Anyone knows about primegrid

I see the site http://primegrid.com/ . In this site they run project like NFSNET for factoring a RSA challenge number.But they are not mentinoning algorithm they are used.I hope they are using GGNFS implementations.
Anyone knows about it?

[R.D. Silverman]

Originally Posted by koders333

I see the site http://primegrid.com/ . In this site they run project like NFSNET for factoring a RSA challenge number.But they are not mentinoning algorithm they are used.I hope they are using GGNFS implementations.
Anyone knows about it?

They are using..... (drum roll please!)

TRIAL DIVISION.

[ColdFury]

The Neo Project part 2.

I'll check back on their progress in a couple billion years.

[garo]

Not just <drum roll> trial division </drum roll> but
<cymbals>
trial division by all odd numbers
</cymbals>

[jasonp]

Originally Posted by ColdFury

The Neo Project part 2.

I'll check back on their progress in a couple billion years.

I've been wondering what this project could do better. It basically is run by one guy, who has zero time and whose primary interest is building a BOINC scheduler in perl. Going after RSA numbers is just a hook to lure people.

So, the obvious question: if you need a bunch of volunteers to test your code, and you have zero time to come up with something, and the something has to be interesting, what should you do?

jasonp

PS: NFS happens not to be an option with PrimeGrid, because of the data storage needs. ECM is not an option either, because the ECM that people care about requires even more memory in the client than an NFS siever.

[Jeff Gilchrist]

Originally Posted by garo

trial division by all odd numbers

Hey, that increases their search speed by cutting out half of the numbers to try right????

[ewmayer]

Originally Posted by jasonp

I've been wondering what this project could do better. It basically is run by one guy, who has zero time and whose primary interest is building a BOINC scheduler in perl. Going after RSA numbers is just a hook to lure people.

So, the obvious question: if you need a bunch of volunteers to test your code, and you have zero time to come up with something, and the something has to be interesting, what should you do?

How about trial division of Fermat or iterated Mersenne numbers with exponents > 30 bits or so? That way, even if their trial-division code sucks they at least have >= (a snowball's chance in hell) of finding a previously unknown factor. Tony Forbes' MFAC code isn't all that fast (though probably an order of magnitude faster than the app in question here), but still managed to find the first known factor of F31. With RSA challenge numbers they have what is for all intents and purposes zero chance of success. Why doom oneself from the start? The only plausible answer I can come up with is ... because one is an idiot.

[ColdFury]

Originally Posted by jasonp

I've been wondering what this project could do better. It basically is run by one guy, who has zero time and whose primary interest is building a BOINC scheduler in perl. Going after RSA numbers is just a hook to lure people.

So, the obvious question: if you need a bunch of volunteers to test your code, and you have zero time to come up with something, and the something has to be interesting, what should you do?

jasonp

PS: NFS happens not to be an option with PrimeGrid, because of the data storage needs. ECM is not an option either, because the ECM that people care about requires even more memory in the client than an NFS siever.

I've heard the Dutch Power Cows are working at starting another ECCP project and are looking for developers. Perhaps he could offer his services to them?

[wblipp]

Originally Posted by jasonp

ECM is not an option either, because the ECM that people care about requires even more memory in the client than an NFS siever.

He could easily generate a huge amount of small ECM work by deciding to work on Brent's factor list for bases larger than 1000. But then he would generate lots of factors, and have to spend time managing the results, another distraction for a manager already short on time. His present project will never overload him with results.