Which statement best describes memory-mapped I/O?

• A. Memory-mapped I/O maps device registers into the normal address space.
• B. It uses a separate I/O space with in/out instructions.
• C. It cannot be used for peripherals.
• D. It is used only on mainframes.

Answer: (A)

Memory-mapped I/O maps device registers into the normal address space, so software accesses peripherals using the same load and store instructions used for RAM. A dedicated range of physical addresses corresponds to device registers, meaning reads and writes to those addresses interact with the device rather than memory. This makes programming peripherals straightforward: you can use ordinary pointers, pointer arithmetic, and even byte-level access if the registers support it. Because those addresses correspond to hardware, the region is typically marked non-cacheable and may require memory barriers to ensure correct ordering and visibility of side effects. This approach contrasts with systems that keep a separate I/O space and require special in/out instructions to communicate with devices. It’s a common method across many architectures, not limited to mainframes.