SoB Sieve?

[Biggles]

Just thought of another project that you don't have listed in the DC Vault. It's the sieving sub project of Seventeen Or Bust, which has seperate scores. It, like PSP and RieselSieve, uses Proth. More information on the site: http://www.aooq73.dsl.pipex.com/index.htm

As far as I can tell it meets the criteria.

Cheers
Biggles

[NeoGen]

I had also asked about that project sometime ago...

[Biggles]

My bad. I some how totally missed your topic when I had a look before posting. My excuse is the time :)

Still, the fact I made a thread too suggests that this should be one of the projects!

[NeoGen]

Yep, that's two votes in favor alrealdy! :)

[NeoGen]

I had heard about the classic climate prediction still be running but never got to see it. I ran the classic climate prediction but it was years ago, then boinc came and I moved over to boinc.
I went around the site a couple of times recently and couldn't find a link anywhere to download it.

[russkris]

So far from what I have read.. I can't see where it tells you how to create a team, join a team...

Maybe you could put together a case to review
maybe you start by linking to the following

The project must:
1. be active
2. accept new members and teams immediately upon registration
3. have parsable team stats
4. have team stats that are updated regularly
5. provide a client program which runs on a local PC

The project must not:
be a keylogger or mouseclick counter
have a maximum number of teams or members

I'm not saying it cant be added to the Vault, but even the "Offical" Forum hosted by FreeDC, doesn't have any links.... Maybe you could provide some links

What do you say??

[garo]

Just like in PSP and RieselSieve, the user and teams are shared with the main SOB project. So once you have a username and team for the main SOB project it automatically adds you to the sieve stats. Sieve stats are submitted online using the SOB submission page http://www.seventeenorbust.com/sieve or using the Sobistrator which does it automatically. In any case, you log in using the same username and password as with the main project and the points are credited automatically to you and your team in the next stats run (every 6 hours). So if the criteria are met for SOB, they are also met for SOB sieve as far as user/team creation goes.

[russkris]

OK well it sounds like the criteria has been met, but what about Links to clients, a how to page......?

Maybe if you could get as many links as possible, so far the top post has a external source link to 17oB Seive, but what about the "offical" site

Sorry to be a pain but if a project has been requested to be add we need to have everything all in front to reveiw

[garo]

Not at all. Your request is quite reasonable.

Here is the forum page that contains all the links as well as the manual reservations thread. This can be treated as a meta-page that has links to all links given below as well.
http://www.free-dc.org/forum/showthread.php?t=3501

Client download:
http://81.233.34.242/proth_sieve.php
permalink: http://mklasson.com/proth_sieve.php

Sieve Info/overview page:
http://www.aooq73.dsl.pipex.com/

Page for creating an account (main SOB project as well):
http://www.seventeenorbust.com/signup/

Complete setup with instructions inside:
http://www.teamprimerib.com/vjs/Sieve-setup_with-instructions-inside.zip

Stats Page:
http://www.aooq73.dsl.pipex.com/scores.htm

Hope these are sufficient.
Garo

[NeoGen]

I would also add the Automatic Sieve Range Reservation page:
http://mattford.org.uk/sob/

and the Sieve Result Submission page:
http://www.seventeenorbust.com/sieve/


To use the automatic sieve range reservation page, you have to create a username/password on it.

[russkris]

OK that is great now the Admin have some linky's to goto..

But the question I have is Why aren't these links to the offical 17oB site...

[NeoGen]

Maybe because it's a bit more complicated, with all the things about client versions, sob.dat files, reserving ranges, etc...
Their main SOB client is pretty simple, just register, download, install, connect to the net and let it work.
They probably prefer to keep this project in the background to have only a small core of people that really know what they're doing instead of a million people all asking the same questions because they don't understand how all this works.

[Ragnarog]

Which leads to the obvious question....is it time to revise the criteria to say "installation must be reasonably straight forward"?

I see that this is really giving advantages to those "few" tecchies out there, leaving well over 95% of the people out of the loop due to, in this case (IMHO) very sloppy project support (if they'd try harder it would work as simple as their other client, they show it can be done)

Rag

[russkris]

Well that was going to be my next post.. I don't know how PSP and RieselSieve submit the results. But I don't think it's an online thing...

[shauge]

I have been running PSP and I have noticed that it is possible to do a combined sieve with SoB. This is very sensible because it prevents the same ranges from beeing sieved twice by the same program, but how does this come out in the stats? Is it separate stats or do you get double credits? How to join this?

[garo]

Actually both PSP Sieve and RieselSieve (sieve) have exactly the same modus operandii as SOB sieve. Infact, both these projects started after SOB and gained major inspiration from SOB. The client ProthSieve was written for SOB and later adapted to work with RieselSieve (it works with PSP without any changes).

Shauge - PSP and SOB may have overlapping p ranges (p=potential factor) but not overlapping k and n so they is no potential overlap in the work. It is true howeve that doing joint sieving is more efficient.

As regards criteria for project inclusion, it's your site and you decide. But SOB sieve is as easy or as hard as PSP and Riesel so I see no justification in adding two of the projects and not the third.

SOB actually has a zip file (instructions included) where all you have to do is unzip and run the executable. I mean how much easier can installation get? Far easier than PSP sieve which is also included in your stats.

Finally, I would say that I disagree with Neogen's assessment of why SOB sieve is the way it is. Sobsieve was never integrated into the automatic client and thus the project scoring was kept separate. it started as an afterthought and that is why there are so few links on the main site. But note that the first link I gave you is to the official SOB forum - it is hosted by free-dc - so it is as official as it gets. Also the sieve submission page is on the main SOB site.

russkris - yes submitting results is an online thing for Riesel and an emailing things for PSP.

You may wish to do a bit of research on the projects you have already included as many of them don't seem to meet the criteria you are now inventing!

[shauge]

Shauge - PSP and SOB may have overlapping p ranges (p=potential factor) but not overlapping k and n so they is no potential overlap in the work.

Can you explain what is meant with the statement "combined testing with SOB" in the PSP sieve guide thread http://www.mersenneforum.org/showthread.php?t=2666, if it does not mean a potential overlap in the work?

(It is OK to add SoB sieving, but everybody should have the same information)

[russkris]

OK, Now we have your thought and links... Excellent.. Give it a day or so the Admin can chat or should I say PM

[Ragnarog]

Gets my vote to be added, after all its not more hassling than the other sieving projects :)

Rag

[russkris]

Well if Rag's is Happy, I'm happy

[garo]

Thanks guys! I may have come across as being a bit hot under the collar there in my last post but that was not my intention :p

shauge: Both SOB and PSP are trying to eliminate candidates of the form k*2^n+1. This is done through sieving (basically finding factors of each such number). SOB has 8 such k's for which it is trying to find factors. Proth_sieve works by processing a file called sob.dat which contains all n's from 1 to 50 million for each k which have not yet been proven composite. PSP does the same but for a different set of k's. It was discovered that if the files that contained the k's and n's were combined, we get a speedup over doing the sieving separately. Of course, this constrains the sieving range to be the same for both sets of k's so the set of p's (potential factors is the same).

[Mystwalker]

Can you explain what is meant with the statement "combined testing with SOB" in the PSP sieve guide thread http://www.mersenneforum.org/showthread.php?t=2666, if it does not mean a potential overlap in the work?

The ranges that are sieved are specified by their "n" value (all these projects search for primes of the form k*2^n +/- 1, with k as a rather small set of numbers (~75 for Riesel, 12 for PSP and 8 for SoB) and n as continuously incresing values), but the k values differ between these projects.

SoB and PSP both search for "+1" numbers, so it is possible to search for all 20 k's at the same time.
The benefit is that doubling the "k" count doesn't half sieving speed. In fact, sieving all 20 k values instead of PSP's 12 results in a ~20% slowdown. Hence, it is faster than sieving the k sets one after another.

[shauge]

garo and Mystwalker: I now understand what combined sieving is, thanks. I guess DC-Vault will as it catches on, increase the interest in sieving, but there will be asked some "stupid" questions from some of the newcomers. Thanks for your patience. I am also sorry if I sounded harsh, it was not intended.