[quote=Andi47;167577]Today I got a squfof error with yafu 1.08:

[code]error on N = 0x6279c2a5af1
[/code]

[/quote]

Thanks for the info. The error is mostly harmless :) Meaning that everything should (and did) continue to work. The error and file are generated if the squfof routine detects an infinite loop. I've found and fixed one bug already in the routine due to info mklasson provided from errors he saw. Apparently there is another...

thanks,
- ben.

[quote=Andi47;167577]Today I got a squfof error with yafu 1.08:

[code]error on N = 0x6279c2a5af1
[/code]
[/quote]

That is in hex, the "0x" is the hex indicator, and the rest is the number (6767193512689). squfof probably went infinite-loop because the number is the square of a prime (2601383).

[quote=10metreh;167585]That is in hex, the "0x" is the hex indicator, and the rest is the number (6767193512689). squfof probably went infinite-loop because the number is the square of a prime (2601383).[/quote]

Well, it works fine if I try it using squfof from the command line. And squfof checks for perfect squares, so I don't think that's the problem. This one's still in the unsolved bucket for now... but doesn't appear to be causing major problems and is fairly rare so I'm going to put it on the shelf for now.

If anyone sees more of these errors though feel free to let me know about them.

- ben.

Some more squfof errors with 1.08:

[code]error on N = 0x3cde25fe1a11
error on N = 0x142216efaaa01
error on N = 0x880e3d8d1b69
[/code]

[quote=Andi47;167632]Some more squfof errors with 1.08:

[code]error on N = 0x3cde25fe1a11
error on N = 0x142216efaaa01
error on N = 0x880e3d8d1b69
[/code][/quote]

All squares of primes.

Some more squfof errors - they seem to be quite common.

[code]error on N = 0x39e3998f411
error on N = 0x3179fd0eb99
error on N = 0x136056b72401
error on N = 0x68544904261
error on N = 0x334baf81c09
error on N = 0x61cacf4a10e9
[/code]

[quote=Andi47;167876]Some more squfof errors - they seem to be quite common.

[code]error on N = 0x39e3998f411
error on N = 0x3179fd0eb99
error on N = 0x136056b72401
error on N = 0x68544904261
error on N = 0x334baf81c09
error on N = 0x61cacf4a10e9
[/code][/quote]

Do the primes they are the squares of have anything in common?

[QUOTE=10metreh;167879]Do the primes they are the squares of have anything in common?[/QUOTE]

Maybe the routine which should catch the squares of primes is bugged?

[quote=Andi47;167883]Maybe the routine which should catch the squares of primes is bugged?[/quote]

The odd thing is that it works when you run squfof manually (squfof()).

[quote=10metreh;167884]The odd thing is that it works when you run squfof manually (squfof()).[/quote]

Yeah, which is what is making this bug exceedingly hard to find.

@andi47: There is a slight oversight in the code in that perfect squareness is only detected when the squfof multiplier is one. The routine tries several other multipliers first, and one last, so maybe something happens before the squareness can be checked. But everything works when I try these test cases standalone...

If you are just seeing these few over the course of several days, then the occurance of the bug is still extremely rare (a few out of perhaps a hundred million cases), and still seems to be harmless.

I have seen yafu crashing a few times when factoring composites in the c40 ... c50 range. One of these was:

[code]04/10/09 17:32:55 v1.08 @ ANDREAS, Processing expression: siqs(58219450013341002773353091641937404175104466027)

04/10/09 17:32:57 v1.08 @ ANDREAS, Initializing with Tom's Fast Math (x86-32 asm)...
04/10/09 17:32:57 v1.08 @ ANDREAS, cached 78504 primes. pmax = 1000099
04/10/09 17:32:57 v1.08 @ ANDREAS, detected cpu 4, with L1 = 16384 bytes, L2 = 2097152 bytes
04/10/09 17:32:57 v1.08 @ ANDREAS, New random seeds: 698544311, 4251155528
*crash*
[/code]