What is the purpose of the multiprocessing in computer system structure?

Answer 1

The primary purpose of multiprocessing is to allow more than one process to run, or to appear to run, at the same time. This allows more work to be done more quickly, because one process can be running while another is waiting on something else, so it allows the CPU to be more efficiently utilized. Multiprocessing is an architecture that allows the concept of multiple processes, each lined up to compete for execution by the CPU. The operating system manages the dispatching of each process according to a priority scheme. Processes can be ready, which means the OS can dispatch them to run, or they can be blocked, which means they are unable to be dispatched because they are waiting on some event, such as IO completion, keyboard entry, another process, something. Done correctly, which most modern OS's do very well, it will seem that you can run multiple processes at the same time. If there is only one CPU, then only one process can be running, or dispatched, at any one time. Still, the appearance of concurrent execution will exist. If there is more than one CPU, such as in a dual or quad core processor, then the OS is able to dispatch more than one process at the same time. This is true parallel execution. In modern OS's, the concept of threading exists. Processes and threads are somewhat similar. While different OS's have different terms for processes and threads, it is general convention that a process is an address space supporting a program, with one or more threads running within it. Threads, on the other hand, are what the OS actually manages in term of dispatchability. A thread, then, is an execution path through a process. You can write a program that has only one thread, and the OS will handle it accordingly. If you have several "things" to do within that program, you are responsible for the dispatching of those "things". Well, those "things" can be threads. You can write code that initiates additional threads within your process, each designed to handle one "thing". You then turn dispatch management over to the OS, and you have now written a multi-threaded application. Whether we call them threads or processes or something else, the issue of thread safety is paramount. Thread safety is the ability of a thread to properly access and potentially modify a data structure that is shared between threads (or processes) without creating corruption of data. Various kinds of interlocking mechanisms exist to handle this - it is a big topic, and one that requires great care in its design.