Hm, I have a query, I happened to check on the progress of factor5 as it was passing 50%. But then I noticed as it proceeded to ... bounce off. And it had done it once before too. Third times the charm I guess because next time it got through.

I've uploaded a screenshot and a status.txt file. Don't suppose you know what's happening here?

[QUOTE=lavalamp;172782]Hm, I have a query, I happened to check on the progress of factor5 as it was passing 50%. But then I noticed as it proceeded to ... bounce off. And it had done it once before too. Third times the charm I guess because next time it got through.

I've uploaded a screenshot and a status.txt file. Don't suppose you know what's happening here?[/QUOTE]

I think I know what happened. I had similar issues when testing my code on quad processors not owning one. Nothing affecting the final results, anyway.

The routine implementing the percentage done crawls over 8 threads and makes assumptions about the progress of calculation. It seems that version 5.02 in pre-alpha testing will need some more tweaking.
Thanks for pointing it to me :smile: .

Luigi

Ah I see. Well apparently it's not just 50% that it bounces off of, it's just done it again higher up:

51.078 %
50.989 %
51.002 %

Would it do this each time it finishes one of those groups of 4 million if one or two threads are lagging behind?

[QUOTE=lavalamp;172789]Ah I see. Well apparently it's not just 50% that it bounces off of, it's just done it again higher up:

51.078 %
50.989 %
51.002 %

Would it do this each time it finishes one of those groups of 4 million if one or two threads are lagging behind?[/QUOTE]

No, it should behave that way only around 50% (16 classes/8 threads).

Luigi

It's over 76.667% now so I've passed through 76 bits for 3321933661.

Edit: Typo.

3321933661 taken to 77 bits, no factors.

Taking 3321933577 to 77 bits, eta is early tuesday morning.

2^3321933893 - 1 has a factor:

36285087986156170392041

74.94 bits ish

I kinda wish I'd done 74 - 75 then 75 - 76 instead of running a straight 74 - 76, 'cause right now it's 50% done and my built-in perfect solution fallacy won't let me stop it mid way.

Fun fact, this factor actually fell out 2.6 days ago but I didn't notice.

Edit: Oh no, I've turned into one of those people that triple posts.

:wblipp:

Congratulations!

Please triple post more often if the triple posts are going to contain factors!

William

[QUOTE=lavalamp;173004]2^3321933893 - 1 has a factor:

36285087986156170392041

74.94 bits ish

I kinda wish I'd done 74 - 75 then 75 - 76 instead of running a straight 74 - 76, 'cause right now it's 50% done and my built-in perfect solution fallacy won't let me stop it mid way.

Fun fact, this factor actually fell out 2.6 days ago but I didn't notice.

Edit: Oh no, I've turned into one of those people that triple posts.[/QUOTE]

Nice catch! Maybe you'll find out a new, fatter factor while finishing your range... :grin:

Did you send the factor to George and Will Edgington? Do you prefer I will?


Luigi

No, I just posted it here. I'll send George a PM now, but am I right in thinking that Will Edgington is wblipp?

Incidentally, does anyone know how deeply MM127 has been trial factored? I can't find much information about it but did read from a very old newsgroup post that Will Edgington kept a record of trial factor depths for it, though I couldn't find it.

I wrote some very basic PHP code using the bcmath functions to trial factor it to 145 bits which took absolutely ages to run (9 mins). Does anyone know of a (much) more optimised utility that can handle trial factoring of a mersenne with such a large exponant?

[quote=lavalamp;173033]No, I just posted it here. I'll send George a PM now, but am I right in thinking that Will Edgington is wblipp?

Incidentally, does anyone know how deeply MM127 has been trial factored? I can't find much information about it but did read from a very old newsgroup post that Will Edgington kept a record of trial factor depths for it, though I couldn't find it.[/quote]

No, Will Edgington AFAIK does not use these forums. His email address is wedgingt AT acm DOT org[COLOR=black].[/COLOR]

MM127 has been trial factored to at least at least 176 bits by Ernst Mayer.