Factoring assignment
 

Hi,
I've been given a factoring assignment. In the worktodo file it says
Factor=20313121,64
on the account report page it also says fact bits=64, but the client
says factoring to 66 bits (using the latest v.22, downloaded yesterday).
What does this mean?
Also, most other exponents in that range have fact bits=59

Norbert

Hi norbert,

As far as I know, your exponent has been factored to 64 bits by previous runs, or factors found gave that order of precision; anyway it should be factored up to 66, and this is where you come to play.

Other exponents have been previously factored to 59, and this is why they show "59" on the list.

Note that on "undoc.txt" file there is a setting that permits you to factor up to 72 and more bits.

Finally, what you see on your user statistics (and on "worktodo.ini") is updated as long as you proceed on factoring.

Hope this helps.
forgive my poor English... :surprised:ops:

Luigi

Luigi,
Thanks for the answer. BTW, I can't find any poor English in your post ...

Norbert

does running it at 77 bits run fastor

or does it mean it just puts it threw more test?

[quote="crash893"]does running it at 77 bits run fastor

or does it mean it just puts it threw more test?[/quote]

Factoring to 77 bits would take quite a while! 73 bits will take twice as long as 72 bits. 74 bits will take twice as long as 73 bits. And so on.

what is the advantage of 77 then

does it do a better test?

WELL, if you factor to 77 bits you are more likeky to find a factor. But it is not worth the extra time it takes as it would be faster to do a Lucas-Lehmer test thaan to factor to 77 bits and even then perhaps not find a factor.

would it be a better option in the future

i saw some one say that a 64 bit cpu would factor lots quicker

The program's factoring limits take into consideration the relative time it takes to factor to a certain point as opposed to the time it takes to do an LL test. If factoring time speeds up then that will be taken into consideration on future versions of the program.

The way Prime95 works is it tries to see what are the chances of finding a factor vs time taken as opposed to doing an LL test. Usually, for a larger exponent since the LL test takes longer it's a good idea to factor more. Look at http://mersenne.org/math.htm to see what are the current factoring bounds. If you factor any more than what that page says - and what is preset in the Prime95 client - you are wasting your time as the increased chance of finding a factor doesn't justify the extra time it takes.

Of course, with a 64 bit processor if you can show that it really reduces the time to factor significantly, those bounds can change. Also, there are people out there who just want to find factors no matter how big and how long it takes. If you are one of those, go ahead by all means but it won't "benefit" the project as much. Or there is "better" use for your CPU time if you let the client decide how much to factor.