[QUOTE=Rodrigo;231843]
I will run the exponent a second time -- it'll take 11 days to get the result again.
[/QUOTE]
You should be able to just run it for the 68 to 69 bit range


[QUOTE=Rodrigo;231843]
Anybody reading this who has experience with this sort of thing -- feel free to pitch in! Besides re-running the test, what do we do?
[/QUOTE]

I don't know what to suggest. Perhaps Luigi would have some suggestions.

Grant.

[QUOTE=gjmccrac;231845]You should be able to just run it for the 68 to 69 bit range.[/QUOTE]I put the factor into windows calculator, flipped it to bin and I get 63 bits.

[QUOTE=Uncwilly;231847]I put the factor into windows calculator, flipped it to bin and I get 63 bits.[/QUOTE]
log(378121774679999658041)/log(2) = 68.357ish. Wouldn't that put it in the 68 to 69 bit range?

[QUOTE=gjmccrac;231845]You should be able to just run it for the 68 to 69 bit range

Grant.[/QUOTE]
Grant,

As you suggested, I'm running the same number from 68 to 69 bits tonight. I'll report the result as soon as I see it.

Why just from 68 to 69?

Rodrigo

[QUOTE=Rodrigo;231860]Grant,

As you suggested, I'm running the same number from 68 to 69 bits tonight. I'll report the result as soon as I see it.

Why just from 68 to 69?

Rodrigo[/QUOTE]
See the post directly above yours. (#487)

[QUOTE=Rodrigo;231843]gjmccrac,

This is disturbing and important, thanks for letting me know.

I'm using factor4, as my understanding is that that's the most suitable version of the software for that PC (Windows 98 on a Pentium II).

I will run the exponent a second time -- it'll take 11 days to get the result again.

Now, the thing is: How often does something like this happen? And, regardless of what the new result is, can we really trust its output anymore?

Anybody reading this who has experience with this sort of thing -- feel free to pitch in! Besides re-running the test, what do we do?

Rodrigo[/QUOTE]

Houston, we have a problem...

Factor4 and Factor5 use completely different data structures; while Factor5 seems to work correctly, it seems that you ran into a bug on Factor4...

I'm afraid I can't patch it right now, as I'm leaving tonight for vacation (holidays, as people in UK say). Factor4 has not been updated since Factor5 was out, and the only way to make Factor5 run on Pentium and Pentium II is to rebuild it on those platforms (at least, that's what I'm aware of, if anybody around here can tell me how to trick the GMP configuration tool to make it think I have a plain vanilla Pentium, I'll be more than happy to try).

Luigi

Luigi,

Great to hear from you!

If you can fix Factor4 or get Factor5 to work on a Pentium II -- whichever way is less work and/or takes care of the problem will be good.

Should I re-run the other exponents I've done, on a different machine? I have an XP with a Pentium 4 that I could bring over from LL-D work so that it could verify my previous reports.

Enjoy your vacation! We'll see you when you come back.

Rodrigo

[QUOTE=KingKurly;231861]See the post directly above yours. (#487)[/QUOTE]
KingKurly,

Ah, so that means that's where the problem lies. Interesting, thanks! It helps to read it in the morning rather than late at night...

Rodrigo

[QUOTE=gjmccrac;231781]It actually found a factor in this range.
[CODE]M3321934709 has a factor: 378121774679999658041[/CODE]

I then double checked it with factor5.

The win32 version of factor5 also finds it.
[CODE]M3321934709 has a factor: 378121774679999658041 - Program: L5.0x[/CODE][/QUOTE]Factor4p finds it, on my Core2Duo laptop under vista. Will try other versions later.

[QUOTE=Uncwilly;231994]Factor4p finds it, on my Core2Duo laptop under vista. Will try other versions later.[/QUOTE]
Uncwilly,

I'll report the 68-to-69 result as soon as I get it. But whether or not the computer finds the factor this time -- what (if anything) does that imply for the reliability of that PC's output? I doubt that (say) Factor4 would work fine on your computer and make a mistake on mine, which leads me to think the problem would be in the PC itself and not in the software. (Unless Factor4p is a different thing from Factor4.)

FWIW, before signing up for OBD I ran the Prime95 stress test on that PC for 30 hours, and it passed without a hitch.

Rodrigo

[QUOTE=Rodrigo;232007](Unless Factor4p is a different thing from Factor4.)[/QUOTE]It is SSE2 enabled. I plan on running Factor3 and 4 on this to check for program problems. Factor4p had different dll's in the package than plain 4.

Edit: Plain Factor4 with it's version of the .dll's finds it. On to Factor3 (much slower and no partial bit depths.)