Discrepancy?

[DJones]

I can't see this anywhere in the FAQ or in the introduction to newbies, and I hope I'm asking this in the right place. Apologies if I'm not. Anyway, according to the Primenet Assignments Report (snippet below) there are still some 14xxxxxx exponents being LL-ed for the first time. Whereas according to Mr. Woltman's GIMPS status page "All exponents below 18,000,900 have been tested at least once." So... what's going on?

 prime      fact  current          days
exponent    bits iteration  run / to go / exp   date updated     date assigned   account ID     computer ID
-------- -- ---- ---------  -----------------  ---------------  ---------------  -------------- ------------
14264087     66   8991949     2.6  77.4  85.4                   23-Nov-06 21:58  aquazer        aquazer
14271473     66   2214507    66.0  45.2  86.2  24-Nov-06 16:30  21-Sep-06 11:15  S319085        C1885EBB3
14288077     66    484893    28.4  27.6  59.6                   29-Oct-06 02:12  xu7118         xulei
14325617     66   4520449    64.7  -0.7  23.3                   22-Sep-06 18:52  gselabs        SCI-125
14350387     66   7872400    10.6   2.8  59.8  25-Nov-06 06:37  15-Nov-06 20:24  haveland       CB5BECAC7
14355337     66              41.4  10.0  70.0  24-Nov-06 11:59  16-Oct-06 01:29  .              CE4A8D3E7
14360881     66      6455    19.5  -8.5  51.5                   07-Nov-06 00:24  S357064        C85888470
14370313     66   7716927    10.9  17.1  59.1  24-Nov-06 14:30  15-Nov-06 13:47  Team_Prime_Rib Reboot_It135
14371451     66   6287824    10.7 513.4  56.4  21-Nov-06 20:45  15-Nov-06 19:10  WileECoyote    S18365
14395531     66   9496321    51.0   9.0  69.0  17-Nov-06 10:19  06-Oct-06 10:21  anbi           gruppi1
14405561 D   66   9660370   250.7  -3.9  56.1  17-Nov-06 13:58  20-Mar-06 19:44  zwirunio       20
14440567 D   66   2025791   182.8   0.6  57.6  23-Nov-06 02:58  27-May-06 16:21  shaneamy       Laptop
14440661 D   66             182.8   3.6  57.6  23-Nov-06 02:58  27-May-06 16:21  shaneamy       Laptop
14443531 D   66             182.8   7.6  57.6  23-Nov-06 02:58  27-May-06 16:21  shaneamy       Laptop
14443889 D   66             182.8  10.6  57.6  23-Nov-06 02:58  27-May-06 16:21  shaneamy       Laptop
14447681 D   66             182.8  14.6  57.6  23-Nov-06 02:58  27-May-06 16:21  shaneamy       Laptop

[Mini-Geek]

Originally Posted by DJones

I can't see this anywhere in the FAQ or in the introduction to newbies, and I hope I'm asking this in the right place. Apologies if I'm not. Anyway, according to the Primenet Assignments Report (snippet below) there are still some 14xxxxxx exponents being LL-ed for the first time. Whereas according to Mr. Woltman's GIMPS status page "All exponents below 18,000,900 have been tested at least once." So... what's going on?

 prime      fact  current          days
exponent    bits iteration  run / to go / exp   date updated     date assigned   account ID     computer ID
-------- -- ---- ---------  -----------------  ---------------  ---------------  -------------- ------------
14264087     66   8991949     2.6  77.4  85.4                   23-Nov-06 21:58  aquazer        aquazer
14271473     66   2214507    66.0  45.2  86.2  24-Nov-06 16:30  21-Sep-06 11:15  S319085        C1885EBB3
14288077     66    484893    28.4  27.6  59.6                   29-Oct-06 02:12  xu7118         xulei
14325617     66   4520449    64.7  -0.7  23.3                   22-Sep-06 18:52  gselabs        SCI-125
14350387     66   7872400    10.6   2.8  59.8  25-Nov-06 06:37  15-Nov-06 20:24  haveland       CB5BECAC7
14355337     66              41.4  10.0  70.0  24-Nov-06 11:59  16-Oct-06 01:29  .              CE4A8D3E7
14360881     66      6455    19.5  -8.5  51.5                   07-Nov-06 00:24  S357064        C85888470
14370313     66   7716927    10.9  17.1  59.1  24-Nov-06 14:30  15-Nov-06 13:47  Team_Prime_Rib Reboot_It135
14371451     66   6287824    10.7 513.4  56.4  21-Nov-06 20:45  15-Nov-06 19:10  WileECoyote    S18365
14395531     66   9496321    51.0   9.0  69.0  17-Nov-06 10:19  06-Oct-06 10:21  anbi           gruppi1
14405561 D   66   9660370   250.7  -3.9  56.1  17-Nov-06 13:58  20-Mar-06 19:44  zwirunio       20
14440567 D   66   2025791   182.8   0.6  57.6  23-Nov-06 02:58  27-May-06 16:21  shaneamy       Laptop
14440661 D   66             182.8   3.6  57.6  23-Nov-06 02:58  27-May-06 16:21  shaneamy       Laptop
14443531 D   66             182.8   7.6  57.6  23-Nov-06 02:58  27-May-06 16:21  shaneamy       Laptop
14443889 D   66             182.8  10.6  57.6  23-Nov-06 02:58  27-May-06 16:21  shaneamy       Laptop
14447681 D   66             182.8  14.6  57.6  23-Nov-06 02:58  27-May-06 16:21  shaneamy       Laptop

Those are double-checks. A LL test has a ~1% chance to make an error, so all exponents are tested twice (double-checked) to make sure their residues match. All below 18,000,900 have indeed been tested once, but there is a ~1% chance on each of those that the residue returned was incorrect, meaning that it's possible we missed a prime.

[DJones]

Originally Posted by Mini-Geek

Those are double-checks. A LL test has a ~1% chance to make an error, so all exponents are tested twice (double-checked) to make sure their residues match. All below 18,000,900 have indeed been tested once, but there is a ~1% chance on each of those that the residue returned was incorrect, meaning that it's possible we missed a prime.

I'm well aware that double-checking takes place, thank you. The report however puts a "D" next to double-checks; I even deliberately included some on the snippet so that people could see the difference. The first few lines of the snippet though, do not have that "D", so my question still stands.

[Prime95]

When in doubt, ignore the PrimeNet server's report. It could be that these exponents were re-released as first time tests ages ago because the first test had a suspect error code. Or the server may simply be confused.

[S485122]

I asked the same question some time ago and got an answer : Status of PrimeNet

[DJones]

Thank you both for your responses - it's nice to see some people read the question carefully before answering.

Could a simple database query not alter the flag on all tests <18000000 to be double-checks as opposed to first-timers, or would the time that takes to run and the risk of it causing more problems make it not worth the hassle?

[S485122]

GIMPS accepts any result.
PrimeNet only accepts results for assigments it distributed.

Users who got their assignments from PrimeNet played according to the rules and should not be penalised.

On the other hand there is some cleaning to do in the PrimeNet database : assigments that are more than 5 years from expiring for instance and no progress report...

[DJones]

Originally Posted by Jacob Visser

GIMPS accepts any result.
PrimeNet only accepts results for assigments it distributed.
Users who got their assignments from PrimeNet played according to the rules and should not be penalised.

Ah, yes, I'd forgotten about the scoring - I'm not that bothered by points but I appreciate that others are only concerned with leaderboards and stuff though. Whilst I agree that people who got their numbers from PrimeNet shouldn't have their points affected by results sent in by those who didn't, it's a little unfair to suggest that those who don't get their numbers from PrimeNet somehow aren't playing "according to the rules" - they are, after all, still voluntarily contributing to the project. Most of the people who test numbers without using PrimeNet (LMH, Mersenne-aries, etc) aren't interested in points in the first place.

Originally Posted by Jacob Visser

On the other hand there is some cleaning to do in the PrimeNet database : assignments that are more than 5 years from expiring for instance and no progress report...

Slow machines with sporadic internet access maybe? </clutching at straws>
I'd be (slightly) more concerned if they'd been *running* for five years and hadn't done a progress report...
I still can't believe someone is willing to let their machine run for five years just to test one number, but then humans have always baffled me as a most bizarre race.