Answer: Operating systems evolved from no OS (manual operation) to batch systems, multiprogramming, time-sharing, personal-computer, network/distributed, real-time and today's mobile and cloud systems, each stage driven by better hardware and the need to keep the CPU busy and users productive.
- Serial processing / no OS (1940s to early 1950s): programmers worked directly on the machine (ENIAC, UNIVAC I) through switches, plug boards and card readers, booking the machine by time slot. Set-up time was large and the CPU sat idle most of the time.
- Simple batch systems (1950s to 1960s): jobs with similar needs were grouped into batches on punched cards or tape and a resident monitor loaded and ran them one after another, using job control cards. Examples: GM-NAA I/O (for the IBM 704), IBSYS (IBM 7090/7094). There was no interaction with the user and the CPU still waited during slow I/O.
- Multiprogrammed batch systems (1960s): several jobs are kept in memory; when one waits for I/O the CPU switches to another. This needs memory management, CPU scheduling and protection. Example: IBM OS/360. Spooling (SPOOL) also overlapped I/O of one job with computation of another.
- Time-sharing systems (1960s to 1970s): an extension of multiprogramming where the CPU is switched among users at terminals so quickly (time slices) that each user feels the machine is his own; gives interactive response. Examples: CTSS, MULTICS, UNIX.
- Personal computer systems (late 1970s to 1980s): single-user machines; early OSs such as CP/M and MS-DOS were single-user and single-tasking with a command-line interface. Later the graphical user interface arrived with Apple Macintosh and Microsoft Windows, and PC systems became multitasking (Windows NT, Linux).
- Multiprocessor (parallel) systems: two or more CPUs share memory and a clock; the OS must schedule work across processors (symmetric or asymmetric multiprocessing). Gives higher throughput and reliability.
- Network and distributed systems (1980s onward): computers connected by a network share files, printers and processing. A network OS lets users access remote machines; a distributed OS makes many machines appear as one system (remote procedure calls, distributed file systems).
- Real-time systems: results must be produced within strict time limits (hard and soft real time); used in industrial control, avionics, medical devices. Examples: VxWorks, QNX, RTLinux.
- Mobile and embedded systems (2000s): OSs for phones and devices, optimised for touch input, battery life and wireless networking. Examples: Android, iOS.
- Cloud and virtualised systems (2010s): hypervisors and virtual machines/containers let one physical server run many OS instances; services and storage are offered over the Internet with on-demand scaling.
Trend: each step improved CPU utilisation, convenience and response time, while the OS took on more duties: process and memory management, protection, file systems, networking and security.