When 64-Bit Makes Sense

Sixty-four-bit chips gain most of their performance improvements from their greater number of registers (used to temporarily store program variables), larger registers, and the ability to address more memory. Most desktop applications, such as desktop productivity suites, do not require more memory than current PCs can provide, and thus are likely to remain 32-bit applications for the foreseeable future. However, two major categories of applications are likely to benefit.

Memory-intensive business applications. Some business applications, such as online analytical processing (OLAP) or enterprise resource planning (ERP) systems, can benefit from very large amounts of memory, typically by storing entire databases in memory rather than accessing them from disk, which is orders of magnitude slower. In many cases, organizations might consider deploying a 64-bit version of SQL Server for the back-end database of their line-of-business applications, while leaving the middle tier and front end as 32-bit applications. This is also true of some Microsoft products, such as Systems Management Server or Microsoft Operations Manager: the database for these systems could be a 64-bit edition of SQL Server, with other components of the application running as either 32- or 64-bit versions on an x64 processor.