What is page thrashing?

Just ! A  wonderful Explaination

"Programs that frequently access many widely separated locations in memory are more likely to cause page thrashing on a system. "   why ???

"So is running many small programs that all continue to run even when you are not actively using them""  why??

thank you :)

this explanation is more than enough

I have some doubts.

1) As you said the page size is generally 4KB or 8KB. What factors decides the page size? Is it the kernal or what?

2) As per I understand the total RAM available is called the physical memory, and the memory which you map to your disk (Hard disk etc..) is called the virtual memory.

But you said that all the memory address that is refered by the process is called the virtual memory, and when PMM maps that address to actual address (RAM, HDD) its called physical address.

Please clarify

  1 User has rated as useful.

  

Login to rate this answer.

Excessive page fault is called thrashing. CPU spend most of the time in page fault related activity than the useful work.

  1 User has rated as useful.

  

Login to rate this answer.

Page size you have to choose as per the requirement. If you choose page size big then processing of page fault takes more time but if you choose page size small then there is possibility that more number of page fault occures but processing time of page fault will reduce.

During each page fault you need to bring complete page from VM to PM and transfer time will be more if page size is big.

I hope this will make picture little bit clear.

  1 User has rated as useful.

  

Login to rate this answer.

I believe this can be described in a simpler and more complete form:

1) Understand what Virtual Memory is - it is functionality implemented by the running OS (Linux, Windows, Mainframe OS, etc) to make a system appear to have more real memory than it does.

2) The granularity is normally applied at the process/task level.

3) VMem requires an MMU - Memory Managemen Unit.

4) Each process is assigned a given amount of VMem, for example 5 processes can be assigned 1GB of memory each (adding up to 5GB) while the actual system only has 1GB of REAL memory.

5) Each processed is assigned a Page Table (or a Segment Table and multiple Page Tables in more complex systems). The Page table defines which Chunks of memory (usually 4K/8K) belonging to that process are actually residing in Real memory at any given moment. The rest of the memory has been "Paged Out" which means save to disk.

6) It is possible for an idle process to completely Paged Out. In that case, that process is not consuming any memory or processor time at all. An active process will usually have its most active Pages (the ones it is reading and writting from) in Real Memory. When it accesses those pages, no "Paging" is required.

7) Paging occurs when the process tries to access a memory address that is in a Page that has been paged out to disk and is not currently in Real memory. This results in an interrupt (a Page Fault interrupt) that the OS passes to one of its components - The Page Manager (PM) to handle. The Page Manager, keeps track of where the corresponding Page is on disk and reads it out to Real memory as well as it updates the Page Table to indicate the page is now in memory. This all transparent to the process - it resumes execution and is able to store/read from the corresponding page.

8) A phenomenom can occurr when there are multiple processes running that forces the OS to perform much process switching and paging in and out process pages.

9! When the OS reaches the point that it is using more processor time Paging (In and Out) rather than allowing processes to complete their task we have reached the point of *Page Trashing*

I hope this is of help.. or write to crimola AT gmail DOT com for more info.

Carlos

  2 Users have rated as useful.

  

Login to rate this answer.

First of all there is a one sentence definition of Page Thrashing (I believe someone posted a similar answer here):

Page Thrashing only comes about when you use Virtual Memory which requires an MMU and "fools" a process into believing that it has memory available than actual Real Memory.  However. it requires Page Management and an OS that brings needed (accessed) pages from Disk to Memory On Demand (when the process accesses a memory address that is currently on disk, it must be read from disk to real memory an *mapped* to that process's memory address space.  *Page Trashing* occurs when the OS's Page Manager (or let's just say the OS) is spending more time, or an excessive amount of time, *paging in and out* than running needed tasks.  The process of paging requires substantial processor time.


"Programs that frequently access many widely separated locations in memory are more likely to cause page thrashing on a system. "   why ???

I believe this would only be the case in processes that access shared memory.  The Page Manager would be forced to first make room for the new page to be bringed in and then bring in the needed page.

Most commonly, Page Trashing occurs with high activity and time slicing leading to the running process being timed out and a new process going into "running" state and requiring its pages to be "swapped" to real memory.  This is what causes Windows to go into a long delay while you see your disk activity light quickly flashing.

"So is running many small programs that all continue to run even when you are not actively using them""  why??

It's not the size of the program is the amount of memory they use.  As little as two or three processes can lead to Page Trashing if your "time slice" is short and each of those processes access their virtual memory in a dispersed form.

thank you :)

Hope it helps! :)

  2 Users have rated as useful.

  

Login to rate this answer.

Thrashing is caused by under allocation of the minimum number of pages required by a process, forcing it to continuously page fault. The system can detect thrashing by evaluating the level of CPU utilization as compared to the level of multi-programming. It can be eliminated by reducing the level of multi-programming.