BGDCL: Assistant Manager (CSE): Exam: 15.03.2024 (BUET)প্রশ্ন ও সমাধান

Definition: Multiprogramming keeps several programs resident in memory at once on a system with only one CPU, so the CPU is never left idle — whenever the running program blocks (typically for I/O), the OS switches to another ready program.

How it is achieved:

• Ready queue: Multiple processes are loaded into memory; the OS keeps a queue of processes that are ready to run.
• CPU scheduler: Selects the next process to run from the ready queue (FCFS, Round Robin, Priority, etc.).
• Context switching: When the running process blocks for I/O or its time slice expires, the CPU saves its state (program counter, registers) and loads another process's saved state — creating the illusion of simultaneous execution.
• Interrupts: A timer interrupt forces periodic switches (preemptive multiprogramming); an I/O-completion interrupt wakes a waiting process and moves it back to the ready queue.
• Memory management: Several processes must coexist in memory (via partitioning, paging, or segmentation) so the OS can switch instantly without reloading from disk.

stateDiagram-v2
    [*] --> New
    New --> Ready : admitted
    Ready --> Running : scheduler dispatch
    Running --> Waiting : I/O or event wait
    Waiting --> Ready : I/O completion
    Running --> Ready : time slice expires
    Running --> Terminated : exit
    Terminated --> [*]

Example: While Process A waits for a disk read, the scheduler immediately dispatches Process B from the ready queue, so the single CPU keeps doing useful work instead of sitting idle.

Exam takeaway: Multiprogramming ≠ multiprocessing — it produces concurrency (interleaved execution) on one CPU, not true parallelism. Its entire purpose is maximizing CPU utilization by overlapping one process's computation with another's I/O wait.