I've shuffled off some of the less-important calculations to another queue system. The factor-processing queue is now pleasantly fast (90k records processed in 5 minutes, about 300 results per second). Then the other slow calculations can run when they get around to it without slowing down the results queue.

[URL="https://www.mersenne.ca/exponent/180221"]Exponents around 180000[/URL] report 295.935 % chance of finding a factor in unsuccessful P-1 runs. Seems a little bit odd.

[QUOTE=Viliam Furik;554392][URL="https://www.mersenne.ca/exponent/180221"]Exponents around 180000[/URL] report 295.935 % chance of finding a factor in unsuccessful P-1 runs. Seems a little bit odd.[/QUOTE]Those are some seriously-big bounds for tiny exponents!

I'll pass the observation along to Mihai who developed the updated P-1 probability/bounds code. Obviously 295% probability of factor isn't right, especially since no factor was found!

Coincidentally, [URL="https://www.mersenne.ca/exponent/1800221"]1800221[/URL] is also prime (the previous link was to 180221, this one is to 1800221)...

It also displays very high probabilities, but most interesting is, that they are kind of the same as in 180221.

Ok, as I am writing this, I found out that [URL="https://www.mersenne.ca/exponent/18000221"]18000221[/URL] is also prime (180221, 1800221, 18000221). Also, it displays the same probability...

Same at all other exponents, which had P-1 done.


I think you may have broken something when doing P-1 updates :wink:

[QUOTE=Viliam Furik;554401]I think you may have broken something when doing P-1 updates :wink:[/QUOTE]Mostly a matter of being clueless. :cmd:

I consulted some people smarter than me and they :richd: and pointed me in the right direction.

Numbers should be sane(r) now. Let me know if they still* seem off.
(*after doing a Ctrl-F5 forcible refresh on any page you've loaded recently that shows wacky probabilities)

After some additional weirdness with importing results from the last two days, I [b]think[/b] everything should be running both correctly and reasonably fast. I'm some someone will point out where that's not the case :smile:

[QUOTE=James Heinrich;554509]After some additional weirdness with importing results from the last two days, I [b]think[/b] everything should be running both correctly and reasonably fast. I'm some someone will point out where that's not the case :smile:[/QUOTE]

AFAICT, most things are fine. I have a question about the "Factors missed by P-1 factoring." Are these only factors found by trial-factoring after a P-1 run, or should the lists also include factors found by subsequent P-1 or ECM searches?

[QUOTE=masser;554637]I have a question about the "[url=https://www.mersenne.ca/p1missed.php]Factors missed by P-1 factoring[/url]". Are these only factors found by trial-factoring after a P-1 run, or should the lists also include factors found by subsequent P-1 or ECM searches?[/QUOTE]It should list any factor (found by whichever means) on which a NF-PM1 result exists with bounds that should have found the factor. There are known to be old (pre ~2005?) versions of Prime95 that were PM1-buggy and missed factors so that's one probable reason (also the usual small chance of faulty hardware or faked results).

Hm. I mentioned a few recent finds (M14086463 and M14085161) [URL="https://mersenneforum.org/showthread.php?t=20047&page=6"]here[/URL] that I believe meet the criteria (factor missed by previous P-1 attempt), but they don't appear listed [URL="https://www.mersenne.ca/p1missed.php?s=x&o=d&p=0&min=14000000&max=14100000"]here[/URL]. Am I missing something?

[QUOTE=masser;554644]Am I missing something?[/QUOTE]No, but I [STRIKE]am[/STRIKE] was. :smile:

The code that generates that list was referencing some data that's no longer used/updated so it wasn't catching new entries. I have updated the code and your quoted exponents now show on the list (along with about 300 other missing ones).

[QUOTE=James Heinrich;554671](along with about 300 other missing ones).[/QUOTE]

Awesome - thanks!