Re: How to get Vista to use more memory?
In message <firstname.lastname@example.org> araibob
<email@example.com> was claimed to have wrote:
>Sorry, 'page file'. I use "swap file" and "page file" interchangeably.
I only mention it because this is actually one case where a more
traditional "swap" system and Windows "page file" system are very
different, and understanding those differences are critical to
understanding pagefile use.
>Old O S programmer Slang is still slang. I don't have specific
>'knowledge that the swap file is in use, however, The Performance tab in
>Windows Task manager has these values that make me think so.
>Total Physical Memory (MB)
>Cached 6712 << using the page file?
Cached is actually the opposite, this is memory that is otherwise
available, but is being used to cache something on disk in memory.
So despite being used, this is as good as free RAM in that it can be
released at a moment's notice, but should the resources that it's
caching become needed, it will result in a performance increase.
>Page File 2300M / 16565M << using the page file?
Page File is a bit of a misnomer here, this actually indicates that,
aside from caching, the OS has 2300M of allocated memory to running
This allocated memory can be backed by physical memory, or by disk
(either in the pagefile, or elsewhere in many cases.) For example, when
a large program loads into memory, the EXE or data files are mapped to
memory, but the data is not necessarily kept in memory at all times.
Rather then writing data back out to the pagefile, Windows know that it
can release the memory when needed, and re-read the original data from
disk on demand.
The pagefile counter gives you an idea of the total number of allocated
pages in use, when combined with the data from the Memory tab, you can
get an idea of how many pages would need to come from disk vs physical
>Kernel Memory (MB)
>Paged 294 << using the page file?
"Paged" here is actually the kernel's version of a pagefile, it's not
necessarily paged to disk right, but rather, it's the memory that is
paged (whether those pages are in RAM, in the pagefile or elsewhere on
disk), vs NonPaged which is address space directly used by the kernel
(typically for lower level drivers, and lower level kernel components
including the paging subsystem itself)
>I have also seen threads on this topic (some time ago) and they noted
>that even when there is lots of memory MS insists on writing things to
>the page file.
Many things are actually paged automatically, but few are actually
"written" to the page file as much as just space in the page file is
For example, if an application requests a 1GB chunk of memory address
space, but doesn't write anything to it and the OS can't immediate
satisfy the request from available memory, this address space is "paged
out" to the pagefile without anything actually being written (although
the pagefile may be expanded to satisfy the memory request) -- This is
the most common situation people are referring to when they talk about
Windows insisting on writing some things to the page file.
In your case, with the ratio of total ram minus cached vs the page file
allocation being where it is, I doubt you're seeing any performance
impact due to paging.
You can set up performance counters to monitor page file use as well as
read and written pages if you want, but unless you're doing something
that uses 8GB of memory (virtual machines, video or large photo
encoding, etc) I'd be surprised if you ever see any significant amount
of any data being written to the pagefile.