Why are page files used?

Answer 1

On a 32-bit system it is only possible to address a maximum of 2^32 bytes of memory (4,294,967,296 bytes which is 4 gigabytes). Memory is allocated in contiguous blocks such that if your program requires x bytes of memory, a block of x contiguous bytes must be available to meet the allocation request. If there is no single block large enough to meet the required allocation, you will get an out of memory error, even if there's more than x bytes of physical memory available in total.

To avoid these problems, we use virtual memory. If there is no single block available to meet an allocation, existing allocations can be temporarily moved to a page file, thus freeing up the physical memory. When memory that has been paged out is required, other allocations can be paged out to allow the required memory to be paged in. Given that the contents of memory are swapped to and from the page file, the page file is often called a swap file.

Data that was paged out of one physical address may be paged in to a completely different physical address. However, the virtual address does not change; the memory manager performs the translation between the physical and virtual addresses. This makes it possible for a 32-bit system to operate with a full complement of 4 gigabytes of memory even when the system has less than 4 gigabytes of physical memory installed.

64-bit systems allow us to address 18,446,744,073,709,551,616 bytes of memory which is far more memory than physically exists on the planet! For practical reasons, we can only use a small fraction of the full address space, however page files and virtual memory make it possible to use far more addresses than might be physically available.