Operation: Billion Digits

[Uncwilly]

M3321928483 no factor from 2^62 to 2^64.

Just pushing back the darkness a little...

[ixfd64]

Where's Ewmayer when we need him? :D

[Uncwilly]

Isn't Operation Billion Digits (OBD) at level 3? i.e. 3 numbers upto 69 bits?

Looks like:
3321928171
3321928241
3321928307
are all at or above 69.

3321928307 is above 69.7 at the very moment. I hope to be able to report it in @70bits before I am away from the machine for a few days.

Are any of us foolish folk up for another challenge? I would like to make a multipronged assualt on M(971). I am thinking of looking at lowM and:
having multiple machines working on different levels of Tfactoring (ET_ can your factor3_1 take decimal bits as input?, like 23.45),
get someone with a machine with around 4GB to P-1 it deep (Maybe P-1 Dave),
keep a sustained ECM attack (if I understand ECM, one can start at a curve of any number they choose with out having to also run all previous) (we could check out ranges and return them when done),
and get a single machine working on a "top down" approach.

The "top down" idea is the following:
Assume that M(971) is the product of only 2 very large primes (100+ decimal digits). Take an educated guess at the square root of M971, 1.41274212421x10^141 for example. Test the number for primality (if it is too hard, at least check it against the first 1000 primes and insure that it meets the criteria for a factor of a mersenne), if it isn't move up to the next number that passes this test. Divide M971 by the number. Check the result.
If the remainder is 0, We Win! If not, note the divisor and quotient and mark them off the list of suspects (even though there is a remainder). Find their mean and start the process again. Continue until the root is found and if it is an interger. If so, WE Win!

If that does not find it, have the original machine working on it work the area around the square root. Use presumed primes (it may be worth it put effort into checking them a goodly amount). Have a second machine join in the hunt, but starting in on the range of a cube root. Divide twice and check result, take mean, start over. When that machine finds the cube not to be whole, start it hunting the area for 3 prime factors. Then start a third machine (maybe pull one from the TF work), to start in on the 5th root. My best understanding from the ECM work is that the 7th root might have been taken care of.

I know this sounds like a wierd way of doing it, but until the NetSieve folks take up M971, this might be the way for independant machines to best attack it.

[wblipp]

Originally Posted by Uncwilly

Isn't Operation Billion Digits (OBD) at level 3? i.e. 3 numbers upto 69 bits?

Looks like:
3321928171
3321928241
3321928307
are all at or above 69.

I missed the '171. Luigi shows it completed on the Italian language site but didn't announce it here.

Today my '219 passed 69 digits on it's way to 70. That puts us at Level 3.2 (Level 3 (3 at 69 bits) plus two steps towards level 4 (a fourth at 69 and one at 70)).

I also took the next four unclaimed numbers to 66 bits.
M3321928381 no factor from 2^62 to 2^65.
M3321928483 no factor from 2^62 to 2^65.
M3321928601 no factor from 2^62 to 2^66.
M3321928381 no factor from 2^65 to 2^66.
M3321928483 no factor from 2^65 to 2^66.
M3321928697 no factor from 2^62 to 2^66.

http://ElevenSmooth.com/Billion.html
http://www.gimps.it/billion/billion.htm

William

[HiddenWarrior]

Originally Posted by Uncwilly

I would like to make a multipronged assualt on M(971).

I saw somewhere in internet the project 'Random-971'. People are searching for factor of M(971) by checking randomly generated factors. Sorry, but the link is lost for now (or the project doesn't work?).

[nitro]

Originally Posted by nitro

Neither is there one to 68 bits

M3321928373 no factor from 2^67 to 2^68.

M3321928373 no factor from 2^68 to 2^69.


Now at 69 bits, no factor, switching back to M3321928319

[geoff]

Originally Posted by Uncwilly

Are any of us foolish folk up for another challenge? I would like to make a multipronged assualt on M(971).

According to http://www.mersenne.org/ecmm.htm it should still be worth doing more ECM curves at the 50 digit level.

One particularly efficient way to tackle this if a few people are working together is for those with P4s or P4-Celerons to do the stage one ECM using Prime95/mprime, save the results by putting the GmpEcmHook=1 option in prime.ini, then those with Athlons or P3s to do a deep stage two using gmp-ecm resuming from the saved file.

This would involve emailing the results.txt file from stage one, but it should compress fairly well since each curve has some repeated data.

[ET_]

Originally Posted by wblipp

I missed the '171. Luigi shows it completed on the Italian language site but didn't announce it here.

Today my '219 passed 69 digits on it's way to 70. That puts us at Level 3.2 (Level 3 (3 at 69 bits) plus two steps towards level 4 (a fourth at 69 and one at 70)).

I'm taking '171 to 70... :-)

Luigi

[nitro]

Originally Posted by nitro

M3321928373 no factor from 2^68 to 2^69.


Now at 69 bits, no factor, switching back to M3321928319

M3321928319 no factor from 2^62 to 2^66.

[ET_]

No more volunteers?

Luigi

[PrimeCruncher]

Originally Posted by ET_

No more volunteers?

Luigi

I think everyone's a bit more concerned with this at the moment as opposed to such a 'whimsical project'. Speaking of which, there's only one 0.01M range left; I don't suppose someone here would be willing to donate a little computing muscle to finish that up?