• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

If you like SOB then try Riesel Sieve

SlangNRox

Golden Member
I just started a team at http://www.rieselsieve.com/ This project is similar to seventeen or bust, in that it is trying to find primes, but there are 81 k that they are trying to find primes for unlike the 13 for sob. Also a prime was just found on 12/21 and on 12/6. I started a team a couple of weeks ago but it is getting lonely being the only one on the team. I thought I'd try to drum up some support since I initially formed the TA sob team and kick started the dormant lifemapper TA team.

You can either run the automated llrnet client to search for primes or the much less automated proth sieve to eliminate k/n pairs. The information on their website and forums can explain it much better than I can. They also have an irc chat where everyone is really helpful.

I just thought someone might want to try it out. Stats were done manually and not on a regular basis, but the stats are automatically being run every 30 minutes now (or at least I think they are) I can test a k/n pair in aboutn 2.5 hrs while a sob test now takes me over 50 hrs to do. If you want to find a prime this would be the place to do it at.

Thats all the promotion I'm going to try doing. If you do join me, please tell them on irc to put you on Team Anandtech or pm me and I'll tell them to add you to TA.
 
I have been crunching on RieselSieve since early last spring. Primarily sieving, but have also done some llrnet. The stats are now working, which is a great improvement. More people are starting to jump in there, both for sieving and llrnet. Athlon XP's are the cpu of choice for sieving, while the intels rule the llrnet. I see now that Team Rechenkraft is there, along wiht a couple other of the heavy hitters.

If some people want to play on this project, then I'll put some systems back on it and move my points and stuff over to TeAm Anandtech. I'm currently in #2 for sieving, and #25 for llrnet. As SlangNRox said, it gets lonely crunching by yourself. I'd crunch with a team but I don't care to do it by myself, and will put my systems on something else otherwise. As the infrastructure continues to improve, you are going to see some heavy crunching going on here. So it is a good time to get in on the ground floor. New/reworked clients will be on the street before much longer, and work is proceeding on trying to get a 64bit client, initially for linux (Suse9.1).
 

Yo SlangNRox,

They just started an open beta on client v3.5 for llrnet. Really is a massive improvement in speed. If you are running llrnet, you need to get the beta downloaded and start using it. Both a win32 and a linux beta.
 
Originally posted by: Brucifer
I have been crunching on RieselSieve since early last spring. Primarily sieving, but have also done some llrnet. The stats are now working, which is a great improvement. More people are starting to jump in there, both for sieving and llrnet. Athlon XP's are the cpu of choice for sieving, while the intels rule the llrnet. I see now that Team Rechenkraft is there, along wiht a couple other of the heavy hitters.

If some people want to play on this project, then I'll put some systems back on it and move my points and stuff over to TeAm Anandtech. I'm currently in #2 for sieving, and #25 for llrnet. As SlangNRox said, it gets lonely crunching by yourself. I'd crunch with a team but I don't care to do it by myself, and will put my systems on something else otherwise. As the infrastructure continues to improve, you are going to see some heavy crunching going on here. So it is a good time to get in on the ground floor. New/reworked clients will be on the street before much longer, and work is proceeding on trying to get a 64bit client, initially for linux (Suse9.1).



How many systems do you have ...
 
Send a PM to b2 on the rieselsieve site, and tell him what team you want to belong to, and he will see that you are placed in it. 🙂

So are you sieving or running llrnet?

I had b2 move me to the Anandtech team. I'm running hard on the llrnet now, have really slowed down on sieving. llrnet is the part where you are testing for actual primes.
 
Originally posted by: Brucifer
Send a PM to b2 on the rieselsieve site, and tell him what team you want to belong to, and he will see that you are placed in it. 🙂

So are you sieving or running llrnet?

I had b2 move me to the Anandtech team. I'm running hard on the llrnet now, have really slowed down on sieving. llrnet is the part where you are testing for actual primes.

I'll PM him when I get home (IIR). I'm sieving, since you mentioned that was better for Athlon XPs. 🙂
 
Originally posted by: n0cmonkey

I'll PM him when I get home (IIR). I'm sieving, since you mentioned that was better for Athlon XPs. 🙂


Well it was better before the 3.5 beta came out for llrnet. And sieving at present is still ripping through numbers, however we are reaching a point where other issues come into play and the previous benefits to sieving are beginning to wan due to software limitations. Meanwhile the llr (llrnet) portion is the testing that actually finds the primes, which is a goal of the project. Bryan (on the rieselsieve project) posted this morning that 543 llr tests were completed yesterday, which he (they) feel is a phenominal number based on past performance. So I shifted my stuff over to running llrnet.

The 3.5 client really rocks on P4's and celerons. The new client makes it possible to use XP's now. Even while the XP's don't match the speed of the intels, they are putting out more now than the intels used to put out before the 3.5 client. So I guess what I'm saying is that running XP's on llrnet isn't a waste of a cpu. For instance in early December on a 2.4c HT P4 running one instance of llrnet, it was taking on average 9.218ms to complete one iteration of a test. That same system averages 4.110ms now. I have an xp2400 system taking 6.908ms per iteration, which is definitely slower that the 2.4c system, but it is quicker than what the 2.4c was running before the latest 3.5 beta llrnet code. And this means that the XP users now have a much better chance of getting in on the game of finding a prime number! 🙂 By the way, my timing examples above are on linux.

In talking with b2, the head banana on rieselsieve the other day, he mentioned that we were reaching a point where sieving more than likely would be taking a back seat now with the advances in llr software. And while the stats are not automated yet, that is in the works at present. Even with the new llrnet client there is still a ton of work to be completed yet, so the project is not going to be running out of work for years. And also, while it is not as high profiled as some of the other math projects, it keeps slowly growing as any project does in it's infancy. Just part of the ground floor thing. 🙂
 

Sounds like an interesting project...
What about Hyperthreading/dual processors, can it take advantage of that, and how do you do it?
I didn't see any descriptions of this at the projects website/forum.

I'm running an instance right now of the Win32 CLI LLRNET client, and it seems that the times I see for one iteration is approximately 12.3ms.
I have a P4 2.8@3.1, and based on the above description it seems a tad slow I think?
Is there some tweaks one should know about, or???
 
Things they need to work on:
DOCUMENTATION- There isn't enough of this (especially relating to installation
Client installation- It shouldn't be this difficult 😕
 
Originally posted by: n0cmonkey
Things they need to work on:
DOCUMENTATION- There isn't enough of this (especially relating to installation
Client installation- It shouldn't be this difficult 😕



Agreed !!!
There should some kind of whitpaper describing the basic steps for setting it up, and not just a small thread describing settings in a file. I'm pretty sure this alone scares off some people.
Maybe n0cmonkey and I haven't looked at all the right places, but I also think something should be done.
Right now I'm running the client just out of curiosity, to see if it sends some results automatically, or if I have to do something manually... 😕 (Also to see if it runs smooth on my PC)

Still looks interesting tho. 🙂

 

Yes, documentation would be a help! The "download" page has the standard client, not the beta. To get the beta go to:

windows:
http://download.rieselsieve.co...a-llrnet.win32-cli.zip

linux:
http://download.rieselsieve.com/beta-llrnet.linux.tgz

For setup on windows, put the zip file in a directory unzip it. After unziping you should see a sub directory called llrnet.win32, change into this directory. Using notepad, edit the file named llr-clientconfig.txt ---- what you need to do is on the line username="zip" or whatever name it is, I can't remember, place your username you want to use between the quotes. There is also a line:
--llrVerbose(1)
which you need to delete the two dashes, if you want to see timing information written to the screen.

To fire it of, in a command line window just type in llrnet and press enter and it will be off and running, or else just click on llrnet.exe if you are using filemanager. To stop the program, when the window it is running in is highlighted, press CTRL-C and it will halt. When you restart it, it will run again from the save point it made when shutting down. It will automatically contact the server to get work units, and to deliver completed ones.

For linux, just unzip it via tar, or else use gunzip beta-llrnet.linux.tgz which will change it to a tar file, and then just use tar -xvf beta-llrnet.linux.tar and it will blow the tar apart, creatting a subdirectory, and then you just edit the same file as explained in the windows section above. As with windows, it just runs, takes care of contacting the server for more work or uploading completed work units.

You can also change the number of cached work units via the config file, up to a maximum of 100 units. At the completion of each work unit, it will try and contact the server to upload the result, and download a replacement work unit. If it can't contact the server, it saves the result and starts in on the next result in the cache.

Yes you can use hyperthreading, and just create a second directory, I set mine up with /riesel/node1/llrnet.linux and /riesel/node2/llrnet.linux and run two instances that way under linux. I haven't used HT under windows, but I would assume the same basic setup would work.

As for optimizations........... for linux/unix you can recompile for your own system if you want to go through that. I just use their precompiled client. For linux, recompiling may be necessary if you have a non-installed library issue. I haven't had that problem though and I use slackware 8 through 10.0, Suse 8.2, and also Libranet 2.8.1 updated.
 
Hello all.

Let me introduce myself. I'm Lee Stephens, founder of Riesel Sieve. I'd like to thank SlangNRox and Brucifer for all the hardwork they have been doing at our humble little project. I'm very grateful to you guys for spreading the news about Riesel Sieve. I'll take a few moments to explain where we have been and where we are going.

Our project started out as a few friends reserving a 'k' from www.prothsearch.net and trying to find a prime for the Riesel Problem. What we quickly noticed was 101 k's being sieved individually and one user would never hand off any kind of paper trail to the next. I decided that a talk with Mikael Klasson and Paul Jobling was in order to take the very new prothsieve.exe and turn it from the Seventeen Or Bust version which was just about 13 k's at the time and numbers in the form of k*2^n+1, and turn it into a monster sieve that could rip thru 101 k's in the form of k*2^n-1. The first client to come out was Paul Joblings code of merely doing NewPGen and making it work on multiple k's and was called Rieselsieve.exe...hence the name of our project. Later Mikael Klasson converted his Prothsieve.exe to handle numbers of our form and we've been tweaking that client ever since.

Originally, I sieved 101 k values by myself from 2 to 1 billion individually for each k. I then took these 101 files and combined them and edited them to fit into Riesel.dat. Riesel.dat is the data input file for Prothsieve.exe. I then sieved all 101 k values myself from 1 billion to 10 billion. At this point I started a PHPBB2 forum and asked my friends to help me with sieving. Everyone grab a range and see if we can remove a lot of numbers. We offered our sieved numbers to the rest of the community. Those people on the www.prothsearch.net website who had a 'k' reserved in their name. As people 'gave up' or 'abandoned' their k reserves and moved on to other projects, I reserved the k values in my name at first and then under Riesel Sieve later on.

We individually, co-ordinated via forum posts and IRC only, started to work using the original LLR.exe and tried to find primes. Late in 2003 we found 261221*2^689422-1 to be prime. In 2004 we found 14 more primes and excelerated the completion date of solving the Riesel Problem. Our Primes are as follows:


----- -------------------------------- ------- ---- ---- --------------
rank description digits who year comment
----- -------------------------------- ------- ---- ---- --------------
20 192089*2^1395688-1 420150 L49 2004
46 502541*2^1199930-1 361221 L93 2004
49 71009*2^1185112-1 356760 L47 2004
54 350107*2^1144101-1 344415 L35 2004
56 500621*2^1138518-1 342734 L73 2004
57 504613*2^1136459-1 342114 L84 2004
60 412717*2^1084409-1 326446 L76 2004
61 150847*2^1076441-1 324047 L73 2004
78 309817*2^901173-1 271286 L64 2004
80 170591*2^866870-1 260960 L47 2004
81 93997*2^864401-1 260216 L49 2004
102 460139*2^779536-1 234670 L47 2004
104 246299*2^752600-1 226561 L42 2004
136 261221*2^689422-1 207543 L35 2003
184 279703*2^616235-1 185511 L53 2004
----- -------------------------------- ------- ---- ---- --------------

Early in 2004 we created LLRNET. LLRNET is the core of LLR on with a LUA wrapper to make it a network client. The client and server are the same code. So if you have the client...then you have the server. Only a command line switch and config file are needed. We encourage people to use a server themselves for a proxy server on their LAN. One word of note though....currently the server in proxy mode strips out the Userid and sends as the ServerUserid.

Our next version of LLRNET will by tied to MySQL as to help us automate stats and create what we hope to be the most extensive stats of any Primes based project and both informative and fun.

Someone mentioned documentation. This area of our project is woefully pitiful. I know we need to document everything in an easy and quickly informative way. Unfortunantly I get tied up in the other parts of the project and ignore the issues that are my weakness. I may like to write 500 page forum posts but I detest writing how-to's and instructions. I will make a promise to get someone working on this issue as soon as possible though.

My biggest concern at the moment is building PHP pages for user registration, personal stats pages, personal admin pages, and tying all that in with the client/server. If anyone has experience with building professional looking windows installers that can install the client and register new users...please feel free to contact me.

Our goals for this year are to find no less than 20 new primes. At first our primes will be Top 50 primes and as we move thru the ranges we will end the year with most likely our first Top 10 prime number. Finding primes in the millions of decimal digits will be common for us next year. With our project, we have formulas for knowing...well educated guessing...how many primes we will hit, where and when. This being said, our project WILL have results. We will hit primes and each one will be bigger than the last. We will have a long way to go to find 81 primes and prove the Riesel Problem, but each year we will bring that goal closer and closer.

Thanks to all that have read this far. I hope to see each and every one of you involved in our project. May you join, may you find primes. If anyone has any questions at all. Please feel free to drop by the forums on www.rieselsieve.com or stop by IRC irc.freenode.net #rieselsieve and ask me or one of the staff questions about our project.

Lee Stephens
B2
Riesel Sieve
 
b2riesel welcome to our little corner of the internet. 😉 Thanks for the info, it's an interesting project so far. If I get some time, I might be able to help out a bit with the documentation. Too many projects, not enough time. 😉
 
sounds like pretty much the same as lifemapper for HT setup (just create 2 seperate folders). im gonna join and install this bad boy! thanks for the information guys 🙂

here is a more detailed description of the configuration file that brucifer mentioned above. seems pretty darn easy to configure based on the minimal config options.
 
a few questions here:

1) can i upgrade to the latest beta edition of the llrnet cli while i still have work in my cache from the previous cli?

2) are there any 3rd party windows addons for llrnet?

TIA
 
Originally posted by: Unforgiven
a few questions here:

1) can i upgrade to the latest beta edition of the llrnet cli while i still have work in my cache from the previous cli?

2) are there any 3rd party windows addons for llrnet?

TIA


The uncrunched work units are in a file called workfile.txt. Anything that needs to be sent to the server is in tosend.txt, and the file z00000... is the current workunit. The only one I'd even worry about would be the workunit.txt file, which you can just copy over to the new beta directory.... or save it in a sub-directory and just unzip the beta zip file contents into the same directory and then copy the workfile.txt back into the directory. The other thing is to just forget it and start anew with the beta installation. After six days or something like that, the server will reissue those numbers to someone else anyway.

As for 3rd party addons, I don't know of any. the llrnet software can be a server or a client, depending how you set your stuff up. Once things stabilize, and a release client replaces the beta one, maybe something will come along from a 3rd party. I don't think I'd hold my breath though 🙂
 
Back
Top