--- In

primenumbers@yahoogroups.com, "julienbenney" <jpbenney@...>

wrote:

>

> Re: "A PRP test of a 2M digit number consists of several million

multiplies. Even

> testing 100 multiplies is sufficient to work out how long 2M will

take. The variation

> between runs is normally minimal."

>

> I have had this problem: I have been trying to find the previous

prime to 1000

> factorial - which as you would probably know is a number of 2568

digits - using

> Dario Alpern's site "http://www.alpertron.com.ar/ECM.HTM". Whilst I

was aware that

> site would take some time to find the number, it gave so few clues

as to how long the

> number would take to find that after about forty minutes with no

hint as to when the

> number would be found I just stopped and gave up.

>

> Whilst I am well aware that there is much better software for doing

the job, I wonder

> if any of you have done that or a very closely related experiment

and what you

> findings are?!

>

Julien,

The expression evaluator I programmed for the applet is not very

smart and it computes SPRPs for all odd numbers before or after the

argument. If sieving is added to the process, it could be about 10

times faster.

Best regards,

Dario Alpern

http://www.alpertron.com.ar/ENGLISH.HTM