how big should the page file partition be & do you really need one?

[~Aart]

I've read the page file in XP should have it's own partition, and I've read
you don't really need one.

Can anybody straighten me out on this?

If you shuld need it, how big shuld it be?

Thanks,

Art

 

[Carey Frisch [MVP]]

For best performance, leave the page file on the same
partition Windows XP is installed on. Moving the page
file to a different partition, on the same drive, will actually
degrade performance due to additional seek time.

Virtual Memory in Windows XP
http://aumha.org/win5/a/xpvm.php

--
Carey Frisch
Microsoft MVP
Windows XP - Shell/User

Be Smart! Protect your PC!
http://www.microsoft.com/security/protect/

--------------------------------------------------------------------------------------


| I've read the page file in XP should have it's own partition, and I've read
| you don't really need one.
|
| Can anybody straighten me out on this?
|
| If you shuld need it, how big shuld it be?
|
| Thanks,
|
| Art

 

[Spinner]

Having the page file on it's own partition will only stop it from becoming
fragmented.
You will hear arguments on both sides, but in my opinion, it's not needed.
It would make more sense to place it on a separate drive from the OS.
As to size, that depends on what your doing with your computer.
Using word, cruising the net, a small one is fine.
Playing high end games or editing graphics or movies, larger is better.
Unless you have specific needs, just let Windows handle it. It does a fine
job
on it's own for the majority of people.

 

[Crusty \(-: Old B@stard :-\)]

And moving the pagefile to another hard drive on a different IDE controller
will increase performance by decreasing the seek time (assuming that both
hard drives are identical)!

When the page file is on a 2nd drive on IDE 1 the controller performs
sequential seeks (it can only read/write to one device at a time). When it
is on a 2nd hard drive on IDE 2, the read/writes are concurrent with those
of IDE 1. Each IDE controller is doing "it's own thing", when necessary!

--
Regards:

Richard Urban

aka Crusty (-: Old B@stard

 

[Al Dykes]

Having the page file on it's own partition will only stop it from becoming
fragmented.
You will hear arguments on both sides, but in my opinion, it's not needed.
It would make more sense to place it on a separate drive from the OS.
As to size, that depends on what your doing with your computer.
Using word, cruising the net, a small one is fine.
Playing high end games or editing graphics or movies, larger is better.
Unless you have specific needs, just let Windows handle it. It does a fine
job
on it's own for the majority of people.

Making a partition just for the swap file can slow things down because
the disk head will do more ling seeks. You can prevent swap file
fragmentation by setting pagefile size MIN = MAX in
controlpanel/system. If, after that you do a defrag that includes
system files the swap file (and everything else) will be 1 fragment,
each. There is at least one free downloadable tool that does that. I
think it's from System Internals web site. The paid-for defrag tools
do it too.

You need to decide how big to make the swap file. Start all the
appplications you waht to be able to run concurranlty then start Task
Manager (right mouse click on the task bar, pick Task Manager)
and look at the Performance tab. The lower half of the panel graphs
your pagefile usage. You want a generous # but too much doesn;t buy
you any additional performance. If, later, you get an Out Of Space pagefile
message you can expand it in Control Panel and defrag, again.

 

[Alex Nichol]

I've read the page file in XP should have it's own partition, and I've read
you don't really need one.

Can anybody straighten me out on this?

I regard a separate partition on the same drive as your system as a bad
idea. The most important point with the page file is to keep time spent
on the heads 'seeking' to it as low as possible: such a separate
partition is directly against that.

Putting it on a separate physical drive is OK, but even then I would
have it in a data partition, as the first thing put there, setting its
*initial* size as being as much as is likely to see actual use. Read up
more on background and reasoning at my page
www.aumha.org/win5/a/xpvm.htm

 

[Alex Nichol]

Making a partition just for the swap file can slow things down because
the disk head will do more ling seeks. You can prevent swap file
fragmentation by setting pagefile size MIN = MAX

That is wasteful, if you have the Max big enough to cover contingencies
and provide for all the potential space that programs have asked for but
never brought into use. You need an initial size big enough to cover
the size it will actually grow to in practice in *your* workload (which
may be surprisingly small) and a high max to allow expansion if needed

 

[Al Dykes]

That is wasteful, if you have the Max big enough to cover contingencies
and provide for all the potential space that programs have asked for but
never brought into use. You need an initial size big enough to cover
the size it will actually grow to in practice in *your* workload (which
may be surprisingly small) and a high max to allow expansion if needed

I'm the proponant of MIN=MAX pagefile setting. I came to that because
most of my daytime job is with software development tools that were
huge memory pigs. Setting up NT with the defaults, and then letting
the pagefile expand in increments as a tool went thru it's first-use
initialization produced an amazingly fragmented pagefile.

For average users I recommend the default memory settings. The OP
was looking for an unfragmented pagefile and I threw in my 2 cents.

 

[Spinner]

I'm the proponant of MIN=MAX pagefile setting. I came to that because
most of my daytime job is with software development tools that were
huge memory pigs. Setting up NT with the defaults, and then letting
the pagefile expand in increments as a tool went thru it's first-use
initialization produced an amazingly fragmented pagefile.

For average users I recommend the default memory settings. The OP
was looking for an unfragmented pagefile and I threw in my 2 cents.

Your kind of jumping to conclusions there.
Where did he say anything about a non fragmented pagefile?

 

[Alex Nichol]

I'm the proponant of MIN=MAX pagefile setting. I came to that because
most of my daytime job is with software development tools that were
huge memory pigs. Setting up NT with the defaults, and then letting
the pagefile expand in increments as a tool went thru it's first-use
initialization produced an amazingly fragmented pagefile.

The point you miss - and AFAIK it was new in XP - is that the
'potential' file space represented by Max - Min can be used to assign
the pages of VM space that programs have had ALLOCed but not brought
into use. And probably never will. This can easily run into hundreds
of MB. So having Max = Min ties up a lot of disk for these - without
need. On a largish RAM - 512 MB up, I suggest Initial 100, Max say 1GB.
The file will likely never expand. *If* it does, you can turn it off,
defrag and turn it on again with optimal initial size