mersenneforum.org

mersenneforum.org (https://www.mersenneforum.org/index.php)
-   PrimeNet (https://www.mersenneforum.org/forumdisplay.php?f=11)
-   -   OFFICIAL "SERVER PROBLEMS" THREAD (https://www.mersenneforum.org/showthread.php?t=5758)

Viliam Furik 2021-07-27 20:39

2 Attachment(s)
Screenshots. It seems to be correct now... I swear it was switched last time. I'll tell you if I see it happen again. Unless you made a stealth fix.

James Heinrich 2021-07-27 20:47

[QUOTE=Viliam Furik;584116]Screenshots. It seems to be correct now... I swear it was switched last time. I'll tell you if I see it happen again. Unless you made a stealth fix.[/QUOTE]I have not changed anything. If you notice it showing up incorrectly again for whatever reason please do provide another screenshot.

Zhangrc 2021-07-28 02:41

[QUOTE=Viliam Furik;584116]Screenshots. It seems to be correct now... I swear it was switched last time. I'll tell you if I see it happen again. Unless you made a stealth fix.[/QUOTE]

That error "TF result was not needed" is a bug, I doubt. The TF assignments still help to find more factors, and have not previously been done, so credit should be given.

James Heinrich 2021-07-28 14:31

[QUOTE=Zhangrc;584149]That error "TF result was not needed" is a bug[/QUOTE]Not a bug. The exponents have known factors, therefore are certainly not Mersenne Primes, therefore as far as GIMPS (Great Internet Mersenne [b]Prime[/b] Search) is concerned no further effort should be spent on this exponent. Users can of course spend their cycles wherever they want, but [u]for the purpose of finding the next Mersenne Prime[/u] any effort on an exponent with known factors is wasted effort.

kriesel 2021-07-29 16:58

PRP DC manual assignments are misrepresented as Cat 0 PRP
 
Assignments obtained today as 84M PRP DC (PRP second test of an exponent with PRP first test already reported) manually assigned are indicated in [URL]https://www.mersenne.org/workload/[/URL] as PRP, Cat 0, with 7 days allowed to start (or 7 days to complete and be reported, in the case of manual assignments on GPUs, which have no way to report progress). As PRP DC, 84M should be indicated as Cat 4, not Cat 0. ([URL="https://www.mersenne.org/thresholds/"]Cat 3 DC[/URL] today is exponents below 63724104.)
This is the case even after they are put on prime95 instances behind other work and expected completion dates reported by PrimeNet API to the server. This issue has lingered on for years and been repeatedly reported previously. Please look into implementing a fix. Otherwise the issue will continue on for several more years. I believe it also affects automatically issued PRP DC via PrimeNet API.

[CODE]CPU Name Core Exponent Work type Current cat Original State% Assigned Last update Est. Complete Days left Expires days
Manual testing 1 84661939  PRP 0 0 2021-07-29 2021-07-29 7
Manual testing 1 84662069  PRP 0 0 2021-07-29 2021-07-29 7
Manual testing 1 84662239  PRP 0 0 2021-07-29 2021-07-29 7
Manual testing 1 84662323  PRP 0 0 2021-07-29 2021-07-29 7
Manual testing 1 84663617  PRP 0 0 2021-07-29 2021-07-29 7
Manual testing 1 84663907  PRP 0 0 2021-07-29 2021-07-29 7
Manual testing 1 84665089  PRP 0 0 2021-07-29 2021-07-29 7
Manual testing 1 84665459  PRP 0 0 2021-07-29 2021-07-29 7
Manual testing 1 84666083  PRP 0 0 2021-07-29 2021-07-29 7
Manual testing 1 84666587  PRP 0 0 2021-07-29 2021-07-29 7
Kilroy_was_here 1 84666823  PRP 0 0 PRP 2021-07-29 2021-07-29 2021-08-04 6 30
Kilroy_was_here 2 84666931  PRP 0 0 2021-07-29 2021-07-29 2021-08-06 8 7
Manual testing 1 84667057  PRP 0 0 2021-07-29 2021-07-29 7
Manual testing 1 84667201  PRP 0 0 2021-07-29 2021-07-29 7
ostrich 1 84668011  PRP 0 0 2021-07-29 2021-07-29 2021-08-05 7 7
ostrich 2 84668407  PRP 0 0 PRP 2021-07-29 2021-07-29 2021-08-04 6 30
[/CODE]It seems only exponents checked out as PRP DC for exponents with LL first tests already completed, which is a separate manual assignments page menu selection from ordinary PRP DC, get correctly indicated as PRP-D currently and show appropriate Cat #'s and expiration dates:
[CODE]CPU Name Core Exponent Work type Current cat Original State% Assigned Last update Est. Complete Days left Expires days
Manual testing 1 61503457  PRP-D 2 3 2021-06-28 2021-06-28 89
Manual testing 1 61823711  PRP-D 3 3 2021-07-28 2021-07-28 119
Manual testing 1 61823731  PRP-D 3 3 2021-07-28 2021-07-28 119
[/CODE]

Uncwilly 2021-07-29 21:33

All PRP-CF suffer from this too.

S485122 2021-08-01 08:41

Exponent Status Distribution page
 
5 exponents are missing in the 57M range :[code] 57000000 55901 1 | 36254 19211 435 | 430 | |[/code]The 430 assigned exponents can be found using the active assignment page.

5 exponents are missing and they were already missing at the end of June. Other ranges also have missing exponents.

Jacob

S485122 2021-08-01 09:44

Strange P-1 results in the database
 
[url]https://www.mersenne.org/report_pminus1/?exp_hi=99[/url], M3 has three P-1 factoring attemps in the database : [url]https://www.mersenne.org/report_exponent/?exp_lo=3&exp_hi=&full=1[/url]
One would say their computer, their time, their electricity, ... [noparse];-)[/noparse] but I can't understand how those strange attempts found their way in the database. I strongly suspect the results in the database are fakes : the bounds are obviously inane. Why would new results be accepted for a known Mersenne prime ?

Jacob

Prime95 2021-08-01 17:44

[QUOTE=S485122;584536]

5 exponents are missing and they were already missing at the end of June. Other ranges also have missing exponents.

[/QUOTE]

Five exponents where the server is (too?) patiently waiting for a proof upload: [CODE]57310949
57311099
57326809
57326917
57331919[/CODE]

Prime95 2021-08-01 18:18

[QUOTE=Prime95;584561]Five exponents where the server is (too?) patiently waiting for a proof upload: [/QUOTE]

Something weird going on. The server is supposed to wait 60 days for a proof. At that point it deletes the assignment and makes the exponent available again. In this case, the assignment was deleted (or maybe never assigned in the first place?), but the exponent was not made available again. I've manually corrected the 5 exponents.

Two more that I did not fix:

60800659
60822823

There are also quite a few over 100 million, which does not concern me at the present time. I need to understand why this is happening to create a true fix.

kriesel 2021-08-01 18:47

These are all LL first test / PRP proof tandem exponents. Three PRP userids.

[M]57310949[/M] tservo
[M]57311099[/M] tservo
[M]57326809[/M] tservo
[M]57326917[/M] tservo
[M]57331919[/M] tservo

[M]60800659[/M] nescro
[M]60822823[/M] Lukas_J


All times are UTC. The time now is 03:00.

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, Jelsoft Enterprises Ltd.