Anyone factoring <5M?

[WVU Mersenneer]

Originally Posted by markr

Indeed, prior to version 5 of the server, about three years ago, GIMPS only kept one factor for each exponent in its database. (Of course other folk tried to keep track of all factors.)

So that explains why I have been able to find factors < 2^60 via ECM on exponents, such as 1761695163450679, which is between 2^50 and 2^51, for M84299. Interestingly, I tried TF against M84299 starting at 2^51 to see if there are any other factors, and it found one under 2^52, but it was the product of M84299's smaller factors. Continuing TF from 2^53, nothing found by 2^54, but the server returns TF is not needed. I guess it isn't since factors are already known, but I'd think it would be glad someone's trying to "solve" the exponent.

Does TF run a PRP on the remaining cofactor like ECM does, or does that only occur with ECM?

Is there an undocumented command similar to ContinueECM which allows factoring to continue even after a factor is found? Maybe that's what lorgix was looking for.

[lorgix]

Yes, I would like information about factorization attempts to be stored even for numbers with one or more known factors.

I suppose the minimum would be saving the TF level.

[TimSorbet]

Originally Posted by WVU Mersenneer

Does TF run a PRP on the remaining cofactor like ECM does, or does that only occur with ECM?

Only with ECM. Otherwise consider what would happen when you find a factor in the course of a normal first-time test, or TFing: it would want to run a PRP on the cofactor, which would take even longer than just running the LL on the original number.

[wblipp]

Originally Posted by WVU Mersenneer

So that explains why I have been able to find factors < 2^60 via ECM on exponents, such as 1761695163450679, which is between 2^50 and 2^51, for M84299.

I hope you keep Will Edgington informed of factors, too. Many places keep track of factors of various subsets of Mersenne numbers that intersect their own interests, but Will keeps ALL Mersenne factors and makes them available for download from his page. Telling Will is the way to make it most likely that other people will be able to find your factors without duplicating your effort.

http://www.garlic.com/~wedgingt/mersenne.html

Just send him an email that includes "factors" as part of the subject line. Batching the inputs is a kindness to reduce his workload.

William

[petrw1]

All exponents above 500,000 now TF'd to 61 bits.

Oh and monst, George and chalsall are making short work of the last few (13 more) very low exponents <60 bits.

[lycorn]

It has gone down to 9.
It surprises me how fast the number is diminishing. Are people finding factors, or just increasing the TF level? Either way, I would expect the task to take a long time, in view of the size of the exponents and the amount of ECM already done. But apparently, it is going pretty well...

[petrw1]

Originally Posted by lycorn

Are people finding factors, or just increasing the TF level?

The latter

Either way, I would expect the task to take a long time

Check here: http://mersenne-aries.sili.net/credit.php

[lycorn]

Thanks for the link.
Taking these low exponents from 58 to 60 bits takes much less time than I was anticipating. That explains the progress observed.

[gjmccrac]

Carsten Kossendey - could you please direct your GPU to larger exponents.

I have a couple of old CPU's that works well in the < 64 bit range.

You recently completed exponents in the 3,420,*** that I had reserved.

I had to spend a bunch of time cleaning up reservations because of the exponents you just processed.

Thanks,
Grant.

[gjmccrac]

Carsten - please ensure that you reserve exponents you are going to TF in the 3M range. You are TF'ing exponents that are already reserved.

Grant.

[ckdo]

Final exponent will finish in 15 minutes. Forces redirected towards 8.2M, going down.