GOVERNMENT

For the third consecutive year, HP is the revenue leader in the $9 billion high-performance computing (HPC) market. HP took the No. 1 position with more than 31 percent market share, according to 2005 figures released by research firm IDC. Fueled by the success of its Unified Cluster Portfolio, HP produced industry-leading revenues across the HPC market segments of workgroup, departmental and enterprise as well as in overall clustered systems. The market for technical clusters has been growing at more than 70 percent for the past four years. To learn more, we asked Ty Rabe, director of research and development for Hewlett-Packard’s HPC business, to share some insights. SC Online: What's driving sales growth today? Director Rabe: Dramatic reductions in the cost of supercomputing are fueling big increases in demand. Just a few years ago (circa 2001), one TF had a street price in the $5-10 million range. Today, the price is $500 thousand to $1 million per TF and still dropping. So everyone in the market is getting a lot more for their money, and plenty of new entrants see the value of HPC for their mission—whether it’s academic research, designing widgets, or predicting the value of a portfolio. SC Online: Why do you think your company has been so successful? Director Rabe: HP has remained successful during the last few years of rapid change, because we’ve been driving the emerging HPC technology—industry standard clusters--rather than simply protecting our traditional business. Today we offer tremendous choice, but we also have the expertise to help our customers make the best decision about how to solve their HPC challenges. And, of course, HP delivers great quality and customer support. SC Online: Do you see a bright future for Itanium? Director Rabe: I do. Itanium consistently delivers top performance in many HPC applications and the next generation (Montecito) Itanium processors will enhance that advantage in the second half of 2006. Longer term, I expect a constant shift from year to year in which industry standard processor architecture—Itanium, Opteron, or Xeon—has the performance lead. That competition will keep driving performance up, keep reducing the cost per TF, and keep fueling rapid growth in the market. Supercomputing Online wishes to thank Director Rabe for his time and viewpoints. Today, HP also announced that in the fastest growing segment -- technical combined workgroup and departmental systems -- HP has the top position with 30 percent market share. These systems now account for more than 55 percent of the HPC market, up from 37 percent in 2002. IDC defines workgroup systems as those costing less than $50,000 and departmental systems as those costing between $50,000 and $250,000. HP also holds the No. 1 position in the enterprise segment, which includes through-put oriented systems costing more than $1 million, with a 47 percent revenue share. In addition, HP is No. 1 in technical cluster revenues with a 31 percent market share. "HP is delivering affordable, easy-to-use industry standard-based supercomputing solutions to an increasingly broad range of research teams in diverse application areas around the globe," said Winston Prather, vice president and general manager, High Performance Computing Division, HP. "The expansion of HPC is fostering more, better and faster scientific discovery and HP is at the forefront -- advancing the power of computing to boost scientific productivity." Some HP customers that exemplify the trend of HPC growth include: -- Saoirse Corporation, Cambridge, Mass., focuses on nanosystem-based biological technology and recently purchased a five-node HP Cluster Platform 4000 system running Accelrys Materials Studio and Accord software for primary research in drug delivery. Saoirse will use the cluster to perform critical research tasks to build and simulate the interaction of drugs and materials with local media, tasks that are not feasible using traditional experimental methods. -- At the University of Texas Medical Branch (UTMB), scientists employ a 15-node HP Cluster Platform 4000 running United Devices GridMP software to manage an evolving campus-wide grid computer. The cluster and grid technologies are used to rapidly screen extensive chemical databases to identify drugs that can combat bioterrorist pathogens and treat emerging infectious diseases. The HP Cluster Platform also runs high-memory and I/O-intensive applications to determine structures of highly pathogenic viruses to near atomic resolution. UTMB is one of 10 National Institute of Health (NIH) Regional Centers of Excellence in Biodefense. UTMB's Galveston National Laboratory is one of two national centers established by the NIH for biodefense and infectious disease research. -- Canada's Shared Hierarchical Academic Research Computing Network (SHARCNET) uses HP Cluster Platform 4000 systems in a vast computational grid that provides HPC capability to 16 leading Ontario universities, colleges and institutes for advanced scientific research. SHARCNET is accelerating breakthroughs in such areas as human genomics, the containment of infectious human and animal diseases, improving weather prediction, simulating the collapse and formation of stars and planets, and the development of nano-scale electronic devices. -- South Africa's Council for Scientific and Industrial Research (CSIR) is implementing a 52-node HP Cluster Platform 4000 as part of its new HPC facility, C4, established to enhance its scientific computing infrastructure and develop skills and experience for South Africa's HPC community. The C4 facility will be used for applications in bioinformatics, aerospace, material sciences, geosciences and other large-scale modeling and simulation activities and will act as a general purpose capacity computing infrastructure. -- Norway's University of Tromso upgraded to a 100-node HP cluster system to provide a 10-fold increase in computing power and meet the growing demands of its chemistry and biotechnology researchers. More intensive, time-critical projects can now be undertaken and the facility is attracting more research funding and scientists to the university. -- Spain's National Institute of Aerospace Technique (INTA) uses a 32-node HP Cluster Platform 6000 to perform its principal aerospace research involving simulation, structural analysis and electronic design. -- China's Daqing Oilfield is using a 16-node HP Cluster Platform 3000 system running Linux and Schlumberger Eclipse reservoir simulation software to locate and exploit petroleum resources more quickly and cost-effectively. -- China's Fudan University has acquired a 128-node HP DL360 cluster and a 16-node HP rx2600 cluster system for a supercomputer center focused on research in life and materials sciences, chemistry, physics and mathematics. -- Shanghai University has acquired a 174-node HP DL360 cluster for use as a shared grid research platform. More information about HP in high-performance computing is available at its Web site.