Fast machine is slowed down by Virtual Memory.

dennis wrote:
<!--coloro:blue--><span style="color:blue <!--/coloro-->
> It doesn't.<!--colorc--><!--/colorc-->

You made a reference to "Windows Internals". It says:

"Sections mapped to committed memory are called page file backed
sections because the pages are written to the paging file if memory
demands dictate. (Because Windows can run with no paging file, page file
backed sections might in fact be "backed" only by physical memory.)"

And that is what happens. As long as there is enough free ram, it will
not fail.

 

Laurence wrote:<!--coloro:blue--><span style="color:blue <!--/coloro-->
> Okay, since it gets more complicated with applications using virtual
> memory (even when there are GB's of unused physical memory!) then can
> Vista be instructed to first use all RAM and only when ALL RAM is used
> then switch to virtual memory when needed or when an application
> requires it?<!--colorc--><!--/colorc-->

This is under the control of the individual application, and how that
the writer of that app chooses to handle memory management. There's
nothing in the operating system's memory manager to enforce this
behaviour for all applications.

An application can ensure that regions of its address space are pinned
into physical memory, by calling VirtualLock(). These locked regions
will never be written to the paging file, even if one is available.

See


Just to clarify, there's a difference between "virtual memory" and
"paging" - they are not synonymous. As I described earlier, NT *always*
uses virtual memory: in other words, every process gets a virtual
address space, with its own addresses and all, which aren't the same as
the addresses as the real physical pages in memory. Every memory access
will involve a virtual-to-physical address translation. If (and only if)
the page was been paged out to the backing store (ie, the paging file),
there will be a page fault, and the page will be read from disk back
into RAM. This allows performance optimisations, such as (for example)
"moving" data around by changing the virtual addresses - without needing
to actually shift the data to some other physical location, which would
be slower even in memory-to-memory moves.

Cheers
Andrew

 

dennis wrote:
<!--coloro:blue--><span style="color:blue <!--/coloro-->
> It doesn't.<!--colorc--><!--/colorc-->

Whatever; I was probably remembering wrongly, in that case.

 

Andrew McLaren wrote:<!--coloro:blue--><span style="color:blue <!--/coloro-->
> dennis wrote:
> <!--coloro:green--><span style="color:green <!--/coloro-->
>> It doesn't.<!--colorc--><!--/colorc-->
>
> Whatever; I was probably remembering wrongly, in that case.<!--colorc--><!--/colorc-->

"Whatever" ?