An important hardware design decision is the size of page to be used. There are several factors to consider.
- Clearly, the smaller the page size, the lesser is the amount of internal fragmentation. To optimize the use of main memory, we would like to reduce internal fragmentation.
- On the other hand, smaller the page, the greater is the number of pages required per process which could mean that some portion of page tables of active processes must be in virtual memory, not in main memory. This eventually leads to double page fault for a single reference of memory.
Rate at which page fault occurs:
- If the page size is very small , then ordinarily a large number of pages will be available in main memory for process,which after some time will contain portions of process near recent references leading to low page fault rate.
- As the size of page is increased, each individual page will contain locations further and further from any particular recent reference. Thus, the effect of the principle of locality is weakened and the page fault rate begins to rise.
- Eventually, however the page fault rate will begin to fall as the size of page approaches the size of the entire process.
Size of physical main memory and program size:
- For a given size of TLB, as the memory size of processes grows and s locality decreases, the hit ratio on TLB declines. Under these circumstances, the TLB can become a performance bottleneck.
modified 2.1 years ago
written 2.1 years ago by