Riesel base 2 discussion

[Jean Penné]

Hi All,

Did you remark that http://www.rieselsieve.com/ is now again active ?

It says :

"The Riesel Sieve project is currently undergoing reconstruction."

" Thanks for your patience!"

Regards,
Jean

[ValerieVonck]

Any news of the project?
It is already a month ago when the last was made...

[KEP]

There is no news, except the project is undergoing reconstruction, so with a little luck, maybe around Christmas or New Year we will start to see the RS project get moving again

I just hope that they abandon their sieve part untill a new sievefile is needed, and then re-launch their project with an overall LLR attack on the k's remaining and thereby starts to bring down the k's remaining below the 3.4M pairs, as a result of k's being primed in stead of k/n pairs being factored

KEP!

[gd_barnes]

Originally Posted by Jean Penné

Hi All,

Did you remark that http://www.rieselsieve.com/ is now again active ?

It says :

"The Riesel Sieve project is currently undergoing reconstruction."

" Thanks for your patience!"

Regards,
Jean

It's been that way for a month now. They don't say what kind of reconstruction they are talking about, which is typical of the way it's been for the last ~3-4 months.

Unfortunately I've been so busy with NPLB and other CRUS efforts that I haven't had time to start any kind of effort here on it. The main thing I'm concerned about is getting the n-range for each k where there is ZERO remaining k/n pairs to be tested below that range. There have been several postings here but none has clarified that. If it's the "min n", I need to know that. I know it's not the "max n". The problem that I'm having is that the TRUE search range is what I think to be the "min n" and if that is the case, it is only around n=2.3M to 2.4M on most k's; not the 3M to 3.4M that people have been stating. We can't just start there and leave untested n-values below, even if it's only a few of them.


Gary

[gd_barnes]

Originally Posted by KEP

There is no news, except the project is undergoing reconstruction, so with a little luck, maybe around Christmas or New Year we will start to see the RS project get moving again

I just hope that they abandon their sieve part untill a new sievefile is needed, and then re-launch their project with an overall LLR attack on the k's remaining and thereby starts to bring down the k's remaining below the 3.4M pairs, as a result of k's being primed in stead of k/n pairs being factored

KEP!

Yes, the sieving needs to be completely abandoned for YEARS!! (no joke) The sieving is likely sufficient for testing up to n=6M or higher! Primality testing is all that is needed at this point for a very long time.


Gary

[ValerieVonck]

Originally Posted by gd_barnes

It's been that way for a month now. They don't say what kind of reconstruction they are talking about, which is typical of the way it's been for the last ~3-4 months.

Unfortunately I've been so busy with NPLB and other CRUS efforts that I haven't had time to start any kind of effort here on it. The main thing I'm concerned about is getting the n-range for each k where there is ZERO remaining k/n pairs to be tested below that range. There have been several postings here but none has clarified that. If it's the "min n", I need to know that. I know it's not the "max n". The problem that I'm having is that the TRUE search range is what I think to be the "min n" and if that is the case, it is only around n=2.3M to 2.4M on most k's; not the 3M to 3.4M that people have been stating. We can't just start there and leave untested n-values below, even if it's only a few of them.


Gary

Why shouldn't we start from n=2M for the remaining k's?
That to show we have good "cuttoff" point?

[gd_barnes]

Originally Posted by CedricVonck

Why shouldn't we start from n=2M for the remaining k's?
That to show we have good "cuttoff" point?

Well, we know we are up to n=2.3M-2.4M per the above posted web page that was archived in 2006 so there's no use to start so low. When the project gets going again, they can double-check themselves. We don't need to do it. So...perhaps starting from n=2.8M or 3M would be OK.

KEP, do you happen to remember what the "Min n" was on most k's before RieselSieve went down?

[KEP]

Originally Posted by gd_barnes

Well, we know we are up to n=2.3M-2.4M per the above posted web page that was archived in 2006 so there's no use to start so low. When the project gets going again, they can double-check themselves. We don't need to do it. So...perhaps starting from n=2.8M or 3M would be OK.

KEP, do you happen to remember what the "Min n" was on most k's before RieselSieve went down?

The min. n was about 2,947,000 for all k's. Most tests for n's from 2,947,000 to n=3,000,000 was tested, since 2,947,000 apparently referred to an outstanding n, which had not yet been returned. So All numbers below n=2,947,000 has been doublechecked, since that was practically all that RS ever managed to do while using BOINC.

KEP!

[gd_barnes]

Originally Posted by KEP

The min. n was about 2,947,000 for all k's. Most tests for n's from 2,947,000 to n=3,000,000 was tested, since 2,947,000 apparently referred to an outstanding n, which had not yet been returned. So All numbers below n=2,947,000 has been doublechecked, since that was practically all that RS ever managed to do while using BOINC.

KEP!

OK, very good. Thank you for the specific answer.

What does everyone think of us setting up a server to restart Riesel base 2 from n=2947000 on all k's except k=2293 and 342847. Clearly those k's were at n>4M and likely tested higher by individuals. There is no reason to test them right now.

Before doing this, I'll post our intentions in the RieselSieve forum.


Gary

[KEP]

Originally Posted by gd_barnes

OK, very good. Thank you for the specific answer.

What does everyone think of us setting up a server to restart Riesel base 2 from n=2947000 on all k's except k=2293 and 342847. Clearly those k's were at n>4M and likely tested higher by individuals. There is no reason to test them right now.

Before doing this, I'll post our intentions in the RieselSieve forum.


Gary

I support this idea. However, I'm probably never gonna be able to support this in the near future. But to all of you who have the resources to support the effort, please feel free to count on my support. Also it's most likely a good idea to post a message on your intention in the RS forum, since there is really no need for a multi-run on 2 different servers, testing the same numbers, since this is big numbers.

Regards

KEP

[gd_barnes]

Originally Posted by KEP

I support this idea. However, I'm probably never gonna be able to support this in the near future. But to all of you who have the resources to support the effort, please feel free to count on my support. Also it's most likely a good idea to post a message on your intention in the RS forum, since there is really no need for a multi-run on 2 different servers, testing the same numbers, since this is big numbers.

Regards

KEP

I posted our intent to revive Riesel base 2 by requesting that Carlos and/or David possibly host an LLRnet server, but then I changed my mind and promptly deleted the post at http://dc.rieselsieve.com/viewtopic....st=0&sk=t&sd=a.

If you look at the end of that thread, Bryan did finally respond on Oct. 9th. I hadn't seen it because I wasn't used to the way the site navigates. Unfortunately his post doesn't really give much useful info.

At this point, as all-encompassing as CRUS is, I'm going to bow out of any kind of administration of the Riesel base 2 effort. As much as it pains me to see it go dormant for so long, I feel that the controversy that it might create is just not worth it.

I'll be perfectly willing to support anyone who wants to set up threads, a sub-project, and/or a subforum for running the effort here but I won't make an effort myself to get the ball rolling in that regard. I'll also support and advertise any server that anyone might be interested in hosting for the effort. I have the sieved file from KEP and can provide it if needed.


Gary