Answer 6

First of all, note that these chips (sometimes characterized as RISC) use fixed-size instructions in order to simplify the instruction decoding process and achieve high performance. The size of an address field embedded in an instruction is necessarily limited. Arbitrarily adding bits to the address field forces the width of the entire program memory to increase, driving up the overall cost. Bank-switching to access data memory makes sense for several reasons: The instruction format does not need to change across members of a chip family. The memory decoding for the bank-select bits does not need to be as fast as the decoding of the address bits from the instruction field, so large memories do not incur a speed penalty. Deselected banks can also be put in a low-power state, reducing the power penalty for large memories. When data is laid out in memory to take into account locality of access, relatively few bank-switching instructions are required, maintaining high performance.

Contributor: Dave Tweed

Published: September-2002