The rise of 64-bit computing in technical environments is attributed to two primary benefits of 64-bit systems: floating point performance and virtual address space.

Floating point performance of 64-bit Itanium 2-based systems is significantly faster— approximately 45 percent, by one measure—than the floating point performance of IA-32 systems. On industry benchmark results reported to the SPEC organization, the 1GHz HP Workstation zx6000 posted the world’s fastest floating point performance— 1,356, and HP expects to soon post the even-higher result of 1,410. By comparison, the fastest IA-32 posting—which was for a 2.8GHz Dell 340 Precision workstation—was just 938.

With 32-bit computing, there is a growing disconnect between the kernel’s virtual address space and the memory requirements of large servers and workstations. Despite the availability of 32-bit Intel systems that can be configured with up to 64GB of physical memory, most operating systems will support only 16GB of physical memory. That’s because the kernel doesn’t have the room to support any more. These systems cannot support data set requirements larger than 3 GB per process—a severe limitation for many technical computing applications.

By comparison, 64-bit systems have truly massive virtual address spaces, and therefore allow the kernel to manage the massive physical memories required for large databases, high-performance computing (HPC) environments and systems that run many applications concurrently.

For a closer look, read the white paper.