Microsoft KB Archive/102025

From BetaArchive Wiki
Knowledge Base

Article ID: 102025

Article Last Modified on 11/1/2006


  • Microsoft Windows NT Advanced Server 3.1
  • Microsoft Windows NT Workstation 3.1
  • Microsoft Windows NT Advanced Server 3.1

This article was previously published under Q102025


When designing computers, there are two different architectures for handling memory storage. They are called Big Endian and Little Endian and refer to the order in which the bytes are stored in memory. Windows NT was designed around Little Endian architecture and was not designed to be compatible with Big Endian because most programs are written with some dependency on Little Endian.


These two phrases are derived from "Big End In" and "Little End In." They refer to the way in which memory is stored. On an Intel computer, the little end is stored first. This means a Hex word like 0x1234 is stored in memory as (0x34 0x12). The little end, or lower end, is stored first. The same is true for a four-byte value; for example, 0x12345678 would be stored as (0x78 0x56 0x34 0x12). "Big End In" does this in the reverse fashion, so 0x1234 would be stored as (0x12 0x34) in memory. This is the method used by Motorola computers and can also be used on RISC-based computers. The RISC-based MIPS computers and the DEC Alpha computers are configurable for Big Endian or Little Endian. Windows NT works only in the Little Endian mode on both computers.

Windows NT was designed around Little Endian architecture. The Hardware Abstraction Layer (HAL) is written so that all operating system-related issues are automatically handled. Therefore, it is possible to create a HAL that could work on Big Endian architecture. The basic problem with porting the code has to do with the way the code is written for all programs. Code is often written with the assumption that Big Endian or Little Endian is being used. This may not be specific to the HAL; it could be something as simple as bit masking for graphics. To clarify this concept more, two programming examples follow.

Example 1

      WORD y;
      WORD x;
   } POS;

   lparam = (DWORD) POS;

Basically, there is assumption in the code that Little Endian is being used. The switching of the bytes is being assumed in the 'C' structure. This is faster on Intel architecture, but will not work with Big Endian.

Example 2

Another example is a common practice of using bit masks. The following is an example of defining a bit mask:

   #define BITMASK    0x0008

This allows you to check if the 4th bit is a 1 if you AND it with another number. It also allows you to set the 4th bit by OR-ing it with another value. The problem comes when you OR or AND this with DWORD (Double Word) or anything other than a WORD size value. This causes strange things to happen and unexpected results. You might make assumptions about how it works with Little Endian, yet it won't work the same way with Big Endian. A large amount of code is already created with these assumptions built in.

Note that the PowerPC and Sparc chips are also switchable between Little Endian and Big Endian. However, the Apple PowerMac implementation of the PowerPC chip is stuck in Big Endian mode. Hence, Windows NT may port to the PowerPC, but probably not to the PowerMac implementation of the PowerPC.

Additional query words: prodnt uuh apple mips

Keywords: kbhardware KB102025