13. Multithreading vs Multitasking vs Multiprogramming

1. Multiprogramming (1960s) — problem: CPU sat idle 80%+ of the time waiting on I/O. Fix: load several programs in RAM; switch when one blocks on I/O. Goal: maximize CPU utilization. No time slicing, no user interaction (batch).

2. Multitasking (1970s–80s) — problem: a CPU-bound job could hog the CPU forever under multiprogramming. Fix: a hardware timer forces a switch every quantum, regardless. Goal: responsiveness + fairness. Preemptive.

3. Multithreading (1990s–present) — problem: switching between heavy processes (full TLB flush) is expensive. Fix: multiple threads within one process, sharing memory, cheap to switch. Goal: concurrency + efficiency within a program.

Aspect Multiprogramming Multitasking Multithreading
Unit	Program	Process	Thread
Switch trigger	I/O only	Timer + I/O	Timer + I/O + yield
Memory	Separate	Separate	Shared
Goal	CPU utilization	Responsiveness	Concurrency + efficiency
Context switch cost	High	High	Low
Era	1960s	1970s–80s	1990s–now

Modern OSes run all three at once: multiple processes in RAM (multiprogramming), timer-preempted for fairness (multitasking), each running several threads (multithreading) — e.g., Chrome uses separate processes per tab and multiple threads within each tab.

---