5 questions

[OmbooHankvald]

Hi, I have a couple of questions (yes, I'm trying to add this ECM stuff to my guidecollection)

1) If I pick some numbers from the file archive , how is the easiest way to find out what I should make as B1? I know that this place has a table, but I believe the numbers have far more digits.

2) ECMNET. Is the B1=3000000 still the preferred?

3) Still ECMNET. At http://www.loria.fr/~zimmerma/ecmnt/ , what does the "-20" in the example stand for?

4) Cunningham. What are the Cunningham Tables? I thought people should just run ECM on the list to search for random factors, so what's the deal with the tables?

5) I still don't know how to compile ecmclient, but I'm already asking here . If you can help, please do!

I'm really confused here

OH

[geoff]

I don't know anything about the client/server, so hopefully someone else will help with that.

1) If I pick some numbers from the file archive , how is the easiest way to find out what I should make as B1? I know that this place has a table, but I believe the numbers have far more digits.

The number of digits in this table refers to the size of the factor, not to the size of the number being factored. B1=3,000,000 is ideal for finding factors of 40 digits regardless of the size of the number being factored.

4) Cunningham. What are the Cunningham Tables? I thought people should just run ECM on the list to search for random factors, so what's the deal with the tables?

To choose the right B1 for ECM you need to have an idea of how much work has already been done. If the optimal amount of curves have been done at B1=1,000,000 then it is likely that any remaining factors are larger than 35 digits and so it is better to use a larger B1. To know when to give up doing ECM and try another method like SNFS/GNFS you need to know how long these methods would take, which usually depends on the form of the number and how many factors have already been found, not just on its size in digits. The Cunningham tables have all this information.

[OmbooHankvald]

I've got one more question: What is ECMNET? Is it the project searching for Cunningham factors? Is it a pseudonym for the crew behind GMP-ECM? Is it the client/server ecm application?

[Xyzzy]

http://www.loria.fr/~zimmerma/ecmnet/

It is frequently down... I wonder if he needs a different host or something? We have room here if he needs it...

[OmbooHankvald]

Thanks Xyzzy, but I already know that site (though right now I can't remember what it contains).
On another forum people have described ECMNET as the ECMclient/server application and on Kirk Pearsons site ECMNET is a project.

[Citrix]

Originally Posted by Xyzzy

http://www.loria.fr/~zimmerma/ecmnet/

It is frequently down... I wonder if he needs a different host or something? We have room here if he needs it...

May be you can host a mirror to the site, so people can access it when it is down.

Citrix

[wblipp]

Originally Posted by OmbooHankvald

I've got one more question: What is ECMNET? Is it the project searching for Cunningham factors? Is it a pseudonym for the crew behind GMP-ECM? Is it the client/server ecm application?

Here's my impression - perhaps somebody with more history can provide corrections.

I think ECMNET was a project to find large factors using ECM - I think "large" originally meant 50 digits. During this era they created the ECM client and server programs. With the finding of large factors and the rise of NFS the sense of a specific project with a mission diminished. What is left is a home site for the client and server software and record pages for large ECM factors. I think Paul (aka xilman) now manages the server that is the ECMNET server. I believe it is usually stocked with Cunningham Project numbers.

In summary - ECMNET is a project that now has a low profile and is better known as the home of the client-server software and the ECM records pages.