2003-11-30, 23:18 | #1 |
Aug 2002
100000100_{2} Posts |
New to GIMPS? Start here!!!
New to GIMPS? Start here!
Welcome to the mersenneforum.org, the official forums for the Great Internet Mersenne Prime Search (GIMPS) project. This forum is for anyone participating in the GIMPS project regardless of team affiliation. It is moderated by volunteer users. Users can read and search this forum without registering. Users who want to post to it must register first, apart from on the Public Forum Ask Questions Here! where no registration is required. Questions about this forum (including any problems) can be directed to the moderator(s) via email, or to Xyzzy, the forum Manager. The purpose of this thread is to give new users, or people that are just curious, some information about the project, a little bit of information about the background of the project, information on how to get started with GIMPS, pointers to where you can find help, links to other GIMPS sites, and some other more or less useful pieces of information. For those of you that are only interested in how to get started, jump directly to the How to get started with GIMPS post. Contents
Material for the thread has been collected from various sources available on the net. Most important of these sources are in no particular order: The GIMPS Home Page, the Prime Pages, Ars Team Prime Rib, Ars Distributed Computing Arcana, Mersenne Prime Mailing List, mersenneforum.org, The Prime Monster Site, MathWorld, School of Mathematics and Statistics University of St Andrews Scotland, Intel, Landon Curt Noll, and others I do not remember. I am also very grateful to all the people that have helped me with the posts in this thread. A special mention goes to GP2, nfortino, Maybeso and Scott. They have pointed out errors, omissions and spelling mistakes. If there are any mistakes, omissions, inaccuracies or actual errors left, I am sure you can work out where to put the blame. -- PM PS: If you feel something is missing or there is something which is not properly explained, please send me a private message, and I will se what I can do about it Last fiddled with by Prime Monster on 2004-06-10 at 17:38 |
2003-11-30, 23:19 | #2 |
Aug 2002
2^{2}×5×13 Posts |
What is GIMPS?
The Great Internet Mersenne Prime Search, also known as GIMPS, is a prime example of Distributed Computing project at work and no pun intended. The goal of the GIMPS project is to discover new Mersenne Prime numbers. Mersenne primes are named after Marin Mersenne, a French monk and mathematician, who was born in 1588. Mersenne investigated a particular type of prime number: 2^{P}-1 (2 to the power of "P" minus one), in which "P" is an ordinary prime number. (How to pronounce Marin Mersenne) Mersenne primes are much rarer than ordinary primes, of which there are an infinite number. The GIMPS effort, exhaustively searching for possible candidates since 1996, has been responsible for discovering the six most recent examples. Altogether, in all of history only 41 Mersenne Primes have been discovered, and GIMPS have discovered the 7 largest. GIMPS was formed in January 1996 by George Woltman, as one of the first open Distributed Computing projects, to discover new world-record-sized Mersenne primes. All the necessary software can be downloaded for free. Most GIMPS members join the search for the thrill of possibly discovering a record-setting, rare, and historic, new Mersenne prime. All you have to do to be part of it, is to contribute spare or idle CPU cycles. Pretty cool. I emphasise that you do not need to be a mathematician to take part in GIMPS, in the same way as you do not need to be a mechanic to drive a car. It is one of the few projects where ordinary people can contribute to fundamental scientific research. Last fiddled with by Prime Monster on 2004-05-29 at 13:13 |
2003-11-30, 23:20 | #3 |
Aug 2002
2^{2}·5·13 Posts |
The 41st Mersenne Prime!
The 41st Mersenne Prime!
If you came here just to find out about this new Mersenne Prime number then here are the bare facts. The newly discovered Mersenne Prime number is 2^{24,036,583}-1, or 2 to the 24,036,583rd power minus 1. It was discovered by Josh Findley, on May 15th 2004, using prime95 on his office computer. He used a 2.4 GHz Pentium 4 Windows XP PC running for 14 days prove the number. Findley is an independent IT consultant living in Issaquah, Washington. He joined GIMPS in June 1999. The new prime was independently verified by Tony Reix of Grenoble, France using half of a Bull NovaScale 5000 HPC running Linux on 16 Itanium II 1.3 GHz CPUs for five days using the Glucas program by Guillermo Ballester Valor of Granada, Spain, and by Jeff Gilchrist of Elytra Enterprises Inc. in Ottawa, Canada using eleven days of time on a HP rx5670 quad Itanium II 1.5 GHz CPU server at SHARCNET. The discovery is the seventh record prime found by the GIMPS project. In recognition of every GIMPS contributor's effort, credit for this new discovery will go to "Findley, Woltman, Kurowski, et al". Findley, a consultant to the National Oceanic and Atmospheric Administration (NOAA) in La Jolla, California, described the find. "I'm still surprised at the discovery. Even after five years running GIMPS on my computers, I didn't expect to find a new Mersenne prime! I joined GIMPS because it seemed the logical choice for using my spare CPU cycles." Why, you may ask, do you have to verify the Mersenne prime using a different client? There is no difference in the math between the clients. Both implement a Lucas-Lehmer primality test using discrete weighted transforms. Prime95 was written in x86 assembly code. Glucas is written in C code. Since the clients were written by different people and run on different hardware, that virtually eliminates a programming bug or hardware flaw as a cause for a false M40 report. The new prime was unofficially double-checked by Mike and Alf Refsum (Xyzzy and Prime Monster) on a 3.2 GHz P4 machine, but since it was run using the prime95 client this run does not count as an official verification run. Some trivia about the 41st Mersenne Prime! The new Mersenne, 2^{24,036,583}-1 has 7,235,733 decimal digits, and is currently the largest known prime number. Other places to read about the new Mersenne
Last fiddled with by Prime Monster on 2004-05-29 at 21:32 |
2003-11-30, 23:21 | #4 |
Aug 2002
260_{10} Posts |
What is Distributed Computing?
Distributed Computing (DC) is a branch of Computer Science that tries to solve large problems by giving small parts of the problem to many computers to solve and then combining the solutions for the parts into a solution for the complete problem. Distributed Computing projects have been designed to use the computers of hundreds of thousands of volunteers all over the world, via the Internet, to look for extra-terrestrial radio signals, to look for prime numbers so large that they have more than ten million digits, and to find more effective drugs to fight the AIDS virus. These projects are so large, and require so much computing power to solve, that they would be impossible for any one person, computer or even a large corporation to solve in a lifetime. Also see the post More on Distributed Computing Last fiddled with by Prime Monster on 2003-12-01 at 19:24 |
2003-11-30, 23:22 | #5 |
Aug 2002
2^{2}·5·13 Posts |
A little bit of history
The study of Mersenne primes has been central to number theory since they were first discussed by Euclid in 350 BC. The man whose name they now bear, the French monk Marin Mersenne (1588-1648), made a famous prediction about which values of "P" would yield a prime. It took 300 years and many important discoveries in mathematics to prove his conjecture. Prime numbers have long fascinated amateur and professional mathematicians alike. In the past fifty years a number of individuals wrote software and developed databases to aid in the search for Mersenne primes. Slowinski of Cray Research comes to mind among several others. But it wasn't until late 1995, when George Woltman gathered up the disparate databases and combined them into one, it really took off. In early 1996 George placed this database, and a free and highly optimised program to search for Mersenne primes, on the Internet to coordinate the search; the start of the Great Internet Mersenne Prime Search. The first version of GIMPS was based on using manual reservation of exponents and manual reporting of results. You can still use this method today, but most participant use the automatic method that was introduced with PrimeNet. PrimeNet was established in late 1997 by Scott Kurowski (and others) to automate the selection of ranges and reporting of results for GIMPS. When the project began, there were only 40 people and a little over 50 computers using their spare CPU cycles; the project has since grown to thousands of users and computers. The software has also changed quite a bit from GIMPS's beginnings; The factoring part of the program was originally written for 386 computers, in its current incarnation it detects the processor type and, if possible, uses hand-coded support for the P4's SSE2 codeset. Another aspect of the software, which probably influenced how the project is set up and managed, is that at first, there were many errors in checking for the primes, and all discovered Mersenne primes had to be double-checked to ensure their validity. Even today every result, whether a prime or not, is double-checked, as errors are still occurring. Current estimates are that the error-rate is actually increasing slightly. GIMPS is a unique distributed computing project for many reasons. George Woltman publishes the GIMPS software source code, and welcomes anyone to revise the software or make enhancements, as long as they follow the GIMPS rules. And, although the source code has been fully released, there doesn't seem to be any threat of data contamination that is present in some other distributed computing projects. If the results indicate a prime number, other users could check the same exponent and verify that it is a Mersenne prime, and given the prize and fame a discoverer gets, it seems unlikely that someone who found a Mersenne prime would keep the discovery hidden. Last fiddled with by Prime Monster on 2003-12-01 at 19:29 |
2003-11-30, 23:23 | #6 |
Aug 2002
2^{2}·5·13 Posts |
A tiny bit of theory
An integer greater than one is called a prime number if its only divisors are one and itself. The first prime numbers are 2, 3, 5, 7, 11, etc. A Mersenne prime is a prime of the form 2^{P}-1. The first Mersenne primes are 3, 7, 31, 127, etc. In 1644 Mersenne claimed that n is prime if P = 2, 3, 5, 7, 13, 17, 19, 31, 67, 127 and 257 but composite for the other 44 primes smaller than 257. Over the years it has been found that Mersenne was wrong about 5 of the primes less than or equal to 257 (he claimed two that did not lead to a prime (67 and 257) and missed 3 that did: 61, 89, 107). The Mersenne primes are not just pretty numbers, though: they do have many applications. The most common uses for these primes are for forming algorithms, and, interestingly enough, testing hardware (Check towards the end of the page). To show that a small number is prime, we can just divide by the primes below its square root. For example, 53 is prime because it is not divisible 2, 3, 5 or 7. Large numbers need more complicated tests that usually involve a form of LaGrange's theorem. If you look at a list of the largest known primes you will see that for most such numbers n, either n+1 or n-1 factors easily. Mersenne primes follow this rule and are the primes of the form 2^{P}-1. It is easy to show that p here must also be a prime, (at least that is what the mathematicians tell us). Because of the special form of the Mersennes, there is an easy test to see if they are prime, called the Lucas-Lehmer test. Last fiddled with by Prime Monster on 2003-12-01 at 01:22 |
2003-11-30, 23:24 | #7 |
Aug 2002
2^{2}·5·13 Posts |
Some words about the Work and the Client
Prime95 is a free program written by George Woltman and provided on his GIMPS site. Just download the software, install it, and let it run. Simple and easy, but the road is long. GIMPS is not a project for the faint at heart, most of the exponents (work units), depending on type of work you choose to do, are huge! Over 86% of those who have begun to test a Mersenne exponent with GIMPS have never completed even one, and the average GIMPS participant has completed less than three (mean 2.67). GIMPS appeals to those with a lot of patience, the work units take a lot of crunching, but pack serious punch. Very friendly to offline operation and P4's, and runs well on a wide range of x86 machines. Finding a Mersenne prime is no easy task; a first time primality test for a single candidate exponent takes a fast machine three to four weeks, running 24/7. On the fastest Tbirds, PIII's, and P4's, prime95 performs better, due to the SSE2 optimizations in the code. This is without a doubt one of most process intensive project in all of computing. A test runs through three phases before the result is considered final, done gradually and by several different machines. First, each candidate exponent goes through a level of trial factoring, which is a brute force approach designed to eliminate non - prime candidates early on in the process. This works by checking the number for small divisors, using the fact that factors of Mersenne numbers are of the special form q = 2kp+1, and that testing a factor candidate simply requires a binary powering, i.e. one finds 2^{p} modulo q (which involves roughly log_{2}(p) arithmetic steps on numbers no larger than q^{2}) and checks if the result is equivalent to -1 modulo q. One can also do a small amount of factoring work using (say) Pollard's p-1 method or the elliptic curve factorizations method, both of which involve manipulating numbers the size of the Mersenne number itself. If no factors are found then the exponent is released for a first time Lucas - Lehmer primality test. A final double-check is done later by another computer to verify the validity of the result. This project has it advantages, namely the client itself. This has to be the most stable, mature, feature rich DC client on the net. It will use as much or as little memory as you allow it, and will get completely out of the way whenever another program needs cycles. It can be run as a service, completely hidden from view, with the ability to run when the user logs off so no cycles are wasted. Client Highlights[list=1][*]Mature client[*]Easily runs when logged off. Just check Run at Bootup. You need not know about Windows services![*]Caches work automatically, so it keeps going if offline. No helper programs are needed to cache work for it.[*]Low memory needs so PCs with modest RAM can use it. If you are short on RAM you can still run it. (2% of the time it wants more RAM for P-1 and the user can set limit.)[*]Low bandwidth needs: <~1 KB / month. Great for dial-up![*]Stringent verification within each run and across runs, so you know your CPU cycles are not being wasted.[*]GIMPS offers something that many other distributed projects don't: actual regular discoveries[*]A very responsive developer, George Woltman, who frequents this board as Prime95[/list=1] Last fiddled with by Prime Monster on 2003-12-01 at 21:58 |
2003-11-30, 23:24 | #8 |
Aug 2002
2^{2}×5×13 Posts |
Monetary Awards
As if having the best Distributed Computing client on the net is not a big enough incentive to run GIMPS, there is also a substantial award for the person that discovers a ten million digit prime number. The Electronic Frontier Foundation is offering a $100,000 award, see the EFF Cooperative Computing Awards, to the first person or group to discover a ten million digit prime number. If you find such a prime with the software provided, GIMPS will claim the award and distribute the award according to the rules published on the GIMPS site. A Word of Warning!!! If you decide to install this client, or any other DC client, on your employers machines, always, always get written permission before installing idle-cycle software on any machines you do not have absolute administrative control over. Many companies have a strict policy against running any non-business software. You will not get a lot of sympathy from the DC community if you installed a DC client on a lot of machines and then got into trouble because of that. In some countries it is actually a criminal offence. Last fiddled with by ewmayer on 2006-06-04 at 19:16 Reason: updated EFF co-op awars link |
2003-11-30, 23:25 | #9 |
Aug 2002
404_{8} Posts |
The type of work you can do
You have the choice of 4 different types of work units you can do for the project. The client can also do other types of work as well, but they are outside the scope of this page (for more information see the readme.txt that comes with the client and the Other Projects section on this board). [list=1][*]Factoring of a prime number (exponent) candidate This is a pre check of a candidate exponent. These work units take the least time and are given out by the server according to your CPU speed. They may go through several iterations of bit length before clearing and getting sent to a fast machine for the more intensive Lucas-Lehmer testing. This is best for low-end PIIIs and Athlons[*]Double checking a prime candidate Double checking is done to work units which have cleared one round of LL testing. This verifies that the LL test has concluded correctly and ensures that the CPU of first running conformed to spec. These are small to medium units, PIII, Athlons and low end P4s turf here. Basically this is a Lucas Lehmer test, but the exponent size is smaller than for the current first time primality test range. [*]Lucas Lehmer (LL) testing on a prime candidate These candidates are much bigger than the run in the double testing pool. It is a virgin run; the chance of finding a prime candidate is about 1 in 100 000. Pre factoring ensures that no time is spent on lengthily LL work when a simple factor is present.[*]10.000.000 + digit Mersenne prime LL testing This is the grand daddy of all DC WU - only the diehard crunching fanatic need apply. Depending on the size and your box, several months to a year (and even more) of CPU time is needed. Same as above except for size and one other thing: the prize. Find one of these and you win a substantial prize, something like over 50.000$ US$ - enough to cover a good bit of D. Find the first one and you will live forever in mathematics history.[/list=1] Last fiddled with by Prime Monster on 2003-12-01 at 19:30 |
2003-11-30, 23:25 | #10 |
Aug 2002
2^{2}·5·13 Posts |
How to get started with GIMPS
Step 1 Download the appropriate program for your OS.
Step 2 If you downloaded p95v237.exe then run it. Ghost Installer has generously made their setup program free for all. Thus, you can run the prime95 setup program and simply follow the instructions to install Prime95. All other platforms will need to create a directory and decompress the file you just downloaded. Windows users can choose from a variety of decompression programs. Some are PKZip, WinZip, or dozens of others at Tucows. Linux and FreeBSD users should use the standard tar and gzip decompression utilities. Step 3 Read the readme.txt file, that came with the client, for more detailed instructions. Generally, joining GIMPS is as simple as running the program, answering a few questions, and the program does the rest. It contacts a central server to get some work to do. There are several ways to set up the program to run every time you restart your computer. You can track your progress on a stats page the server updates every hour. Step 4 Questions and problems. Consult the readme.txt file for possible answers. You can also search for an answer, or ask for help in the GIMPS forums. Otherwise, you will need to address your question to one of the two people who wrote the program. Networking and server problems should be sent to Scott Kurowski, but please consult his FAQ first. Such problems include errors contacting the server, problems with assignments or userids, and errors on the server's statistics page. All other problems and questions should be sent to George Woltman, but please consult his FAQ first. There are also a few other FAQs available, see A handful of GIMPS Links Sneakernetting the client Instructions on How to Sneakernet the client are in a separate post. Statistics WU credit is in based on P90 CPU hours. If you have a faster box you get more P90 hours. Simple and unchanging. The stats server is constantly updated - turn in a WU and get instant credit. You can find the stats here. Note : It is important to show a bit of patience if you get any problems with registering or requesting work. The main server, www.mersenne.org, has had some intermittent problems in the past. We hope that they are resolved now, but if you do get problems, please check with the forum for the current status. The top left hand corner smilies, on the forum home page, shows the current server status. Last fiddled with by Prime Monster on 2003-12-01 at 19:31 |
2003-11-30, 23:26 | #11 |
Aug 2002
2^{2}·5·13 Posts |
How to set up a Team
The GIMPS project does not really support teams like other Distributed Computing projects. This does not mean that it is impossible to have teams, it is. GIMPS uses a nice little trick to get teams to work, and that is to use the User ID field for the team name. So, if you want to set up a team for you and your friends, you will have to specify this when registering from one of your clients. Tick the CheckBox to create a team. Note: You will only have to do this once. Name and e-mail address remains individual for each computer. Anyone that wants to be a member of your team must then use the same User ID (Team name) and password. It is quite smart to use different Computer IDs for each computer, so it is easier to identify the individual member. Each computer gets its individual tasks and can individually be set to special task types under "Test, PrimeNet". Besides possible psychological effects of competition, building a team has no impact on how soon a task is done. Nevertheless, building or joining a team can be fun, as you get in contact with other people, can find friends and equal-minded people that share the same interests as you. Last fiddled with by Prime Monster on 2003-12-01 at 19:32 |
Thread Tools | |
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
Where should I start? | christian_ | Information & Answers | 9 | 2016-01-22 19:28 |
Before you start... | jasonp | Operation Kibibit | 65 | 2013-09-03 22:06 |
How to start? | Thomas11 | Lone Mersenne Hunters | 29 | 2008-12-21 13:47 |
how to start with P-1? | ValerieVonck | Marin's Mersenne-aries | 8 | 2006-04-29 22:21 |
How to start? | OmbooHankvald | Factoring | 15 | 2005-09-03 13:42 |