Kirk Pearson's Weblog

World's largest computing grid now has over 100 Sites According to a PhysOrg.com article, the Large Hadron Collider Computing Grid (LCG) project is now "the world's largest international scientific computing grid," with over 100 sites in 31 countries. The grid, containing "more than 10,000 central processor units (CPUs) and a total of nearly 10 million Gigabytes of storage capacity on disk and tape" will process data from the Large Hadron Collider (LHC), a particle accelerator which will begin operating in 2007. The current status of the LCG project (and a map of participating sites) can be seen here. Even with the grid's record-breaking size, it only has 5% of the computing capacity that the LHC will eventually need. (2005-03-15 13:13:00.0) Permalink Comments [7]

grid.org's Smallpox Project Completed

grid.org and the University of Oxford have completed their distributed computing project to find a drug to fight Smallpox, a viral disease which could be used as a bioterror weapon and for which currently there is no cure. The project began on February 5, 2003. Phase 1 of the project was completed on September 30, 2003. That phase used 39,000 years of computing time from volunteers in 190 countries to screen "35 million potential drug molecules against eight models of the smallpox protein" to determine if any of them would prevent the virus from replicating. Potential drug candidates from this project will be studied more closely in another phase of research. No news has been posted about the results of the complete project yet, but the project's news page should be updated soon. The results from this project may also benefit grid.org's project to find cancer-fighting drugs.

So what disease is your computer trying to cure? If it's not, pick a project and get started.

distributed.net completes its OGR-24 project

distributed.net successfully completed its OGR-24 distributed computing project on November 1, 2004. This project, which began over 4 years ago, used a brute force method to find the shortest, or optimal, 24-mark Golomb ruler, a ruler in which no two marks are the same distance apart. The project tested over 555,529,785,505,835,800 rulers (twice, to verify its results), and determined that the previously best-known ruler, 24/9-24-4-1-59-25-7-11-2-10-39-14-3-44-26-8-40-6-21-15-16-19-22, is optimal. Almost 42,000 volunteers participated in the project.

distributed.net is currently running an OGR-25 project, which began over 4 years ago and is over 65% complete, and a RC5-72 project (an attempt to solve the 72-bit RSA Labs secret-key challenge), which began over 2 years ago and is 0.16% complete.

Help Design CERN's Large Hadron Collider

Coinciding with CERN's 50th anniversary today, the LHC@home distributed computing project has opened to public participation. This project will help CERN design a stable particle beam for its Large Hadron Collider (LHC) particle accelerator. The project uses a software client, based on the BOINC distributed computing platform, to simulate the path of a group of particles around the LHC ring. The client is available for Linux and Windows now, and will be available on other platforms in the future. Since the public will not be able to help analyze the 15 Petabytes of data the LHC will generate each year, LHC@home is a great way for people to participate in the LHC project.

Sun grid computing will be used to design next generation of nuclear reactors

Nothing illustrates Sun's phrase "The Network Is The Computer" better than grid computing, in which a networked "power grid" of computers acts as a single computer to solve a single, large problem. In the past few years we have read and heard much about the promise of grid computing, but we haven't seen many actual applications of it. I just found a news story, posted today on Yahoo!, which describes an impressive practical application. Sun and the U.S. Department of Energy are creating a computing environment with Sun hardware and Sun's N1 Grid Engine software for the Idaho National Engineering and Environmental Laboratory to use in designing the Department of Energy's Generation IV nuclear reactors. The 230 Sun Fire(TM) V20z servers in the grid will produce 2 TeraFLOPs/sec (2 trillion floating-point operations per second) of computing power, more than 7 times the power of the INEEL's existing, expensive, custom mainframe computers, and enough to rank "the INEEL datacenter as one of the world's top 150 supercomputing sites."

As a comparison of CPU power, one Intel Pentium 4 2.2 GHz CPU produces about 4400 MegaFLOPs/sec (or 4.4 GigaFLOPs/sec or 0.0044 TeraFLOPs/sec), and SETI@home's network of distributed computers is currently producing 58.8 TeraFLOPs/sec.

Print a 3D model of your climateprediction.net simulation

If you are participating in the climateprediction.net global climate simulation project, and if you are a Windows user with Photoshop, you can now download a Photoshop plug-in to make a 3D globe of your simulation results.

Instead of worrying about a possible future runaway greenhouse effect or the next ice age, you can decorate your office or Christmas tree with it.

Just one page a day

What do you do while you're listening to a conference call, or waiting for an application to compile, or waiting for a document to print? Twiddle your thumbs? Surf the web for the latest depressing financial news and gossip on celebrities? Why not use that wasted time to contribute to the world's first online library of free electronic texts, Project Gutenberg?

Project Gutenberg was begun in 1971 by Michael Hart, with the goal of creating a library of 10,000 electronic books which would be free and downloadable to anyone with a computer. The library contains books that have entered the public domain after their copyrights expired (these books were typically published before 1925). Until a few years ago, each electronic book was created by one person, who laboriously scanned a paper copy of the book, one page at a time, on an electronic scanner, then converted each scanned image into ASCII text, then assembled all the pieces of text into one file and proofread and formatted the file. I have contributed five books to Project Gutenberg using this technique.

In 2000, Charles Franks created the Distributed Proofreaders project to help speed up the process of creating electronic books. In this project, volunteer project managers scan paper books to create a set of images for each book, and use Optical Character Recognition (OCR) software to convert each image to a page of ASCII text. Then a project manager uploads the images and text pages to the Disibuted Proofreaders server. Next, volunteer proofreaders view a list of projects at the website and choose one to proofread. Then a simple proofreading interface is displayed and a volunteer can proofread pages of the project, comparing the OCRed text to the image of the page, and making any corrections necessary to the text. After a book is completed, a post-processor assembles all of the text pages into the final e-text and submits it to Project Gutenberg, where it is sent to ftp archives all over the world. This project has made it possible to create an electronic book in hours instead of weeks.

With the help of hundreds of volunteers, the project now proofreads over 6,000 pages (twenty-four 250-page books) every day. On September 3, 2003, the project completed its 2,000th book: soon after, on October 15, 2003, it helped Project Gutenberg reach its first goal of publishing its 10,000th book. As of today Distributed Proofreaders has completed 4,841 books.

So how can you help? Sign up, choose a book that interests you, and proofread just one page a day. Depending on the difficulty of the book, this can take 5 to 30 minutes. You can proofread a sentence while you're waiting for another web page to load, a paragraph while you're waiting for your document to print, or a page while you're waiting for someone else to finish talking on a conference call. If 100 people proofread just one page a day, they will create a new electronic book in just two to three days. You don't need any special tools or skills to proofread. The project site contains detailed proofreading guildelines and tutorials, proofreading mentors, and very active discussion groups.

If you're fluent in another language besides English, you can also help proofread books in other languages in the Distributed Proofreaders Europe project. This project enables volunteers to proofread books in any Unicode-supported script.

Have you BOINCed today?

No, not boinked, but BOINCed. BOINC is the Berkeley Open Infrastructure for Network Computing, the next-generation distributed computing platform. It allows distributed computing organizations to develop and manage client applications on a common distributed computing platform. It also allows project participants to participate in multiple projects without having to download a client application for each project. The BOINC core client is available for Windows, Linux, UNIX, and Mac OSX, and project client applications are available for some or all of those platforms.

SETI@home is the first distributed computing project to use BOINC. It began its BOINC-based project to search for alien radio signals on June 22, 2004. The project has had a hard time generating enough work for all of the participants who want it, but the project server and website are becoming more stable. It has a Solaris client.

Predictor@home is the second project to use BOINC: it is trying to predict protein structures from protein sequences. The project client is still in development, and is not available for Solaris, but you can participate in the project with the other platforms and you can contribute useful work.

Einstein@Home and climateprediction.net are also developing BOINC-based clients, although it doesn't look like they plan to develop Solaris clients.

If your CPU is idle, join one of these projects and start BOINCing!

I'm #1 in search results for "distributed computing" at google.com, anyway. Sometime in the last few days my distributed computing website took over first place in search results for "distributed computing" at Google. And it took me only 5 years to get there :-) Woohoo! (2004-07-20 11:44:50.0) Permalink Comments [2]

What does your workstation/server/PC do in its spare time?

Six years ago I learned about the distributed.net project and became an active participant in its RC5-64 encryption cracking contest. A year after that I was completely hooked on the idea of distributed computing projects and I created a website to track active and upcoming public distributed computing projects (those in which anyone can participate) and to get others involved with them. The site has grown from tracking distributed.net, the SETI@home project to search for alien radio signals, and the GIMPS project to find world-record large Mersenne prime numbers, to tracking over 40 projects in many areas of science, biological sciences, and mathematics.

Since I created my website, it has grown to a viewership of 24,000 people per month and a mailing list of over 200 people who subscribe to website updates, and I have received lots of emails from people all over the world who have said my site got them interested in, and participating in, one or more distributed computing projects. But in the same time I have not gotten a single coworker, friend or relative interested in distributed computing enough to join a project and stay active in it for more than a few days. Why is this? Why is it so hard to convince the people around me of the benefits of using spare CPU cycles for distributed computing, and to participate in a project?

Modern CPUs are so fast that they have nothing to do most of the time. Your CPU is wasting millions of computing cycles doing nothing while you read this blog entry. It wastes billions more cycles while you view web pages, read emails, create documents, go out to lunch, or go home at night. With about 5 minutes of work you can install a project and use all of those spare cycles. While I am writing this message, one of my CPUs is looking for Optimal Golomb Rulers, the other is looking for a factor of 3^491 + 1, a number for which no factors are known, the idle CPU on my lab system is simulating protein folding, and the CPU on my PC at home is predicting the weather 50 years from now. Another Sun employee, lemonsqzz in the Distributed Folding project, is using enough spare CPU cycles to be the 6th-highest participating member in that project.

What is your CPU doing right now? Does it run a screensaver when you're not using it? If it's wasting computing cycles, why not use those cycles to help find a cure for cancer, design better hardware circuits, or contribute to other scientific or mathematics research?