Peripheral Devices

Computer Organization & Embedded System — Input-Output Organization and Multiprocessor, NEC licence examination syllabus (Nepal Engineering Council).

Peripheral Devices: the outside world talking to the CPU

Every keystroke, click, and printed page goes through the same basic idea.

Right now, as you scroll this page, you're using at least two peripherals without thinking about it — a screen (output) and a touchscreen/mouse (input). Peripherals split into three families.

🔌 The Three Families

InputKeyboard, mouse, scanner — send data INTO the CPU.

OutputMonitor, printer, speaker — send data OUT of the CPU.

StorageDisk, tape — data flows both directions.

Each character exchanged is usually coded in ASCII (7 or 8 bits per character).

The core problem peripherals create: they're all wildly different speeds. A keyboard produces a few bytes per second; a disk can move megabytes per second. The I/O subsystem exists to bridge that speed mismatch without wasting CPU time.

💡 Likely one-liner: name 2 input, 2 output, and 1 storage peripheral with their typical use.

How large the speed gap really is

The chapter names the mismatch. Its scale is what makes every later design decision inevitable — take a 1 GHz processor, executing roughly one instruction per nanosecond:

Device Rate Instructions the CPU could run while waiting for one byte ──────────────────────────────────────────────────────────────── Keyboard (fast typist) ~10 B/s 100,000,000 Serial line, 9600 baud ~960 B/s 1,041,667 Hard disk ~100 MB/s 10 SSD ~500 MB/s 2

Eight orders of magnitude separate the slowest device from the fastest. No single transfer strategy can serve that whole range — which is exactly why the next chapters describe three of them, and why choosing between programmed I/O, interrupts and DMA is a question about the device's speed rather than about elegance.

💡 The keyboard row is the one to remember. Polling it in a loop wastes a hundred million instructions per keystroke — an almost incomprehensible amount of work to discover that nothing has happened. That single figure is the entire argument for interrupt-driven I/O.

Why the disk row points the opposite way

At the fast end the problem inverts. A disk delivering a byte every ten instructions cannot be handled by interrupting the CPU per byte — the interrupt overhead alone would exceed the work.

So fast devices need the CPU taken out of the transfer path entirely, which is what direct memory access does. Slow devices need the CPU not to wait; fast devices need it not to participate. Two opposite problems, from the same table.

Syllabus points

Input / output / storage devices; ASCII

Create a free account to tick topics off, take notes as you read, watch the video lessons and get a day-by-day study plan built around your exam date.