Multiprocessing and Multitasking

Computer Organization & Embedded System — Real-Time Operating and Control System, NEC licence examination syllabus (Nepal Engineering Council).

Multiprocessing and Multitasking

One CPU juggling many tasks vs many CPUs actually working simultaneously.

🔄 Three Related Terms

MultitaskingOne CPU switches rapidly between multiple tasks, giving the illusion of simultaneity.

MultiprocessingGenuinely multiple CPUs run tasks at the same time.

MultiprogrammingMultiple programs are kept in memory ready to run, so the CPU is never idle waiting for I/O.

Context switching

When the OS switches from one task to another, it must save the current task's registers/PC/state (into its PCB) and load the next task's saved state — this overhead is called a context switch. Too many context switches waste CPU time on switching rather than real work.

🤝 Cooperative vs Preemptive

CooperativeA task voluntarily gives up the CPU (yields). Simple, but one badly-behaved task can freeze everything.

PreemptiveThe OS can forcibly interrupt a running task (e.g. on a timer). Safer, and what virtually all modern/real-time systems use.

💡 Short-answer: "Why is preemptive multitasking preferred over cooperative in an RTOS?" → Guarantees the scheduler can always regain control to meet deadlines, regardless of whether a task misbehaves.

Concurrency and parallelism are not the same word

The three terms in the table describe two genuinely different things, and the distinction is worth being precise about.

⚡ Overlapping versus simultaneous

ConcurrencySeveral tasks are in progress over the same period. On one CPU they take turns; at any single instant exactly one is running. This is multitasking.

ParallelismSeveral tasks execute at the same instant, which requires more than one processor. This is multiprocessing.

Concurrency is a way of structuring a program; parallelism is a property of the hardware it runs on. A concurrent program runs correctly on one core and simply goes faster on several — which is why the distinction is not pedantry: it decides whether adding cores will help at all.

💡 It also explains why multitasking helps even on a single CPU. While one task waits for disk or network, another can use the processor — so the gain comes from filling idle time, not from doing two things at once. That is the whole purpose of multiprogramming, which is why the three terms belong together.

What a context switch costs

Switching between tasks is not free, and the cost is more than the obvious part.

💾 Where the time goes

DirectSaving registers and the program counter, loading the next task's — measured in microseconds.

IndirectThe new task's data is not in cache, so it runs slowly until the cache refills. This is usually the larger cost and is invisible in the switch itself.

💡 The cache effect is why very short time quanta hurt more than the switch count suggests, and why pinning a task to one core can help — a task that keeps moving between cores loses its cache contents every time, paying the indirect cost repeatedly without any switch appearing to be at fault.

Syllabus points

Concepts and difference
Context switching
Cooperative vs preemptive multitasking

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.