What is a misaligned operand? Why are misaligned operands such a problem in programming?

Question

What will be an ideal response?

Answer 1

If all instructions and data values were the same length and equal to the length of a computer word, the concept of a misaligned operand would not exist. However, in the real world, some data operands are shorter than a computer word. For example, computers with 32?bit wordlengths can operate with 8?bit or 16?bit data values.

A misaligned operand exists if the address of the operand is not on a boundary that is an integer multiple of the operand size. Suppose memory is byte addressed. A byte operand can be on any byte boundary and cannot be misaligned. Now consider a 16?bit value. That takes two bytes. Consequently a 16?bit value is aligned at addresses 0, 2, 4, 8, … ; that is all even addresses. A 16?bit value at an odd address is misaligned because it is split between two aligned words, e.g., a 16?bit value at address 7 would have one byte at address 7 and one at address 8, thereby occupying two word slots at byte addresses 6, 7 and 8, 9. Similarly, a 32?bit word would be aligned at addresses 0, 4, 8, ….

Conceptually, a misaligned word is not a problem; that is, you just read data from memory. Who cares where it comes from? In practice, misalignment is a problem. At worse, the computer may crash (i.e., abort the program because a misaligned operation is illegal). At best, it slows the computer because what should be a single memory access becomes two memory accesses. If a 32?bit word were misaligned at location 9, the word at location 8 (bytes 8, 9, 10, 11) would be read. The desired word is at address 9, 10, 11, 12. Consequently, the words at 12, 13, 14 would have to be read and the byte at 12 extracted and appended to the three bytes already read.