We will brie?y introduce paging to ?nish off this lecture. When a process is loaded, not all of the pages are immediately loaded, since it's possible that they will not all be needed, and memory is a scarce resource. The process' virtual memory address space is broken up into pages, and pages which are valid addresses are either loaded or not loaded. When the MMU encounters a virtual address which is valid but not loaded (or "resident") in memory, then the MMU issues a pagefault to the operating system. The OS will then load the page from disk into RAM, and then let the MMU continue with its address translation. Since access times for RAM are measured in nanoseconds, and access times for disks are measured in milliseconds, excessive pagefaults will clearly hurt performance a lot.