Itanium Growth in Technical Computing

Technical computing in and of itself can be done on a wide range of systems, as illustrated in Figure 5-2. At the base level, there are systems that contain only one or two CPUs and are used by the engineers doing Computer Aided Engineering (CAE) tasks. Today, Hewlett-Packard offers Itanium-based systems consisting of 2-CPU workstatiopns and 4-CPU symmetrical multi-processing (SMP) systems.

At the higher end you see the processing power of a cluster of computers to act as one big computer. The UNIX^®[16] and Linux^®[17] world works this very well. A cluster can focus the power of smaller computers to create a sort of supercomputer. “HP-UX Clustering” is the name we give for production-oriented clusters that supply enterprises with compute and storage resources.

^[16] UNIX is a U.S. registered trademark of The Open Group.

^[17] Linux is a registered trademark of Linus Torvalds.

These are normally—but not always—“mid-range” devices that contain tens of processors. Typically they are used for production-oriented customers who need the robustness of HP-UX and the availability features that HP-UX provides. Prime examples in this arena include aerospace and automobile manufacturers.

Another option available is what we call “Native Linux Clustering”. This is the term for those who normally are the higher-end clusters found often in science and research laboratories. These customers tend to be more price-conscious, more leading-edge customers who are more likely to do their own intense system administration and system integration.

The next boost involves the clustering itself. High-end research institutes are looking more and more to these computing clusters to maximize their price: performance ratio. For these two reasons alone, 95% of HPTC buyers purchased Linux or UNIX-based systems in 2002.

And finally, the last increase in performance comes directly from the Itanium chip's built in ability to handle floating-point equations with extra precision and performance.