Riesel base 2 discussion

[ValerieVonck]

Rieselsieve.com is still not up
Is it not time for action????
I will see that I can liberate a core for it (sieve)

[gd_barnes]

Originally Posted by CedricVonck

Rieselsieve.com is still not up
Is it not time for action????
I will see that I can liberate a core for it (sieve)

Whatever you do, do NOT sieve! The file has been sieved into the ground. All that we should for the next 5 years is look for primes! Perhaps when primality testing is approaching n=7M-10M, the file should be sieved further.

Carlos had indicated a while back that he might be interested in creating an LLRnet server for it. I'm sure that IronBits would also.

If you want to, create a thread and coordinate the effort. If needed, I can edit reservation/status postings as needed. Testing should begin at n=2.947M on all except 2 k's. k=2293 and one other k are already searched to n=4M.


Gary

[ValerieVonck]

Anyone still interested in this project?
I saw that dc.rieselsieve.com is "backup" again.... presumably with my figures

[gd_barnes]

Originally Posted by CedricVonck

Anyone still interested in this project?
I saw that dc.rieselsieve.com is "backup" again.... presumably with my figures

Good for them. I'm not interested in it. Perhaps others are.

Are they actually doing primality testing now or is the site just "fluff"?

One bit of advice for them: Stop the sieving completely and look for primes. Sieving should have stopped long ago.


Gary

[MooooMoo]

Originally Posted by gd_barnes

Are they actually doing primality testing now

I think they are. Here were yesterday's results, from http://dc.rieselsieve.com/

Last 10 tests completed:
(didn't copy the name) 336839*2^3627836-1 4928 points 
NeoLogic: 336839*2^3633644-1 4943 points 
KWSN_Lady_DC: 336839*2^3634196-1 4945 points 
KWSN_Lady_DC: 336839*2^3634020-1 4944 points
KWSN_Al_Dente: 336839*2^3633860-1 4944 points 
NeoLogic: 336839*2^3633956-1 4944 points 
Happy5214: 336839*2^3631364-1 4937 points 
KWSN_Lady_DC: 9221*2^3634086-1 4945 points 
KWSN_Lady_DC: 9221*2^3634014-1 4944 points 
TongZhou: 9221*2^3629154-1 4931 points

Today's results:

KWSN_Lady_DC: 9221*2^3628830-1 4930 points 
NeoLogic: 9221*2^3635790-1 4949 points 
NeoLogic: 9221*2^3635538-1 4949 points 
KWSN_Al_Dente: 9221*2^3635214-1 4948 points 
cswchan: 9221*2^3632970-1 4942 points 
Dr_Vornoffs_Hex_Mach: 9221*2^3634338-1 4945 points 
TongZhou: 336839*2^3629180-1 4931 points 
NeoLogic: 9221*2^3613434-1 4888 points 
KWSN_Lady_DC: 336839*2^3628508-1 4929 points 
NeoLogic: 336839*2^3635084-1 4947 points

[kar_bon]

there're only 3 k's in the test-queue:
9221, 336839 and 444637 with n_min=3.6M

the links on top of the page (Info,Stats,Download) don't work yet!

the current k-status (http://dc.rieselsieve.com/kstats.php )shows only 270 pairs under n=3M.

[gd_barnes]

Originally Posted by kar_bon

there're only 3 k's in the test-queue:
9221, 336839 and 444637 with n_min=3.6M

the links on top of the page (Info,Stats,Download) don't work yet!

the current k-status (http://dc.rieselsieve.com/kstats.php )shows only 270 pairs under n=3M.

Why would they only be testing 3 k's near n=3.6M when there are still k's showing with tests remaining for n<3M?

The status page is woefully out of date. k=485767 is shown as remaining. A prime was found for it 6 months ago. I'm beginning to think that everything n>2M should be double-checked there.

Stuff like this calls into question the actual sieve file that they are working with. That is, are there k/n pairs removed that should not have been? It kind of makes one wonder.


Gary

[mdettweiler]

Originally Posted by gd_barnes

Why would they only be testing 3 k's near n=3.6M when there are still k's showing with tests remaining for n<3M?

The status page is woefully out of date. k=485767 is shown as remaining. A prime was found for it 6 months ago. I'm beginning to think that everything n>2M should be double-checked there.

Stuff like this calls into question the actual sieve file that they are working with. That is, are there k/n pairs removed that should not have been? It kind of makes one wonder.


Gary

Last I heard, they had transferred most of their k's into their BOINC LLR system, and had left just 3 k's (high-weight if memory serves) for LLRnet. I don't think they've gotten their BOINC system back online yet, so right now they're only doing their three LLRnet k's.

[gd_barnes]

Originally Posted by mdettweiler

Last I heard, they had transferred most of their k's into their BOINC LLR system, and had left just 3 k's (high-weight if memory serves) for LLRnet. I don't think they've gotten their BOINC system back online yet, so right now they're only doing their three LLRnet k's.

Ah; that's a little better. At least there is progress now. There is still nothing that I can accurately link to from my web pages for the effort so I'm still showing each k remaining separately.

[jasong]

I think that it's possible that Lee(B2_Riesel) may be intentionally trying to kill the project. I have multiple reasons for thinking this.

Reason #1: Lee claimed that he was court ordered to stay off the Internet for 60 days because the government suspected him of using the RS project as a front for a cracking operation. I know it sounds stupid, but that's basically the gist of it. Yet I know of at least one person who's claimed to have seen a known Ikariam(online browser game) account of his with activity.

Reason #2: Before he disappeared, he had told people he would only be offline for about 15 minutes to move the servers. Now, this is mere conjecture, but I can imagine him having unanticipated troubles and simply saying "F*** it, I quit."

He has given the project over to punchcard. I don't know punchcard's real name, apparently he was a co-worker of Lee's in meatspace at one point. It is my opinion that this was done simply because punchcard is better at being a hardass and not backing down. I believe they intend to kill the project and decided punchcard would be better able to deal with the problematic questions about the project.

This is just my opinion, based on experience and watching #rieselsieve channel, as well as some PMs in IRC. I wouldn't be surprised if this gets me permanently banned from #rieselsieve, we'll see.

I think the Riesel Sieve project is dead, if the Riesel conjecture continues to be studied, I don't think it will be done by Riesel Sieve. I also think that it will be very difficult to get any of the known results in the database if anyone wants to continue the project.

[MyDogBuster]

Quote:
Originally Posted by MyDogBuster http://www.mersenneforum.org/images/...s/viewpost.gif
I don't know who is older, you or I, but you summed it up perfectly.

PS. Riesel Sieve was my favorite project also. Too bad b2 went the direction he went. I wonder whatever happened to those humungous sieved files?

Never fear. I have that file. You don't think I'd be running CRUS and not follow up did you? Technically, it is a direct subset of the CRUS project.

If someone wants to give me a little (OK, maybe a BIG) push, I could be talked into setting it up to start testing on the file again. But PLEASE, no more sieving. The file is WAY WAY over-sieved as it is. It just needs primality testing done on it. We'd likely start at n=3M but I'd need to check into test limits a little more.

If we did something like that, we'd need these things:
1. A hardy server to run it on.
2. It's own stats page like what we have for NPLB.
3. Someone to handle the political fall out from "taking over" the Riesel Sieve project. [Not sure how it could be taking over the project since the project hasn't existed for 9 months now.]

Part of the reason I haven't pursued it is that I didn't want to drag resources away from NPLB. I've strongly felt for a long time that what is needed to find and unlock the "key" to the primes is to find a lot of smaller primes (if you can call n=500K-600K small ); not to find a few huge primes.

But the main reason is the political fallout. I hate politics and self-admitidely don't handle politics well. (Same way in my personal life too, much to my chagrin. lol) So if we did something like that, I'd need a P.R. guy to handle any crap that comes down. BTW, that is why I asked Max to help with CRUS and NPLB. He does the P.R. stuff well and has the techie knowledge to help where I fall short there. I've given him a hard time in these forums sometimes when he gets a little busy but it never phases him. Thanks Max!

Anyway, since it'd be a sub-project of CRUS, it could be handled there with it's own thread or sub-forum.

For that matter, I'd be just as happy to assist someone else who would want to take over such an effort.

If we get into an extended discussion on this, I'll move it over to a thread where we talked about RieselSieve before at CRUS and then provide a link to it from here.

Edit: BTW the RieselSieve file is about half the size of our k=1003-2000/n=500K-1M file...17.6 MB vs. 32.7 MB unpacked in ABCD format. Of course it's only 64 k's vs. 500 k's but it does go from n=100 to 20M. (That's 100, not 100K.)


Gary

Question #1. About the sieve file mentioned above. Did Riesel Sieve remove the remaining tests from the file when they found a prime? I can't remember one way or another.

Question #2. How much flak would be expected. Who is going to have a cow?