Newer milestone thread - Page 52

LaurV · 2012-08-03, 13:17

Hey guys, before some start throwing things in the head of the others, just take a small pause and relax. Nobody is advocating "poaching". Poaching is bad. But there is (and it will always be) a difference in how some of us define poaching. I was personally arguing a lot around, about the subject. Poaching is if I do some activity which will render the other guy's (assignee's) activity useless, making him to waste his time, resources, money. For LL, we anyhow need two tests. There is no harm done, and generally I don't see any harm done in "poaching" first-time-LL-tests. If I would have the resources and if I would find those "special cases" I would do the "poaching" myself, without much fuss. If the other guy finishes his assignment, both of us get credit, and one of the tests will count as double-check. No harm done! (in fact PrimeNet will give to both of us LL credit, not DC credit in this case).

The only "bad" things which can occur is if the exponent happen to produce a mersenne prime. In this case, the "poacher" has to be aware that the glory and the money will go to the original assignee. I am aware of this when I do my "poaching", and I willfully forfeit any rights I would have to that claim, and in fact, if it will happen to step on such a lucky exponent, then the original assignee can be happy too, because I helped him too....

Brian-E · 2012-08-03, 13:49

LaurV:

Fully relaxed as ever, and unlikely to start throwing anything at you or anyone else, I would like to point out that people are motivated to take part by many different considerations. It is therefore not a very good idea to take assigned work from people on the basis that you don't see any harm from your own perspective.

I can assure you that I, for one, would be quite upset if the first-time LL test which I had spent many months doing was relegated to a double-check because someone took the job off me. It may seem irrational to you, and perhaps it is irrational, but I enjoy the very occasional experience of almost certainly being the first person ever to run a primality test on a particular Mersenne number.

Dubslow · 2012-08-03, 15:39

Quote:

Originally Posted by ATH

I already started 39286963 but Signals beat me to it, I think I'll finish it as a doublecheck.

I seriously disapprove of this. Poaching a double check is quite a bit worse than poaching an LL; you'd be rendering the original assignee's work useless, as LaurV points out.

Quote:

Originally Posted by Brian-E

LaurV:

Fully relaxed as ever, and unlikely to start throwing anything at you or anyone else, I would like to point out that people are motivated to take part by many different considerations. It is therefore not a very good idea to take assigned work from people on the basis that you don't see any harm from your own perspective.

I can assure you that I, for one, would be quite upset if the first-time LL test which I had spent many months doing was relegated to a double-check because someone took the job off me. It may seem irrational to you, and perhaps it is irrational, but I enjoy the very occasional experience of almost certainly being the first person ever to run a primality test on a particular Mersenne number.

Indeed, we've been over this before, as I'm sure everyone remembers. There is something to be said for the mystery of the first time LL test. Even (especially) if we don't all agree on a subject, the most conservative option must be the default -- no poaching. I submit(ted) that poaching in a large likelihood of a test never being finished that is way behind the wavefront may have its uses in very limited circumstances.

The 39M expo in question which I already brought up must now be considered a DC assignment, and is thus years ahead of the wave; as I said, I would highly disapprove of further poached results being turned in.

imwithid · 2012-08-04, 06:35

Quote:

Originally Posted by LaurV

Poaching is if I do some activity which will render the other guy's (assignee's) activity useless, making him to waste his time, resources, money.

I have a task that is quite old and have tried to expedite its completion. It should be done in about a month bringing it to a total time of just under 635 days. Had it been recycled or poached, it would have been a waste of resources (based on my calculations, 50% above nominal energy consumption, meaning 35 kWh - could that be right? (1%@7h, hence 700 hours*50 watt hours)).

Perhaps a better way to persuade poachers against the act would be to allocate proportional (or proportionally staggered) time based on the size or range of M to signal poachers that there are term limits to a given task.

Regardless, I stand strongly against poaching and term limits should be increased (except to hoarders).

chalsall · 2012-08-05, 16:16

So, last night Primenet assigned Spidy / GPU72 40902149 to Trial Factor -- the lowest LL candidate. Spidy e-mailed me to let me know.

I then PMed flashjh, James, Dubslow and LaurV to concurrently do the appropriate TFing, P-1'ing, LLing and DCing, respectively.

Sometime in the last hour our "friend" Signals submitted a LL result....

LaurV · 2012-08-05, 16:45

Seems its report is from P95, so I will continue my DC. CL is ~~3M3~~ 3M5 iterations through, with an ETA of 38 hours. Doing it in two gtx580 in parallel and everything matches up to now. I wish I could get a different residue and prove that sonofagun poacher wrong

NBtarheel_33 · 2012-08-05, 19:16

August 5, 2012. All exponents below 41 million have been tested at least once.

Brian-E · 2012-08-05, 20:55

Quote:

Originally Posted by NBtarheel_33

August 5, 2012. All exponents below 41 million have been tested at least once.

No, no! Read the title of chalsall's latest post in this thread!

Off you go, Signals, get testing M42643801. You can have your eggs when the egg-timer on your machine starts beeping. (I prefer mine scrambled.)

aketilander · 2012-08-06, 05:16

Quote:

Originally Posted by Brian-E

No, no! Read the title of chalsall's latest post in this thread!

Off you go, Signals, get testing M42643801. You can have your eggs when the egg-timer on your machine starts beeping. (I prefer mine scrambled.)

All exponents below 41,046,353 have been tested at least once.

Brian-E · 2012-08-06, 08:14

Quote:

Originally Posted by aketilander

All exponents below 41,046,353 have been tested at least once.

Awww, don't confuse her when she's just getting the water to the right temperature for her operations!
Ignore him, Signals. That one isn't healthy, free range.

chalsall · 2012-08-06, 15:36

Quote:

Originally Posted by Brian-E

No, no! Read the title of chalsall's latest post in this thread!

...where I made a cut-and-paste mistake....

2012-08-03, 13:17	#562
LaurV Romulan Interpreter Jun 2011 Thailand 25B5₁₆ Posts	Hey guys, before some start throwing things in the head of the others, just take a small pause and relax. Nobody is advocating "poaching". Poaching is bad. But there is (and it will always be) a difference in how some of us define poaching. I was personally arguing a lot around, about the subject. Poaching is if I do some activity which will render the other guy's (assignee's) activity useless, making him to waste his time, resources, money. For LL, we anyhow need two tests. There is no harm done, and generally I don't see any harm done in "poaching" first-time-LL-tests. If I would have the resources and if I would find those "special cases" I would do the "poaching" myself, without much fuss. If the other guy finishes his assignment, both of us get credit, and one of the tests will count as double-check. No harm done! (in fact PrimeNet will give to both of us LL credit, not DC credit in this case). The only "bad" things which can occur is if the exponent happen to produce a mersenne prime. In this case, the "poacher" has to be aware that the glory and the money will go to the original assignee. I am aware of this when I do my "poaching", and I willfully forfeit any rights I would have to that claim, and in fact, if it will happen to step on such a lucky exponent, then the original assignee can be happy too, because I helped him too.... Last fiddled with by LaurV on 2012-08-03 at 13:19

2012-08-05, 16:16	#566
chalsall If I May "Chris Halsall" Sep 2002 Barbados 9,767 Posts	Lowest untested candidate now 42643801... So, last night Primenet assigned Spidy / GPU72 40902149 to Trial Factor -- the lowest LL candidate. Spidy e-mailed me to let me know. I then PMed flashjh, James, Dubslow and LaurV to concurrently do the appropriate TFing, P-1'ing, LLing and DCing, respectively. Sometime in the last hour our "friend" Signals submitted a LL result.... Last fiddled with by chalsall on 2012-08-05 at 16:17 Reason: s/to do/to concurrently do/

2012-08-05, 16:45	#567
LaurV Romulan Interpreter Jun 2011 Thailand 7²·197 Posts	Seems its report is from P95, so I will continue my DC. CL is ~~3M3~~ 3M5 iterations through, with an ETA of 38 hours. Doing it in two gtx580 in parallel and everything matches up to now. I wish I could get a different residue and prove that sonofagun poacher wrong Last fiddled with by LaurV on 2012-08-05 at 16:48 Reason: 480->580 (typo), 3m3>3m5 (time passed :D)

2012-08-05, 19:16	#568
NBtarheel_33 "Nathan" Jul 2008 Maryland, USA 5×223 Posts	Moving right along... August 5, 2012. All exponents below 41 million have been tested at least once.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Newer X64 build needed	Googulator	Msieve	73	2020-08-30 07:47
Performance of cuda-ecm on newer hardware?	fivemack	GMP-ECM	14	2015-02-12 20:10
Cause this don't belong in the milestone thread	bcp19	Data	30	2012-09-08 15:09
Newer msieves are slow on Core i7	mklasson	Msieve	9	2009-02-18 12:58
Use of large memory pages possible with newer linux kernels	Dresdenboy	Software	3	2003-12-08 14:47

2012-08-03, 13:49	#563
Brian-E "Brian" Jul 2007 The Netherlands 7·467 Posts	LaurV: Fully relaxed as ever, and unlikely to start throwing anything at you or anyone else, I would like to point out that people are motivated to take part by many different considerations. It is therefore not a very good idea to take assigned work from people on the basis that you don't see any harm from your own perspective. I can assure you that I, for one, would be quite upset if the first-time LL test which I had spent many months doing was relegated to a double-check because someone took the job off me. It may seem irrational to you, and perhaps it is irrational, but I enjoy the very occasional experience of almost certainly being the first person ever to run a primality test on a particular Mersenne number.