data link for October 1999

Memory Placement Monitor Released: The Memory Placement Monitor (MPM) is a new graphical tool that lets you observe the placement of an application's memory pages on the nodes of an SGI Origin as the application runs. MPM works on both UNIX and Windows 9x/NT platforms. Read more about it, then check the complete information available online. §
Can I Do That? Scientific Visualization Explained.: NCSA's Dave Bock, a member of the Visualization and Virtual Environments team, is an accomplished professional with countless visualization collaborations to his credit. Learn about two new tools he developed recently and see if you want to delve into visualization yourself. §
Results of User Survey Announced: Earlier this year, a user survey was made available to users to encourage feedback on computing environments and support provided by the Alliance. Your comments and suggestions have helped shape decisions throughout the summer. Take a look at this summary of the input provided to NCSA. §
NCSA's First Million-Hour Month: A National Science Foundation-supported high-performance computer delivered over 1 million normalized CPU hours in one month this August. NCSA's 1,536-processor SGI Origin2000 supercomputer provided the hours to 736 national users, triple the usage from August 1998. §
NCSA NT Supercluster Goes to Production Status: The plans to move the NTSC to production status have been indefinitely postponed.
NCSA Batch Changes Coming to Origin2000: Automated, guaranteed saving of output files from batch jobs to the mass storage system. Sound good? Want to know more? Check this documentation page that explains the changes you need to make to your batch scripts and about the other related changes. §
Running Multiple Jobs in the Dedicated Queues on NCSA's Origin2000: NCSA users whose codes do not scale to the number of processors available in the dedicated queues (128 or 256) can still use the dedicated queues to run multiple jobs. The dplace tool helps obtain optimal performance by initializing the processes and memory of a program on specified nodes, helping to eliminate the potential for poor performance resulting from multiple threads executing on the same processor. §
Deep Blue RS/6000 Installed at Maui HPCC: The IBM RS/6000 SP located at the Maui High Performance Computing Center (MHPCC) will be twice as powerful following an upgrade scheduled for late October. The upgrade will add 50 nodes of IBM's Symmetric Multiprocessing (SMP) POWER3 SP technology to MHPCC's supercomputing suite, providing 178 Gigaflops of additional peak computing power. The POWER3 nodes will be combined with MHPCC's existing IBM POWER2 Super Chip nodes to create a single system that will offer approximately 300 Gigaflops of high-performance computational capability. §
NSF PI Eligibility Statement Updated: NSF recently updated their definitions of eligible PIs under the CISE directorate. Read the newest information to make sure your application can be considered. Also check the NSF Grant Proposal Guide, a PDF document linked in on this page that covers everything you need to know about this important funding agency's grant requirements. §
Rocket Simulation Basis of Hardware Comparison: A rocket simulation provided the computational basis for a recent comparison between NCSA's balder machine (a 256-node system within the center's Origin2000 array), a T3E, an SP2, and the ASCI Red machine on the same ZEUS-MP 3-D radiation hydrodynamics problem. Parallelism was achieved with MPI. Take a look at the results. §
NLANR Packets Available: Issue #2 of NLANR Packets offers more than ten links to high-speed networking information for you to bookmark. Tools, system upgrades, proceedings, and other web-baed information await you at the collaborative site. §
The Internet as a Model: Computer Networking: A Top-Down Approach Featuring the Internet is a new online book by James F. Kurose, University of Massachusetts, and Keith W. Ross, Institute Eurécom (France). The authors start at the application layer and work down the protocol stack. With a special emphasis on Internet protocols, the text also includes hyperlinks and java applets. See what you think of an online textbook -- it could be the wave of the future. §
Workshop on Graph Partitioning & Applications: Graph partitioning is an important problem with extensive applications in many different areas including scientific computing, parallel processing, VLSI design, data-mining, and efficient storage of large databases. If you are interested in joining colleagues in Minneapolis who will be discussing graph partitioning on October 14 and 15, contact conference coordinator Jean Burdick for details. §
BBC Online Highlights Access Grid: The recent Chautauquas featured the Alliance Access Grid. Read the BBC's take on this latest innovation. §
Shorter URL for data link!: To save you some typing, the friendly NCSAers who manage the center's web servers have made possible a shortened URL for this newsletter. Simply enter www.ncsa.uiuc.edu/datalink and you will be on our home page! data link thanks the webadmin team. §