mpz_probab_prime_p(n, reps) : how long?

[Jonathan A. Zylstra]

jojo 29 wrote:

> The bench page "http://www.mersenne.org/bench.htm" is measured in P2K 
> years (P4 at 3.4 GHz, Celeron 2500, Athlon XP 2200+, etc.)... But, in 
> your mind, the function mpz_probab_prime_p(n, reps) needs few months 
> to prove the primality of a 33 million digits number (with 10 
> iterations)?
> How do you estimate this?
>
> Regards,
>
> jojo

Also, please note that the function mpz_probab_prime_p does not actually 
prove any number to be prime. All it does tell you is that n is most 
likely prime. (hence 'probab_prime') So, after your wait - whether 
months or years, you will still need to run another test to actually 
prove n is prime.

Jonathan Zylstra