 11% over the last three weeks, I think the user might, in principle, be allowed to finish it, as it will be done within about 180 days. [However, if the Prime95 server can't track the *rate* at which a number is being worked, I wouldn't fault it for revoking this long-overdue assignment.]

Let me briefly mention the parms George recently proposed for new Prime95 / Manual assignments. I happen to have one assignment in the 331M range. It's on an alternate operating system that I don't run very often. It is over 6 months old and 4% done. Is it necessary to expire that assignment after 1 year? I think I was acting responsibly to choose an exponent that wouldn't hold up the works. I don't honestly know if I will ever finish it. If you expire my exponent, I won't finish it. But if you don't expire it, I have the opportunity to move the partially-completed assignment to a faster computer several years hence. (Many old-time Prime95 users have moved assignments from one computer to another when they upgraded.)

I also have no problem if the main GIMPS server tightens up the rules on allocating preferred assignments. (E.g., a queue length of 30 days or less, plus the existing rules.) But, if that isn't technologically feasible with the current main GIMPS server, I also don't have a problem if the preferred assignment rules remain the same. Prime95 users should be given ample notice to adjust their queue lengths, or whatever is necessary to allow them to continue to receive plum assignments.

Other ways in which GPU72 might help: LaurV proposed performing a triple-check on assignments that are suspect in some way. [Such as being performed by the same user.] I think that's a great idea! As George stated earlier, it may be decades before an independent third party can verify the work done by GIMPS. But that caused me to wonder if, in fact, maybe the time is right to TC (triple-check) all exponents below a very small threshold (2M???). While I don't want to take significant time away from the main GIMPS effort, getting an independent test of the small exponents with a totally different piece of software is inevitable. Does CUDALucas fit the bill? [But see below.] Are there other ideas?

Changes to CUDALucas to improve confidence: As it stands now, nothing prevents a user from faking CUDALucas results, provided they know the residue from a prior LL result (of their own) or from any prior LL/DC pair stored in the GIMPS server. Without getting into complex cryptographic protocols, a very simple change to CUDALucas would be to have it report a little bit more of the residue. Specifically, I would suggest reporting 7 random nibbles of the next-higher 8 bytes of the residue. [E.g., xx3xACxExB5xxx2xD45AC201BF9830E7.] I selected 7 of 16 nibbles, because it permits multiple CUDALucas-style reports without revealing the entire extended residue, while giving a high probability (>99.6%) that a pair of CUDALucas-style reports will have at least one common nibble. (These parameters may not be ideal.) If a user suddenly submits a thousand CUDALucas TC reports, his reports can be spot-checked to confirm their validity. If the main GIMPS server can't handle the extended residues, can GPU72 keep them??? [A technique permitting spot-checks was used on the DESCHALL project. The DESCHALL protocol required the client to return a value, that proved the calculation had been carried out, whenever a "half-match" was found. Statistically, a half-match was present in 1 in 2^32 keys tested. (The size of the keyspace was 2^56.) Any big user who was claiming to have tested large areas of the keyspace could be spot-checked, both by the statistics of how many half-matches they returned and whether or not the half-matches were valid. In practice, I verified *all* half-matches. No client was ever found to have cheated. And no big user reported a suspicious shortage of half-matches.]

Very interesting ideas, maybe one question to be asked: Is George working on including mfakt*/CUDAlucas into P95? I know he's been doing some GPU work lately.

Using your % rule would give Curtisc well over 20% of the 'plum' assignments, but that would also cause a major rewrite to the assignment program. With over 72,000 LL completed in the last year totalling over 7.4 million GHzD, davieddy with ~630 GHzD work would only have a .000084462 chance of getting one of them. Maybe we should have George change the way things are done here, put the lowest 1,000-10,000 exponents into a 'special group' and let people buy them to work on them. Nothing major, say just $1 per exponent. This would ensure 2 things, 1) a modest income for the site to make improvements, fund the found prime monies, etc. and 2) reduce the accusations currently going around. Most people will scream foul over this, but it is a legitimate idea to consider with everything else being thrown out there.

Your 331M exponent is a 'special case'. I would think any assignment over 100M would be exempt from current expiration rules (except maybe the 2 months with no reporting data one)

[Brian-E]

Originally Posted by bcp19

Maybe we should have George change the way things are done here, put the lowest 1,000-10,000 exponents into a 'special group' and let people buy them to work on them. Nothing major, say just $1 per exponent. This would ensure 2 things, 1) a modest income for the site to make improvements, fund the found prime monies, etc. and 2) reduce the accusations currently going around. Most people will scream foul over this, but it is a legitimate idea to consider with everything else being thrown out there.

Legitimate idea, certainly. Source of income, yes.
But I'm unclear why it would reduce "accusations", or even which accusations you are talking about (I've only seen differences of opinion and lively debate so far).
The idea might drastically reduce the number of low exponents which get abandoned by their assignees, which is positive, but it might also have the effect of concentrating these exponents in fewer hands, reinforcing the concept of an elite group which gets the plum assignments - which has been the most dominant criticism of GPU72. (It is rcv's main criticsm if I understand it correctly.)

[davieddy]

Originally Posted by bcp19

davieddy with ~630 GHzD work

You just resent me PERIOD.
What would you expect a 2Ghz Core2 to produce per year?
Mine is one of 10,000 other "core2s" trying to find a prime rather than pissing about. Our syndicate's chance of finding a prime in the next year is ~1/6.
I complete an LL test every 50 days or so.
What I consider to be the most disappointing aspect of GIMPS is that ~80% of LL tests are not completed despite being given the clear warning of expected completion date and the need for patience.

D