Multiprocessors

Depends on	Scheduler

Many modern computers have multiple hardware ‘cores’, meaning they can execute more than one process at once, each with its own context. However most multiprocessor architectures – at least the ones you are likely to meet day-to-day – still have a single address space and share the same memory. Thus it does not matter which processor runs a particular job at a particular time. This does not change the principle of scheduling, so it’s no big deal.

One place where this does matter a bit is in multithreading, because it can be more efficient to schedule threads from the same process together to avoid expense in context switching.

If a processor cache data which is shared between processes – and those processes are (or could be) scheduled on different processors – then some care is needed that the data seen is coherent. This is necessary in caches, networks etc. Fortunately (perhaps?) it is beyond the scope of this course unit.)

---

Experiment

If you can find a ‘System Monitor’ on your computer, it may show you how busy each of the cores (maybe labelled “CPU”?) is; a rolling activity graph makes a nice display. On a workstation, often it/they will be doing little which is useful; perhaps sometimes you will see one core being particularly active although which core it is may change as that process is swapped out and then back into execution by a different core.

---

Also refer to:	Operating System Concepts, 10th Edition: Chapter 1.3.2, pages 16-19

Context Dependencies Dependants

Max. number:

Max. distance:

Next

Articles on Deadlock

Atomicity • Deadlock • File Locking • Locks • Multi Threading • Multiprocessors • Mutual exclusion • Operation Ordering • Race Conditions • Reentrancy • Semaphores • Starvation • Threads