Raising project money!!!!

[mastervince5]

Some of you may allready have heared about the RSA challenge. It's quite well-known. And you can earn money with it. When you read the following page carefully you will see that there are still a couple of nice prices available. And with what I've read on this page, it shouldn't be that difficult. Certainly not for a project like this.
If you want to raise money - just put some resources you have on this project.

http://mathworld.wolfram.com/news/2005-11-08/rsa-640/

Don't you think this is a good idea to get some money, money which could be used on different usefull thinks???

just an idea, greetz

[Greenbank]

The money spent by Franke et al far exceeded the amount they won.

The used 80 Opteron boxes for 4.5 months to do the lattice sieving for relations and solve the final matrix.

Even if they solved the next 3 RSA challenge numbers using the same hardware and software they still wouldn't make a profit. The cost of the hardware, electricity and air-conditioning is quite substantial.

This doesn't even take into account their time in setting up and managing both the software and hardware.

[vjs]

Like the last distributed attempt at RSA-640, they actually tried trial division :rotfl:

Sure it's possible to do a distributed attack at one of these numbers:


RSA-704 212 $30,000 open
RSA-768 232 $50,000 open
RSA-896 270 $75,000 open
RSA-1024 309 $100,000 open
RSA-1536 463 $150,000 open
RSA-2048 617 $200,000 open


RSA-704 or RSA-768 don't look outside the realm of possibility. RSA-1024 perhaps in a couple of years for DC.

What could be distributed would be the polyselection and line-sieve. This is basically what was done over the cluster. However the overhead in such a project :notworthy just isn't worth it, the people participating would really have to know what they are doing and the time invested in get them vs time donated wouldn't be worth it IMHO.

I'd suggest you post a comment like this at nfsnet.net, it's basically what they are doing anyways. Also check out oddperfect and elevensmooth.

At least for the RSA challanges I don't think you need to do any ecm work since they are two large primes of similar size.

I don't even think you could try this with ecm ECM anyways, the 120 digit level (RSA-786)would require bounds of what??? B1=~1T for stage1 and at least 1P for stage 2 for each curve. Even if you had enough memory for stage 2... stage 1 would take at least 10 years by my quick calculations.

[Mystwalker]

Originally posted by vjs
Like the last distributed attempt at RSA-640, they actually tried trial division :rotfl:

And they do not learn:


2005-11-08 14:40 GMT Project resumed with new application
We have resumed the work with the challenge. We are now taking RSA768 (yes, we skipped one. Everyone else is allowed to take it ;)).

:cry:

[smh]

Originally posted by mastervince5
And with what I've read on this page, it shouldn't be that difficult. Certainly not for a project like this.
If you want to raise money - just put some resources you have on this project.

I suggest you read a bit further. Factoring those numbers is not easy.

Like others said, sieving could be done by a couple of thousand users over the internet. It's the matrix step that requires some serious hardware (and software).


Originally posted by vjs
I don't even think you could try this with ecm ECM anyways, the 120 digit level (RSA-786)would require bounds of what??? B1=~1T for stage1 and at least 1P for stage 2 for each curve. Even if you had enough memory for stage 2... stage 1 would take at least 10 years by my quick calculations.

Anyone calculated if the factorization would be faster by trail division or ECM?

[Ken_g6[TA]]

Originally posted by smh
Anyone calculated if the factorization would be faster by trail division or ECM? Or, say, distributed Pollard Rho?

[vjs]

Probably RSA :rotfl:

Seriously, I'm not sure how ecm works on 80-digit numbers. I think the record is still 66-digit but that was sort of a fluke they were using 55-digit bounds. Not even knowing the number of curves it would require I still think trial division would be the slowest.

For RSA-704 one could probably start with 104-digit numbers saving 1% but you'd probably have to trial divide to 106-digit number before it was cracked. That's a huge space.

That's sort of the point of the challenges, not to crack the numbers (although you do win a prize). But the point for them, develop new and faster technique that could crack the number. You then have to show them how it's done so they can try to prevent this type of crack.

Personally if I had a technique I'd report the factors (make it public) but not disclose the technique. You could then sell the program idea to busniess or goverment for much much more than 30k.

Trial division doesn't even take into account that the composite is a multiple of two large primes of simliar size. One has to start with this assumption then add aditional tricks to even consider the possibility of success.

[royanee]

Originally posted by vjs
Personally if I had a technique I'd report the factors (make it public) but not disclose the technique. You could then sell the program idea to busniess or goverment for much much more than 30k.

The problem with that is if you crack the RSA algorithm, the government (NSA and the like) will come after you for being a threat to national security and then lock you away without trial until they know how you did it and the algorithm has been phased out in favor of a newer one that your attack will not work on.