Range of Prime95?
 

Can Prime95 test a 215million digit number for primality?

[quote=iNSiPiD1;211539]Can Prime95 test a 215million digit number for primality?[/quote]

You can manually add Mersenne numbers up to 990M (~ 300 million digits). The problem is that the time will very long (several years).

[QUOTE=iNSiPiD1;211539]Can Prime95 test a 215million digit number for primality?[/QUOTE]

Theoretically, maybe.
According to [url]http://www.mersenne.org/primenet/[/url]
there are currently 2 LL tests with exponents in the 750M range --- about 233M digits; through the years there have been many more at your size or bigger but none have ever finished.

Practically ... NOT likely.
The current supported maximum range for Prime95 is 596M or about 185M digits. Maybe others can explain how those larger assignments exist...could be they are using a program other than Prime95.

Our current BEST theoretical benchmark would have an Intel i7-980x running all 6 cores NON-stop 2.5 years to test 596000000 ... I could only guess that it may take twice as long or more to complete your proposed test.

No, Prime95 can't run a number that big. It won't run LL on anything above 2^596M (179.4 million digits), which is probably the largest number that will work with the largest supported FFT size: 32M. As others have said, other programs might support larger sizes, but in any case, numbers that big would take years (at least without a supercomputer).
Is there a particular 215M digit number you're considering? What drew you to this number? Have you done any TF on this number? Prime95 or [URL="http://www.mersenneforum.org/showthread.php?t=7283"]Factor5[/URL] (which supports multithreading) should work for that. That would be an easy way to prove it composite (if it is).

[QUOTE=petrw1;211546]
According to [url]http://www.mersenne.org/primenet/[/url]
there are currently 2 LL tests with exponents in the 750M range --- about 233M digits; through the years there have been many more at your size or bigger but none have ever finished.[/QUOTE]

That was me. I was manually trying to get some TFs in that range and clicked the wrong box or something. Then I lost the ID code to cancel them(blush). I thought they would time out when no progress was reported on them but its been nearly 6 months now without any activity, not even check ins and they are still there!

[quote=Mini-Geek;211548]No, Prime95 can't run a number that big. It won't run LL on anything above 2^596M (179.4 million digits), which is probably the largest number that will work with the largest supported FFT size: 32M. As others have said, other programs might support larger sizes, but in any case, numbers that big would take years (at least without a supercomputer).
Is there a particular 215M digit number you're considering? What drew you to this number? Have you done any TF on this number? Prime95 or [URL="http://www.mersenneforum.org/showthread.php?t=7283"]Factor5[/URL] (which supports multithreading) should work for that. That would be an easy way to prove it composite (if it is).[/quote]

Well, the story that drew me to this particular number is laughable. I only wish to test the number to satisfy my own curiosity. I have not done a TF on it, but maybe I will sometime.

I suspect that I will have to wait for the computers to get faster, or for a faster algorithm to be found... or something like that. Or start now and wait a few years...

[quote=lfm;211549]That was me. I was manually trying to get some TFs in that range and clicked the wrong box or something. Then I lost the ID code to cancel them(blush). I thought they would time out when no progress was reported on them but its been nearly 6 months now without any activity, not even check ins and they are still there![/quote]
(I was going to suggest logging in and unreserving them, but I just checked and I see they're reserved as ANONYMOUS)
If you have anything to show/suggest that it was you (or maybe even if you don't), you could email George to get it squared away: :woltman:
Or, I think 6 months is the time limit, so soon they should expire on their own.

[QUOTE=Mini-Geek;211551]I don't suppose you checked them out under an account..? If so, you can unreserve them from [URL]http://www.mersenne.org/workload/[/URL]
[/QUOTE]

ya it was anonymous too. I think I was sleep deprived or something at the time. About the only evidence I have that it was me is to note the surrounding TF results are mostly mine. Evidence, not proof of course.

It seems odd to me that this isn't automated yet. I thought there was some discussion about stuff over 90 days without a check-in would be released for re-assignment. Maybe only applies to the more active ranges or something?

i don't think it applies to manual assignments.[quote]I don't remember the LMH policy. It might well be that assignments are
cleared after a year.

Regards,
george[/quote]

[QUOTE=iNSiPiD1;211550]Well, the story that drew me to this particular number is laughable. I only wish to test the number to satisfy my own curiosity. I have not done a TF on it, but maybe I will sometime.

I suspect that I will have to wait for the computers to get faster, or for a faster algorithm to be found... or something like that. Or start now and wait a few years...[/QUOTE]

I can run some trial-factoring for you if you don't mind, just send a PM with the exponent to me :smile:

Luigi

[quote=Mini-Geek;211548]No, Prime95 can't run a number that big. It won't run LL on anything above 2^596M (179.4 million digits), which is probably the largest number that will work with the largest supported FFT size: 32M. As others have said, other programs might support larger sizes, but in any case, numbers that big would take years (at least without a supercomputer).
Is there a particular 215M digit number you're considering? What drew you to this number? Have you done any TF on this number? Prime95 or [URL="http://www.mersenneforum.org/showthread.php?t=7283"]Factor5[/URL] (which supports multithreading) should work for that. That would be an easy way to prove it composite (if it is).[/quote]

I've got one in that number intervall. With one core it takes around 30 years. You can decrease the time significantly. With an 8- or 12-core system you will get down to 5 - 7 years and with Moore's Law you will have finished around 4 years.

[quote=CADavis;211572]i don't think it applies to manual assignments.[/quote]

George explicitly said that policy won't applied to 100M+ numbers ...