Redoing factoring work done by unreliable machines

tha · 2016-11-01, 15:17

I noticed user TJAOI did upload some some factoring work today in the range between 63 and 64 bits. They include factors that were missed earlier when the factoring assignments were handed out by the server or done manually mostly about ten years ago.

A lot of factors were missed by user "1997rj7" who seems to have done a lot of factoring work back then on a unreliable machine.

Can we ( I mean someone ) create a worktodo file that would retest the factoring work by this faulty machine, and maybe other proven faulty machines?

There probably are enough volunteers here who would want to work on such a worktodo file. I do.

petrw1 · 2016-11-01, 15:33

My 2 cents ...

If TJAOI is already closing in on 64 bits is it not safe to assume he/she has found all missing factors by any and all unreliable machines up to that level?

Or are you asking about assignments above 64 bits?

Or as long as TJAOI is there do we just let him clean up?

LaurV · 2016-11-01, 15:55

It is about higher bitlevels done by unreliable machines/users, TJAOI will not reach 65 and 66 soon (years). If we find missed factors, we can save some LL.

I am with tha, but we better see the list first. It may not worth a re-check.

petrw1 · 2016-11-01, 16:03

Fair enough.

Would reviewing the list from TJAOI be the best way to seeing who missed the most and redo all of theirs above xx bits?Do we let TJAOI have 64 to 65 or 66?

I assume 1999rj7 isn't the only one.

Maybe Madpoo can use a similar magic query he used to get the required DC's.

Prime95 · 2016-11-01, 18:17

There are 33075 results between 1999 and 2010 where the user tested to 2^64 or greater.

Can you narrow down which computer or which dates resulted in missed factors?

chalsall · 2016-11-01, 19:05

Quote:

Originally Posted by Prime95

Can you narrow down which computer or which dates resulted in missed factors?

I would be surprised if any new factors, or cheaters, are found.

But whatever rocks your boat. It's your kit / time / energy / money.

Please prove us wrong. Or, alternatively, suggest we are correct.

airsquirrels · 2016-11-01, 19:45

Unless we know with near certainty that a check from a given machine is wrong, the math to recheck all their results just doesn't add up. Better to do other TF work where you have a ~1/70 chance of saving an LL test vs. 1/700 if 10% of their checks are wrong.

VBCurtis · 2016-11-01, 20:16

Quote:

Originally Posted by airsquirrels

Unless we know with near certainty that a check from a given machine is wrong, the math to recheck all their results just doesn't add up. Better to do other TF work where you have a ~1/70 chance of saving an LL test vs. 1/700 if 10% of their checks are wrong.

Unless the checks take less than 1/10th the time... Given the choice of 74-75 bit TF work or double-checking 64-67 for someone with a 5% error rate, I'm going to find more factors with the retest per unit time.

chalsall · 2016-11-01, 20:24

Quote:

Originally Posted by VBCurtis

Unless the checks take less than 1/10th the time... Given the choice of 74-75 bit TF work or double-checking 64-67 for someone with a 5% error rate, I'm going to find more factors with the retest per unit time.

That assumes the person you are retesting was cheating.

As I said before, I did that kind of retesting myself. And I never found a cheater.

Put PLEASE! Rerun that test. Spend that money.

We will all be *very* interested if you can find a cheater. I didn't.

airsquirrels · 2016-11-01, 20:34

Quote:

Originally Posted by VBCurtis

Unless the checks take less than 1/10th the time... Given the choice of 74-75 bit TF work or double-checking 64-67 for someone with a 5% error rate, I'm going to find more factors with the retest per unit time.

My math was based on the idea of doing 64-67 ahead of the wave, all the way up to the 999..M mark. Doing 64-67 for new exponents gets you more exponents cleared than rechecking 64-67 at a 5-10% error rate.

Rechecking where that range is queued by someone else for 74-75 TF or LL/LL-DC, maybe you save some CPU.

I've thought about doing this for some of my own work. I once discovered a card that had a 1/100 error rate on the mfakto self-test, but had been TFing that way for months. I fixed the issue with the card (driver), but at the end of the day I could not justify going back over months of work on another card.

GP2 · 2016-11-01, 21:46

Quote:

Originally Posted by tha

I noticed user TJAOI did upload some some factoring work today in the range between 63 and 64 bits. They include factors that were missed earlier when the factoring assignments were handed out by the server or done manually mostly about ten years ago.

A lot of factors were missed by user "1997rj7" who seems to have done a lot of factoring work back then on a unreliable machine.

You mention TF:63-64 factors uploaded by TJAOI today (Nov 1). Below is the list of all such exponents I see currently in the "cleared" listing, all uploaded by TJAOI on Oct 31 at 20:53.

However, this list doesn't match your description. Most of these exponents are > 100M and presumably done by GPU rather than mprime, more recently than ten years ago, and the "culprits" are all sorts of people, not "1997rj7". Can you clarify which exponents are involved? Perhaps they were uploaded a couple of days ago rather than today, and have now disappeared off the "cleared" listing.

Code:

13128419  9355676489974796977
16556363  14981465658515268727
17841221  14001530903059537847
33088541  15288705018849658519
33122153  15040775811986445743
33142423  16944060969483680321
66104189  14713328536491742193
66195611  17797707556291217479
66219407  18335189697371799671
132131101 12808666766487948431
132150919 16344484188919006721
132308147 17038091606573716753
132343297 17972353937967188791
132385271 15151310052639715127
132395539 16824273427475392943
132545783 18240167095628118409
132833819 15808866343324379257
188141959 15845114013007218697
188376091 12065395018065595007
188376691 11969226960989196959
188377817 11320006072113354703
188385983 12597013385833378609
188398271 14516492533152659617
188403289 9905263616364881711
188416717 12340120614482786327
188422349 11973480210326952257
188423111 10278928002296356679
188424043 13441142530325722361
188516227 9464986542542486983
188603507 10064823530597873743
188644829 13020094912207213609
188659091 9606729918184691623
188661097 10870076138159380591
189342383 15015781421576408897
191135237 9312579326180739689
191138693 11607230100233257247
191146519 10866684654858738943
191148143 11380842450322510967
191154797 14011717158896260159
191199653 9259273790380333793
191202373 13217938705682683801
191206313 10389072086492246041
191212421 11480174189911514753
191214061 12428774210084668199
191272363 16308593842567686599
191280917 13551416122706799847
191283707 13721295112219429967
191284139 17057363340549679129
191284193 16193171405740999511
191284631 10193439850514740711
191285009 11336413647449442047
191289121 10308153713902389887
191289379 12487248387801206927
191298493 16239261693910377457
264306487 17955951716335219583
264406711 18347513166156461233
264483083 14255264859002142497
264728713 17914617690362674297
264750253 17582831426707578127
265111681 18399929025027698543
265204549 15246134958094701127
265278163 16579976415599143049
265299017 18161112424540643177
265409411 14884837943006987201
265463221 18387502900883613209
265475671 14072108164513011031
265670557 17191259690603773153
265791527 16302399353095623863
265832731 15351754224209841583
265841143 16919305513623287513
266031251 17458963290228990097
266065001 13534196407821207281
266098873 16250996031866081911
266220709 14504933610126739609
266235289 17353565049548116639
266266877 16418208917348413039
266283067 16198750327486529623
529153661 15871987080937560823
529187227 13033777671846512369
529283413 15106077604296686671
529481773 17948255019065261431
529523899 14890814210221751609
529551739 14401007882723364313
529617173 16049728830281953151
529669991 15995519075412584833
529742561 14997925876285859399
530067059 16460243170402327097
530325833 14747595488403083087
530455283 16463696040069927247
530477527 17759528532738450391
530684549 17768145060210951743
530804941 13603584024715247887
530807411 17697241496483509999
530894723 15968197876077397583
531009443 15466019095930187177
531073867 15701511812068982017
531110977 13139910783278687081
531348133 13935413648560903241
531373729 17734242856632963457
531687227 16349594804120226871
531760871 14572758899790079553
531998983 18065865027680390959
532077841 16816375713731600401
532139263 17811314838179232641
532268537 14886105606806850649
532405849 15119079382147455191

2016-11-01, 15:17	#1
tha Dec 2002 327₁₆ Posts	Redoing factoring work done by unreliable machines I noticed user TJAOI did upload some some factoring work today in the range between 63 and 64 bits. They include factors that were missed earlier when the factoring assignments were handed out by the server or done manually mostly about ten years ago. A lot of factors were missed by user "1997rj7" who seems to have done a lot of factoring work back then on a unreliable machine. Can we ( I mean someone ) create a worktodo file that would retest the factoring work by this faulty machine, and maybe other proven faulty machines? There probably are enough volunteers here who would want to work on such a worktodo file. I do.

2016-11-01, 15:55	#3
LaurV Romulan Interpreter Jun 2011 Thailand 2³×19×61 Posts	It is about higher bitlevels done by unreliable machines/users, TJAOI will not reach 65 and 66 soon (years). If we find missed factors, we can save some LL. I am with tha, but we better see the list first. It may not worth a re-check. Last fiddled with by LaurV on 2016-11-01 at 15:56

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Coordination thread for redoing P-1 factoring	ixfd64	Lone Mersenne Hunters	80	2021-01-19 17:17
Require P-1 and other factoring work to be done on reliable machines	tha	PrimeNet	66	2015-05-20 15:45
Work transfer between 32 and 64 bit machines	tichy	Software	11	2011-01-07 22:57
Redoing work?	MooMoo2	Riesel Prime Search	2	2010-09-15 17:33
5.98M to 6.0M: redoing factoring to 62 bits	GP2	Lone Mersenne Hunters	0	2003-11-19 01:30

2016-11-01, 15:33	#2
petrw1 1976 Toyota Corona years forever! "Wayne" Nov 2006 Saskatchewan, Canada 4552₁₀ Posts	My 2 cents ... If TJAOI is already closing in on 64 bits is it not safe to assume he/she has found all missing factors by any and all unreliable machines up to that level? Or are you asking about assignments above 64 bits? Or as long as TJAOI is there do we just let him clean up?

2016-11-01, 16:03	#4
petrw1 1976 Toyota Corona years forever! "Wayne" Nov 2006 Saskatchewan, Canada 2³×569 Posts	Fair enough. Would reviewing the list from TJAOI be the best way to seeing who missed the most and redo all of theirs above xx bits?Do we let TJAOI have 64 to 65 or 66? I assume 1999rj7 isn't the only one. Maybe Madpoo can use a similar magic query he used to get the required DC's.

2016-11-01, 18:17	#5
Prime95 P90 years forever! Aug 2002 Yeehaw, FL 2³·919 Posts	There are 33075 results between 1999 and 2010 where the user tested to 2^64 or greater. Can you narrow down which computer or which dates resulted in missed factors?

2016-11-01, 19:45	#7
airsquirrels "David" Jul 2015 Ohio 11×47 Posts	Unless we know with near certainty that a check from a given machine is wrong, the math to recheck all their results just doesn't add up. Better to do other TF work where you have a ~1/70 chance of saving an LL test vs. 1/700 if 10% of their checks are wrong.