Infinite loop in mpz_probab_prime_p

Darin Ohashi darin242 at hotmail.com
Fri Mar 4 15:25:51 CET 2005


>
>While there was too much dividing, it turns out that most of the
>time is spent in the Miller-Rabin code.  You need to be more
>patient, you are requesting a really huge computation.
>

Yeah, I did some more experimentation overnight and found that eventually it
did complete. As long as it is expected to take that long then it is not a 
bug.

I think I was thrown by the fact that I tried a bunch of similarly sized
integers and they all finished very quickly.  However I guess they all had
trival divisors.

Did you add a check for the potential overflow from ln2 = ln2*ln2 ?  For 
very very large integers
this could be a problem.

Thanks for the help

Darin

---
Because all other Floccinaucinihilipilification Homepages are worthless.
http://www.floccinaucinihilipilification.net The 
Floccinaucinihilipilification H
omepage




More information about the gmp-bugs mailing list