Officially poaching very old exponents

Prime95 · 2003-11-10, 04:01

I propose that currently assigned doublechecks below 9 million that have been outstanding for 2 years and at their current rate of progress will not finish within 6 months, be officially poached and reassigned.

Comments? Should the 2 year value be reduced to 1.5 years? 1 year? Other suggested adjustments?

garo · 2003-11-10, 04:52

I agree George. In fact I think the threshold should be reduced to 1 year. There are several exponents that have been assigned for between 400 and 700 days.

GP2 · 2003-11-10, 06:08

Quote:

Originally posted by Prime95
I propose that currently assigned doublechecks below 9 million that have been outstanding for 2 years and at their current rate of progress will not finish within 6 months, be officially poached and reassigned.

As currently formulated (2 years), there are hardly any exponents that qualify.

Here are the longest-running double-checks under 10 million (only one of which is greater than 9 million). Running time is the 5th column:

[Edit: click on an exponent to show a graph of its progress

7137989,D ,64,5737723,730.2,-54.3,5.7,11-Sep-03 11:04,09-Nov-01 14:14,S09625,Cel12
7137769,D ,64,4055939,730.2,-45.5,14.5,15-Sep-03 08:58,09-Nov-01 14:09,S09625,Cel14
7251577,D*,64,4035727,702.8,-35.2,4.8,10-Sep-03 15:25,06-Dec-01 23:22,jamesm,D2
7304699,D*,63,3404512,695.3,-56.6,3.4,04-Sep-03 05:46,14-Dec-01 13:44,older,MYLITV
7336607,D ,64,4055040,689.9,29.2,29.2,11-Sep-03 23:34,19-Dec-01 21:57,S60523,CCDA9E7C7
7501867,D ,64,1118779,650.8,-41.3,16.8,30-Aug-03 13:57,28-Jan-02 00:01,S64180,C9085839F
7516247,D*,64,3675197,646.9,-38.0,10.0,15-Sep-03 18:58,31-Jan-02 21:27,Northwestern,C6C1CD4F1
7562927,D*,63,1769929,636.7,89.8,20.8,03-Sep-03 13:02,11-Feb-02 03:17,ckrusen,library
7569559,D ,64,6583843,634.0,-42.5,4.5,14-Sep-03 06:48,13-Feb-02 20:54,prog,Bertserv
7683989,D*,64,6643098,596.0,-48.0,8.0,15-Sep-03 20:15,23-Mar-02 20:07,Team_LiTH,Miro
7726933,D ,64,1205120,586.1,-0.3,30.7,13-Sep-03 12:34,02-Apr-02 16:21,S69774,C74DDF5F9
7871873,D ,64,1411584,546.9,11.9,24.9,07-Sep-03 16:35,11-May-02 22:56,TimC,mana
7895299,D ,64,6402687,540.1,-48.7,11.3,13-Sep-03 02:12,18-May-02 17:25,molotov,Oshmadu1
7616249,D ,64,6966218,535.5,-49.8,10.2,13-Sep-03 00:37,23-May-02 07:20,S23870,C80B77BFD
7650451,D ,64,2986496,514.5,-52.0,8.0,04-Sep-03 20:35,13-Jun-02 08:38,Mampo,C9DBB4AE8
7979249,D*,63,6562279,512.9,0.2,-0.8,29-Aug-03 00:20,14-Jun-02 21:10,Starchild,Bob3
8015797,D ,64,2344303,504.1,-15.0,25.0,07-Sep-03 20:23,23-Jun-02 17:14,u48900,jk
8033573,D*,63,,500.5,-40.8,14.2,05-Sep-03 00:18,27-Jun-02 07:26,Team_Prime_Rib,Tasuke22
8033539,D*,63,,500.5,-47.8,12.2,05-Sep-03 00:18,27-Jun-02 07:26,Team_Prime_Rib,Tasuke22
8033491,D*,63,,500.5,-54.8,5.2,05-Sep-03 00:18,27-Jun-02 07:26,Team_Prime_Rib,Tasuke22
8033429,D ,64,,500.5,232.8,85.8,07-Nov-03 13:16,27-Jun-02 07:26,Team_Prime_Rib,hydro
8044051,D ,64,4987901,498.0,20.9,80.9,06-Nov-03 17:14,29-Jun-02 19:14,S61506,CDFB9A760
8123959,D ,64,5517631,473.0,23.0,77.0,29-Oct-03 20:50,24-Jul-02 20:02,Juanma,Juanma
8132701,D*,63,,469.7,-33.8,14.2,05-Sep-03 00:18,28-Jul-02 03:04,Team_Prime_Rib,Tasuke22
8133707,D ,63,,469.1,53.2,53.2,03-Sep-03 23:32,28-Jul-02 16:01,CaptainEntropy,
8140369,D ,64,4911423,467.1,-30.2,20.8,03-Sep-03 15:34,30-Jul-02 17:33,S75183,C23251609
8154929,D*,64,5438272,465.2,-47.0,6.0,15-Sep-03 18:48,01-Aug-02 15:37,tsc,zentra01
8186329,D ,64,4632416,457.3,-27.1,21.9,04-Sep-03 16:28,09-Aug-02 12:42,darione,darione
8179289,D ,64,3458623,456.3,20.0,64.0,16-Oct-03 20:53,10-Aug-02 11:46,S78480,C2BCE0698
8313587,D ,65,6036928,439.4,-39.3,20.7,06-Sep-03 12:27,27-Aug-02 09:13,S70018,CF5F22CE2
8344087,D ,64,8126209,429.8,2.5,62.5,09-Nov-03 07:46,06-Sep-02 00:29,bhamrin,Aptiva
8347921,D ,65,7619483,429.3,-55.3,4.7,11-Sep-03 11:24,06-Sep-02 13:12,clown3k,cassiopeia
8008237,D ,64,5029888,416.4,-52.4,7.6,04-Sep-03 09:52,19-Sep-02 10:06,rats,frogs
7735939,D*,64,7021418,413.5,-55.3,4.7,27-Aug-03 12:01,22-Sep-02 08:58,S81156,C58ACE298
8410643,D ,65,6438464,413.0,20.4,80.4,03-Nov-03 04:04,22-Sep-02 19:24,mrvs,texa
8440373,D ,65,6533633,407.8,-24.8,35.2,13-Oct-03 00:17,28-Sep-02 00:50,nevek,C0D5A63AB
8304437,D ,65,985287,403.7,13.8,66.8,09-Nov-03 13:40,02-Oct-02 02:38,Criswell,OSHO
8468137,D ,65,,403.0,34.6,54.6,03-Nov-03 10:53,02-Oct-02 18:49,Mezzo,mw_vu2
7252181,D*,64,,402.4,5.1,5.1,17-Jul-03 22:37,03-Oct-02 09:04,norbert,a2
7233287,D*,64,,402.4,5.1,5.1,17-Jul-03 22:42,03-Oct-02 09:03,norbert,a4
7624003,D ,65,6080914,397.3,34.8,48.8,28-Oct-03 15:53,08-Oct-02 12:50,Z,MarkW
8512033,D ,65,258240,392.2,14.9,60.9,09-Nov-03 16:30,13-Oct-02 14:11,WileECoyote,sharkyjr
8518519,D ,65,4521983,389.9,9.8,9.8,15-Sep-03 13:57,15-Oct-02 21:20,kaffe,jobbet
8528843,D ,65,2293312,388.0,-10.3,31.8,14-Sep-03 13:54,17-Oct-02 18:59,mbond,C172C0D2B
8554901,D ,65,2802312,382.8,-7.3,9.8,15-Sep-03 13:05,22-Oct-02 23:02,crestwicke,h_hcc2
8571053,D ,64,6337141,380.3,-40.3,18.7,01-Sep-03 11:04,25-Oct-02 12:09,belhadj,C0C8D6EA9
8578243,D ,65,8180371,378.0,-24.0,36.0,12-Oct-03 19:32,27-Oct-02 19:49,RG3,3A
8125301,D*,64,5698391,376.5,0.5,60.5,15-Oct-03 07:24,29-Oct-02 08:14,42014685,Sonne
8156299,D ,64,,375.5,57.5,57.5,08-Sep-03 06:26,30-Oct-02 07:54,PaJaSoft,PII
8155897,D ,64,,375.5,57.5,57.5,08-Sep-03 06:26,30-Oct-02 07:54,PaJaSoft,PII
9000209,D ,65,95590,375.0,142.8,66.8,09-Nov-03 13:40,30-Oct-02 20:00,Criswell,OSHO
8597263,D ,64,,373.4,-14.5,18.5,15-Sep-03 08:58,01-Nov-02 09:02,S09625,Cel14
8610169,D ,65,1742592,370.2,151.9,25.9,08-Sep-03 16:30,04-Nov-02 14:24,S82501,C2682F19D
7939829,D*,64,1045017,368.4,6.7,66.7,03-Nov-03 11:49,06-Nov-02 10:59,S09625,Cel04
8621897,D ,65,8969,367.2,125.8,66.8,09-Nov-03 13:40,07-Nov-02 15:19,Criswell,OSHO
8621867,D ,65,9230,367.2,109.8,66.8,09-Nov-03 13:40,07-Nov-02 15:19,Criswell,OSHO
8621863,D ,65,8900,367.2,93.8,66.8,09-Nov-03 13:40,07-Nov-02 15:19,Criswell,OSHO
8621857,D ,65,15698,367.2,77.8,66.8,09-Nov-03 13:40,07-Nov-02 15:19,Criswell,OSHO
8621843,D ,65,8785,367.2,60.8,66.8,09-Nov-03 13:40,07-Nov-02 15:19,Criswell,OSHO
8622473,D ,65,1785547,367.0,-19.0,32.0,14-Sep-03 20:54,07-Nov-02 20:48,S02664,P3400Mz
8638339,D ,65,4156415,363.3,-21.4,30.6,13-Sep-03 10:17,11-Nov-02 13:14,sNYder,1
8136893,D*,64,7667713,362.5,-49.9,7.1,15-Sep-03 21:25,12-Nov-02 06:11,primenator,C33B72AA6
8653657,D ,65,1527616,360.2,22.6,59.6,08-Nov-03 10:40,14-Nov-02 15:42,JC,PDU
8086411,D*,64,,354.3,18.8,60.8,09-Nov-03 15:59,20-Nov-02 11:34,Aillas,Lexa
8085239,D*,64,,354.3,10.8,60.8,09-Nov-03 15:59,20-Nov-02 11:34,Aillas,Lexa
8081419,D ,65,6025177,354.3,2.8,60.8,09-Nov-03 15:59,20-Nov-02 11:33,Aillas,Lexa
8686297,D ,65,,352.4,-10.5,9.5,22-Jul-03 07:33,22-Nov-02 10:07,JC,GVA

garo · 2003-11-10, 17:47

Thanks for the list GP2. I believe that at least some of the exponents should be excluded from the list above. Our discussions in the thread should help refine this. One criteria would be to exclude exponents that have over 1 million completed iterations. That may be overly broad though. My concern is that the "days to go" column is not an accurate predictor of how long it would take to finish an exponent and yet I know for a fact that many of the exponents listed above, if left alone, will complete within the next month or two.

So our concern here is to minimize wasted CPU - by checking an exponent when the original assignee is likely to return a result soon - and yet to not let the exponents drag on forever.

I have been trying to help in this consideration in a legal way - and also to discourage poachers - by mopping up expiring exponents every evening at 0600GMT - I'm on the west coast. I've caught a large numebr of these and have managed to erduce the list of outstanding exponents above 9M significantly.

To avoid unnecessary triple checks I've adopted a "maturing" strategy whereby I let exponents that had more than 1 million iterations complete before they expired wait for about 30 days before I work on them. I have paid close attention to this problem since I have about 40 triplechecks for exponents I doublechecked but then I only test expired doublechecks. George I'd be interested to know your thoughts if any on this.

GP2 since the days to go figure is unreliable, I'd suggest that you look at an older status file and compare the number of iterations completed for each exponent and figure out the approximate rate at which an exponent is being completed. That would give us a much better idea of which exponents fall under the > 6 month to complete category.

Finally, I would like to complete all the Team_Prime_Rib exponents listed above myself so I'm going to ask the assignees if they can either bump those exponents to the top of their worktodo or let me take those exponents over.

GP2 · 2003-11-10, 22:05

[Edit: this had to be split off from the previous post in this thread when I added links to progress graphs for exponents (because the board software imposes a maximum byte length per message]

Click on any exponent to see a graph of its progress

If we add first-time checks under 10M, only 3 exponents are added:

8206057, ,64,,420.8,95.2,95.2,16-Oct-03 00:36,15-Sep-02 00:19,Team_Prime_Rib,trif_pest
8206021, ,64,,420.8,95.2,95.2,16-Oct-03 00:36,15-Sep-02 00:09,Team_Prime_Rib,trif_pest
8206013, ,64,,420.8,95.2,95.2,16-Oct-03 00:36,15-Sep-02 01:16,Team_Prime_Rib,trif_pest

For first-time or double-checking tests in the 10M-11M range, we have:

10412737, ,65,6686448,1260,27.8,59.8,08-Nov-03 14:32,29-May-00 04:01,dtlgc,Kids-p3-900
10964861, ,65,2594367,1171,2808,69.9,05-Nov-03 17:38,25-Aug-00 18:50,gediminas,ihome
10599221, ,65,2135004,1060,543.8,54.8,29-Oct-03 15:36,14-Dec-00 09:13,smoros,padrelin
11840051, ,65,11534336,864.3,-48.0,6.0,15-Sep-03 18:57,28-Jun-01 11:23,MPICE,MS-QUATTRO
11000347, ,66,23304,375.0,44.8,66.8,09-Nov-03 13:40,30-Oct-02 20:01,Criswell,OSHO
10271543,D ,65,8742399,223.9,5.5,60.5,09-Nov-03 07:44,30-Mar-03 21:02,NickGlover,Aptiva
10599223,D ,65,,217.0,223.4,60.4,09-Nov-03 04:56,06-Apr-03 18:11,GW,P4-1600

GP2 · 2003-11-10, 22:11

Quote:

Originally posted by garo
GP2 since the days to go figure is unreliable, I'd suggest that you look at an older status file and compare the number of iterations completed for each exponent and figure out the approximate rate at which an exponent is being completed. That would give us a much better idea of which exponents fall under the > 6 month to complete category.

I modified the previous posts: click on any exponent to see a graph of its progress (if it's not clickable, no iterations have been done yet).

Each graph is scaled so that the top of the graph represents 100% completion.

From visual inspection it's easy to see which ones are hopeless cases, and which ones still have some hope. I'm not sure how you can translate that into a mathematical formula though, because not all the curves are fit by a straight line.

The graphs can be generated automatically and quickly, and only a small set of double-checking exponents have been running longer than a year. So rather than come up with a mathematical formula to estimate time of completion, maybe George could just use visual inspection to decide which exponents to poach.

GP2 · 2003-11-10, 22:21

The graphs show only last-updated time and iteration count. In a handful of cases, the data show more than one user's work (a previous user who let it expire half-completed, etc)

garo · 2003-11-10, 22:40

Cool! These graphs are just what the doctor ordered. I'm sure they'll go a long way in helping George decide what to release. If he feels manual inspection is too time consuming I guess we could try and come up with some automated method. But let us wait to hear from him first.

Again great work!!

Prime95 · 2003-11-11, 00:02

GP2, you are amazing!! I'm perfectly happy with you formulating a list of hopeless causes (no hope for completion in 6 months).

One last check - can you match the userid with some past results. We should be wary of poaching assignments from someone using non-prime95 clients. Mlucas and Glucas do not update the iteration counts - so it goes from zero to done in one fell swoop.

Prime95 · 2003-11-11, 00:05

Quote:

Originally posted by garo
To avoid unnecessary triple checks I've adopted a "maturing" strategy whereby I let exponents that had more than 1 million iterations complete before they expired wait for about 30 days before I work on them. I have paid close attention to this problem since I have about 40 triplechecks for exponents I doublechecked but then I only test expired doublechecks. George I'd be interested to know your thoughts if any on this.

It sounds like you are working within the primenet rules - I see no problems. Good work!

GP2 · 2003-11-12, 06:49

Somebody didn't wait on 7137769... it got poached and completed.

It probably would have (just) made the six-month cut (see graph), with projected completion around April or May of next year.

2003-11-10, 04:01	#1
Prime95 P90 years forever! Aug 2002 Yeehaw, FL 5×23×71 Posts	Officially poaching very old exponents I propose that currently assigned doublechecks below 9 million that have been outstanding for 2 years and at their current rate of progress will not finish within 6 months, be officially poached and reassigned. Comments? Should the 2 year value be reduced to 1.5 years? 1 year? Other suggested adjustments? Last fiddled with by Prime95 on 2003-11-10 at 04:02

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Poaching	blip	Data	8	2016-01-30 01:59
GPU to 72 spider poaching exponents?	ixfd64	GPU to 72	2	2014-07-27 20:38
PrimeGrid has officially joined our effort!	Joe O	Sierpinski/Riesel Base 5	1	2011-01-07 20:38
FYI: GIMPS now officially accepting donations	ixfd64	Lounge	6	2010-03-03 01:47
Poaching and v5	PrimeCruncher	PrimeNet	6	2004-04-05 19:17

2003-11-10, 04:52	#2
garo Aug 2002 Termonfeckin, IE 2⁴×173 Posts	I agree George. In fact I think the threshold should be reduced to 1 year. There are several exponents that have been assigned for between 400 and 700 days.

2003-11-10, 17:47	#4
garo Aug 2002 Termonfeckin, IE 2⁴×173 Posts	Thanks for the list GP2. I believe that at least some of the exponents should be excluded from the list above. Our discussions in the thread should help refine this. One criteria would be to exclude exponents that have over 1 million completed iterations. That may be overly broad though. My concern is that the "days to go" column is not an accurate predictor of how long it would take to finish an exponent and yet I know for a fact that many of the exponents listed above, if left alone, will complete within the next month or two. So our concern here is to minimize wasted CPU - by checking an exponent when the original assignee is likely to return a result soon - and yet to not let the exponents drag on forever. I have been trying to help in this consideration in a legal way - and also to discourage poachers - by mopping up expiring exponents every evening at 0600GMT - I'm on the west coast. I've caught a large numebr of these and have managed to erduce the list of outstanding exponents above 9M significantly. To avoid unnecessary triple checks I've adopted a "maturing" strategy whereby I let exponents that had more than 1 million iterations complete before they expired wait for about 30 days before I work on them. I have paid close attention to this problem since I have about 40 triplechecks for exponents I doublechecked but then I only test expired doublechecks. George I'd be interested to know your thoughts if any on this. GP2 since the days to go figure is unreliable, I'd suggest that you look at an older status file and compare the number of iterations completed for each exponent and figure out the approximate rate at which an exponent is being completed. That would give us a much better idea of which exponents fall under the > 6 month to complete category. Finally, I would like to complete all the Team_Prime_Rib exponents listed above myself so I'm going to ask the assignees if they can either bump those exponents to the top of their worktodo or let me take those exponents over.

2003-11-10, 22:05	#5
GP2 Sep 2003 5035₈ Posts	[Edit: this had to be split off from the previous post in this thread when I added links to progress graphs for exponents (because the board software imposes a maximum byte length per message] Click on any exponent to see a graph of its progress If we add first-time checks under 10M, only 3 exponents are added: 8206057, ,64,,420.8,95.2,95.2,16-Oct-03 00:36,15-Sep-02 00:19,Team_Prime_Rib,trif_pest 8206021, ,64,,420.8,95.2,95.2,16-Oct-03 00:36,15-Sep-02 00:09,Team_Prime_Rib,trif_pest 8206013, ,64,,420.8,95.2,95.2,16-Oct-03 00:36,15-Sep-02 01:16,Team_Prime_Rib,trif_pest For first-time or double-checking tests in the 10M-11M range, we have: 10412737, ,65,6686448,1260,27.8,59.8,08-Nov-03 14:32,29-May-00 04:01,dtlgc,Kids-p3-900 10964861, ,65,2594367,1171,2808,69.9,05-Nov-03 17:38,25-Aug-00 18:50,gediminas,ihome 10599221, ,65,2135004,1060,543.8,54.8,29-Oct-03 15:36,14-Dec-00 09:13,smoros,padrelin 11840051, ,65,11534336,864.3,-48.0,6.0,15-Sep-03 18:57,28-Jun-01 11:23,MPICE,MS-QUATTRO 11000347, ,66,23304,375.0,44.8,66.8,09-Nov-03 13:40,30-Oct-02 20:01,Criswell,OSHO 10271543,D ,65,8742399,223.9,5.5,60.5,09-Nov-03 07:44,30-Mar-03 21:02,NickGlover,Aptiva 10599223,D ,65,,217.0,223.4,60.4,09-Nov-03 04:56,06-Apr-03 18:11,GW,P4-1600

2003-11-10, 22:21	#7
GP2 Sep 2003 3×863 Posts	The graphs show only last-updated time and iteration count. In a handful of cases, the data show more than one user's work (a previous user who let it expire half-completed, etc)

2003-11-10, 22:40	#8
garo Aug 2002 Termonfeckin, IE 2⁴·173 Posts	Cool! These graphs are just what the doctor ordered. I'm sure they'll go a long way in helping George decide what to release. If he feels manual inspection is too time consuming I guess we could try and come up with some automated method. But let us wait to hear from him first. Again great work!!

2003-11-11, 00:02	#9
Prime95 P90 years forever! Aug 2002 Yeehaw, FL 5×23×71 Posts	GP2, you are amazing!! I'm perfectly happy with you formulating a list of hopeless causes (no hope for completion in 6 months). One last check - can you match the userid with some past results. We should be wary of poaching assignments from someone using non-prime95 clients. Mlucas and Glucas do not update the iteration counts - so it goes from zero to done in one fell swoop.

2003-11-12, 06:49	#11
GP2 Sep 2003 3×863 Posts	Somebody didn't wait on 7137769... it got poached and completed. It probably would have (just) made the six-month cut (see graph), with projected completion around April or May of next year.