News & Print

This page is the place for you to find recent news stories regarding Linux clustering, as well as normal "print" stories -- journal and magazine article and such. We also have a list of software and project updates and releases. If you read an article, see a press release or a news story, that isn't here, please feel free to e-mail us the URL and we'll try to get it up in short order.

• SGI has announced that they are building an Altix based supercomputer for the National Center for Supercomputing Applications. It will be running Linux, and be comprised of 1,024 Itanium 2 processors. The big news here is that it will be a shared memory system, which is different than the other Linux clusters NCSA currently has. More information can be found in this Computerworld article, or the SGI press release. (July 20^th, 2004)
  
  
• Linuxworld has a story about Thunder, the newest, and largest, Linux cluster out there, which was recently deployed at Lawrence Livermore National Laboratory. Thunder consists of 4096 Itanium 2 processors in 1024 systems, with Quadrics as the high speed interconnect, and can run at a sustained 19.94 TFlops. (May 14^th, 2004) Update: Here is a link to a press release which contains more technical information about the cluster.
  
  
• C|Net has a story titled Makers of white-box supercomputers hit their stride, talking a bit about LLNL's Thunder cluster, as well as other Linux clusters, and some of the makers and reasons behind Linux clusters and their popularity. (May 14^th, 2004)
  
  
• HP and Dreamworks have been collaborating on new technologies for off-site animation rendering, and is being used in movies such as "Shrek 2," "Shark Tale," and "Madagascar." They are using a 1,000 processor compute farm, linked via a high speed network to Dreamworks' animation studios. For more information, check out the press release. (April 20^th, 2004)
  
  
• Linux Networx has announced the sale of a 556 processor cluster to Japan's National Institute of Advanced Industrial Science and Technology (AIST). The new cluster will join with other systems AIST has to form Japan's largest distributed computing grid. For more information, check out the press release. (April 7^th, 2004)
  
  
• Well, as expected, when you wade into the FUD game, you will likely find that the conspiracies can overshadow the truth. In a followup to the story we posted yesterday (see below entry), we were reading Groklaw, and saw this posting, which points out that the letters sent out to the national labs that are referred to in the story were just attempting to confirm the receipt of letters sent out in the fall. Draw your own conclusions. This is why we try to stick mainly to technical things here. :) (March 21^st, 2004)
  
  
• We don't like to get into the mudslinging that goes on with the various FUD wars that sometimes affect Linux, but we felt we should mention this particular development. News.com is reporting that SCO has sent letters to both Lawrence Livermore National Laboratory and the National Energy Research Scientific Computing Center (NERSC), about their Linux use, attempting to get the facilities to buy licenses that would allow them to operate without fear of legal action by SCO. We will leave the editorializing as an exercise for the reader, but will just say that LLNL has a couple of Linux clusters already, including the #7 system on last fall's Top 500 list, and is the future home of IBM's Blue Gene/L system, expected to be the world's fastest system when it is up and running, and NERSC has a 400+ node Linux cluster in place, as well as having worked closely with IBM on many systems over the years. (March 20^th, 2004)
  
  
• The Grid Wars Challenge will be held at the Cluster World Conference and Expo again this year. The grand prize is an HP cluster with one master and four slave nodes. For more details on the contest and the prize, check out the announcement. (March 4^th, 2004)
  
  
• There was a story posted on Slashdot about a guy who built a cluster using Mini-ITX systems. He used FreeBSD and MPICH, so it's not exactly Linux clustering, but it has some nice pictures and a good walkthrough of the process he took to get it built, up, and running. The story can be seen here. (February 26^th, 2004)
  
  
• Cray has announced that they are acquiring OctigaBay. OctigaBay develops HPC systems, and is based in Vancouver, British Columbia. More information can be found in the press releases from both Cray and OctigaBay. (February 25^th, 2004) Update Feb 26th, The Register has a story about the acquisition, which mentions specifically that OctigaBay specializes in Opteron solutions, which is interesting in light of last year's announcement by Cray of a new line of Opteron based clusters, including the Red Storm cluster sold to Sandia, a system of 16,000 Opterons.
  
  
• IBM has announced that it has built the first of the Blue Gene computers, which is about the size of a television. The Blue Gene/L test prototype came in at number 73 on the Top 500 list released Sunday, with a best run of 1.4 TFLops. It contains about 1000 PPC microprocessors. For more information, there is a news story on the CNN/Money website. (November 19^th, 2003)
  
  
• Supercomputing Announcements and News:
  
  • SGI to Showcase 128 Processor Altix System at SC2003
  • Lawrence Livermore National Laboratory Selects Intel Itanium 2 Processors for World's Most Powerful Linux Cluster (approximately 4000 I2 processors, to be complete in January)
  • IBM Supercomputers are 55 percent of top 100
  • Linux Networx Announces Two Clusters Ranked Within Top 10 Fastest Supercomputers
  • Dell claims number 4 cluster in Top 500, as well as increasing its total entries to 18
  • NASA Ames Installs First 512-Processor SGI Altix-Linux Supercomputer
  • Paracel Introduces the Paracel Cyclone -- a Comprehensive Linux Cluster Solution
  • Mountain View Data announces PowerCockpit optimized for AMD Opteron processors
  (November 19^th, 2003)
  
  
• The 22nd edition of the Top500 List has been released. The new list can be found here. The top system is still the Earth Simulator, number two is still ASCI Q, but after that it gets a little more interesting. Number 3 is the Virginia Tech G5 cluster, coming in at 10.28 TFlops, number 4 is NCSA's cluster of 2500 3 GHz P4 processors from Dell, running at 9.8 TFlops, and number 5 is PNNL's Itanium 2 cluster, with 1900 1.5 GHz I2 processors from HP, weighing in at 8.6 TFlops. Overall, there are seven clusters in the top 10, with a total of 208 in the entire list, up from 149 on the last list 6 months ago. The Top500 press release, with some more interesting facts and information, can be found here. (November 16^th, 2003)
  
  
• Aspen Systems and Dot Hill will be exhibiting HPC systems and Dot Hill's SANnet II storage systems at Supercomputing 2003 in Phoenix next week. For more information, check out the press release from Aspen Systems. (November 11^th, 2003)
  
  
• The 1100 node Macintosh G5 cluster at Virginia Tech, benchmarking at 10.28 Teraflops, is expected to rank as the third fastest cluster in the world when the Top 500 list is announced next week at Supercomputing 2003 in Phoenix. Take a look at this news story for some more information. (November 11^th, 2003)
  
  
• Aspen Systems announced today a distribution partnership with Mandraksoft to sell and support clusters with the MandrakeClustering distribution in the United States. For more information, check out the press release here. (November 10^th, 2003)
  
  
• G5 Cluster at Virginia Tech has reached 10.28 TFlops. Story at the Register. (11/4/03)
  
  
• IBM Offering Opteron and Xeon systems side by side in clusters. Story at the Register. (11/4/03)
  
  
• Cray is offering a new line of clusters based on the AMD Opteron processor. Story at the Register. (10/27/03)
  
  
• IBM's upcoming Blue Gene line of supercomputers will be running Linux. The first system, Blue Gene/L, will contain 65,000 processors. Story at ZDNet. (10/24/03)
  
  
• Sandia National Laboratories has ordered a 16,000 Opteron processor cluster from Cray. Story at the Register. (10/21/03)
  
  
• Spb Software House assembles a cluster of 12 pocket PCs. Check out the story here (10/8/03)
  
  

• Here is the most current letter (Nov 11, 2002).
  
  
• Here is previous letter (Oct 29, 2002).
  
  

• O'Reilly Linux DevCenter has an article titled Introducing openMosix, which is a beginner's introduction to openMosix, what it can do, how to use it, and links to some other openMosix information. (February 25^th, 2004)
  
  
• Linux Journal has an article entitled The Role of Linux in Grid Computing. It talks a bit about what grid computing is, and how Linux can fit into this view of computing. (February 1^st, 2004)
  
  
• Columnist Robert Cringely has written a column titled Why Heat is the Enemy of Server Farms and Sometimes Less Power Means More. It is, predictably, about how heat affects a cluster, and what is being done to deal with what happens when you build bigger clusters that generate more heat. (November 19^th, 2003)
  
  
• NewsForge has a story looking into how Formula 1 racing is using a Linux cluster to build better and safer cars. The story can be seen here. (November 19^th, 2003)
  
  
• As we mentioned earlier, the June edition of Linux Magazine is mainly about clustering. Some of the stories are Beowulf Breakthroughs, The genesis of Linux Clusters in high performance computing, by Dr. Thomas Sterling, Beowulf Infrastructure, How to care for and feed your Beowulf-style cluster, by Dr. Robert Brown, and Coupling Parallel Models, A hands-on guide to some complicated coding, by Forrest Hoffman. The complete table of contents can be found here. (September 15^th, 2003)
  
  
• There is a story on NewsForge titled Is a Linux Supercomputer in Your Future. It discusses clustering working its way into more and more places. (September 9^th, 2003)
  
  
• Linux Magazine's Extreme Linux column for April and May are online. April's is titled Using Derived Data Types with MPI, and May's column is MPI Communicators and Groups. (September 9^th, 2003)
  
  
• The June issue of Linux Magazine has special focus on Beowulf computing, with this year marking the 10 year anniversary of the original Beowulf cluster created by Don Becker and Tom Sterling. An example of a few of the articles are Beowulf Breakthroughs: The genesis of Linux Clusters in high performance computing by Tom Sterling, or Beowulf Infrastructure: How to care for and feed your Beowulf-style cluster by Robert Brown. Or you can check out the issue's Table of Contents. (June 20^th, 2003)
  
  
• The March edition of Linux Magazine's Extreme Linux column has been posted to the web. It is titled Master/Slave Message Passing. (May 31^st, 2003)