Any news?

[hallstei]

This project is very interesting, and I have just started running the client.

But it seems like there has been a long time since the homepage was updated. Is there any news from the project?

I looked at the progress page and there seems like there are one or two composites that one would be able to factor using NFS, instead of continuing with ECM. Any thoughts?

[Mystwalker]

I also thought about GNFS on the C135 - and I can't remember whether or not I've already proposed it to William.

[smh]

Originally Posted by Mystwalker

I also thought about GNFS on the C135 - and I can't remember whether or not I've already proposed it to William.

I did propose it several months back but things never got started.

I can contribute several GHz weeks of sieving, but running the matrix (GGNFS) would be a bit hard with my current hardware.

[xilman]

Originally Posted by smh

I did propose it several months back but things never got started.

I can contribute several GHz weeks of sieving, but running the matrix (GGNFS) would be a bit hard with my current hardware.

Ok, let's do it.

The matrix should be straightforward on what I have available.


Paul

[xilman]

Originally Posted by Mystwalker

I also thought about GNFS on the C135 - and I can't remember whether or not I've already proposed it to William.

I'm missing something. The smallest composite on the web page seems to be C149 --- which is markedly harder than C135.

Someone (William?) please clarify the position and, preferably, post the integer.


Paul

[fetofs]

Originally Posted by xilman

I'm missing something. The smallest composite on the web page seems to be C149 --- which is markedly harder than C135.

Someone (William?) please clarify the position and, preferably, post the integer.


Paul

I can add that, in the "progress" section, it says that C135 is a factor of M1485 (not factored, ECM completed up to p50, 342 curves with B1=110M) Maybe it isn't updated?

EDIT: Oh, and I don't know the integer.

EDIT2:28300649180491855300912615076369109754 264440 134225 412394 813185 799179 827786 479258 863457 226971 724590 980107 599021 820634 948950 561954 778763 954559 736955 520419 203780 944269 761942 624015 867189 522424 175229 598413 857554 824712 310502 172783 420615 262572 946238 198024 536016 638302 542094 239541 373118 708237 453223 is a C290.

[fatphil]

Originally Posted by xilman

I'm missing something. The smallest composite on the web page seems to be C149 --- which is markedly harder than C135.

Someone (William?) please clarify the position and, preferably, post the integer.


Paul

Last bit of news on that number on the website was the creation of the C135:
http://home.earthlink.net/~elevensmo...enFactors.html
M(1485) 3307778274605860019775884877775856144362368451681 49 Mark Rodenkirch 2004-06-11 ECM GMP-ECM5 C184 11M


And the candidate list updated in july contains it still:
http://home.earthlink.net/~elevensmooth/Progress.html
M(1485) C135 done done done done done done done 342

However, july is a long time back...

[fetofs]

As my edit time is over, I had to post again:

795 535258 017305 167145 121871 550373 005335 397768 371315 887482
970506 034490 852226 395158 613855 712286 911706 841452 612879 325276 931891 140828 826431 (C135)

[xilman]

Originally Posted by fatphil

Last bit of news on that number on the website was the creation of the C135:
http://home.earthlink.net/~elevensmo...enFactors.html
M(1485) 3307778274605860019775884877775856144362368451681 49 Mark Rodenkirch 2004-06-11 ECM GMP-ECM5 C184 11M


And the candidate list updated in july contains it still:
http://home.earthlink.net/~elevensmooth/Progress.html
M(1485) C135 done done done done done done done 342

However, july is a long time back...

Thanks, and to fetofs.

Silly me. I thought the "composites" file on the 11-smooth website would contain the composites.

Sander: shall we go for it?

Paul

[fatphil]

Originally Posted by xilman

Sander: shall we go for it?

Is there a DC way others can contribute to such GNFS-ing, or is bringing it home just a job for a few big guys?

[xilman]

Originally Posted by fatphil

Is there a DC way others can contribute to such GNFS-ing, or is bringing it home just a job for a few big guys?

In principle, yes.

Searching for polynomials is trivially parallelizable, as is sieving for relations. The post processing stages are not so easily done. It's rarely worth parallelizing the polynomial search because it takes relatively little computation

We could use any of several different approaches for sieving. The traditional one is entirely human driven. Someone splits up the ranges of special-q and doles out blocks to other humans who are responsible for sieving them by some means they choose for themselves. That is, the subcontractors are free to use one big machine, to dole out tasks to numerous machines, or to employ subcontractors of their own. Many of the landmark factorizations were done this way, but it is rather labour intensive.

The other extreme is exemplified by NFSNET in which virtually all the client labour is automated, though the admin labour is still significant. We could, in principle, set up a parallel NFSNET project. However, the NFSNET admins, and especially Richard, are very overworked at the moment.

There is a middle-way. I wrote an extremely simple client/server harness for the sievers. Much simpler than NFSNET but noticeably less scalable, less robust and more work for the people in charge of the sieving machines. It's most useful for people with a bunch of machines under their immediate control though it could be used amongst a wider spread of systems. The lack of robustness is its major deficiency.

Paul