Limit on the page file segment size?

[Boon]

Folks,

Is there a limit on the page file segment size?

The page file on my 7 year-old computer was fragmented and there wasn't
much space on the drive anyhow, so I used that as an excuse to get a new
computer.

My new computer has 4 gig of memory and two new WD SATA hard drives, a 320
gig and a 750. The system is on C: on the 320. I created a 12 gig
partition called X: on the 750, exclusively for paging. Then I put a
custom size, initial and max, 100 mb page file on C: and a custom size,
initial and max, 11500 mb on X: My thinking was that I'd never have to
worry about the page file on this computer.

The 100 mb page file showed up in one segment on C: -- this is a keeper.
But the large page file on X: was in six fragments! OK, I backed off on
the size of the X: page file. On X: I set the page file to 0, booted,
manually deleted pagefile.sys, booted, and then allocated a custom initial
and max 10000 mb page file. This showed up in three segments. I tried
again with system managed storage on X: This time I got one segment of
3575 mb, and with a recommendation of 5362 mb. That's where I am now. Of
course, I'd rather have a larger one-segment page file. I'm now at the
mercy of the system, maybe making a fragmented mess on X:, even though
there's nothing else on this partition. Any suggestions?

I ran the 5 stage chkdsk against X: and found no problems.

Another question. On the 750 drive I first created one partition for all
but the 12 gig for X:. Then I created X:. This puts X: on the outside of
drive, right? And the outside goes faster, right? I read this somewhere,
but I'm not sure I believe it. The only way this would help is if the bits
are denser per degree of circle on the outside of the disk. Is that the
way they're made?

Thanks much for your comments.

- Boon

 

[db ´¯`·.. >]

you only need one pagefile
and it will stay contiguous if
it is located in its "own" partition,
preferably on a secondary disk.

in regards to its size, you o.s.
will provide you with a
recommendation.

further, use custom and set it
to a min of 2 and the max of
whatever is recommended
by your o.s. + a couple of
hundred megs for good measure.

set it and forget it, as they say.

--

<)))º>·´¯`·.¸. , . .·´¯`·.. ><)))º>`·.¸¸.·´¯`·.¸.·´¯`·...¸><)))º>

..

 

[John John]

The maximum pagefile size is 4095MB. You can have multiple pagefiles on
multiple spindles, Windows will favour the pagefile on the least busy
spindle. Spindle is just another word for a "physical hard disk". Do
not place multiple pagefiles on separate partitions on the same spindle.

John

 

[Boon]

db,

Thanks for your response.

you only need one pagefile
and it will stay contiguous if
it is located in its "own" partition,
preferably on a secondary disk.

I thought that, too, but that's not the way it turned out. I used XP to
create the pagefile on a secondary drive partition dedicated for the
exclusive use of that pagefile. This was a new drive, and a freshly
formatted partition with nothing else on it. The pagefile that XP
allocated was fragmented as I described in my original post.

- Boon

 

[Boon]

John John,

Thanks for your response.

You wrote that the maximum pagefile size is 4095MB. Did you mean segment
size?

My first try as described in my original post got me a 11500MB pagefile.
The problem was that XP allocated the pagefile in multiple segments.

I would like to allocate a 11500MB pagefile with no fragmentatin -- that
is, in one segment.

- Boon

 

[John John]

By segment size, I assume that mean you fragments. The maximum pagefile
size is still 4095MB, be it in one contiguous file or fragmented, the
maximum file size is still limited to 4095MB. The file may become
fragmented but that will not change the size of the file, the total of
the fragments will add up to the size of the file. Windows 32-bit
cannot possibly use a single 11500MB pagefile, it is simply impossible,
32-bit addressing limits the maximum usable size of each pagefile to
4095 MB.

That is a very large pagefile that you are trying to create, few people
need that much paging area. If you do in fact have such a large paging
requirement then you will have to create multiple pagefiles, you will
need three pagefiles to fill your paging requirements.

John

 

[jorgen]

Windows 32-bit
cannot possibly use a single 11500MB pagefile, it is simply impossible,
32-bit addressing limits the maximum usable size of each pagefile to
4095 MB.

I don't know if it is impossible. The rumours say that Windows 2003 SP1
can go beyond that in a single file. Otherwise you can create several
files to make one big

 

[Boon]

John,

You're right, I have 32 bit addressing, so I'm limited to 4095MB. Thanks.
I now understand what I see (I think).

As a test, I just set virtual memory on my X: partition (on a 2nd drive,
exclusively for pagefile.sys) to 9000MB, expecting to see 2 of something.
Yes, the defragmenter analysis shows 2 fragments for the pagefile. But
Windows Explorer reports pagefile.sys as a 8.79GB file. Evidently Windows
XP keeps track of the multiple fragments and reports it as a single file.
Whether it's one file or two is a semantics issue, not related to system
performance.

Folk wisdom is like this:
- the pagefile should be 1 ½ to 3 times memory
- for best performance give it a small allocation on C:
.... and allocate the rest on a dedicated partition on a 2nd drive
- pagefile fragmentation is bad.

There's a trap here for people like me with 4 gig of memory, and I fell
right into it.

I think I'll stick with what I have, as described in my original post:
100MB on C: and System managed size on X:, currently at 3575MB, which is in
one fragment.

John, thanks for your help. I knew about the 32 bit addressing, but hadn't
applied that thinking to my pagefile issue.

- Boon

 

[John John]

I don't know if it is impossible. The rumours say that Windows 2003 SP1
can go beyond that in a single file. Otherwise you can create several
files to make one big

Interesting. First time I hear of that. It isn't inconceivable that it
may be able to do so, I guess it would or may rely on the 36-bit
addressing capabilities afforded by PAE to do so?

John

 

[John John]

Interesting. Without the /PAE switch in the boot.ini file Windows XP
will not permit the creation of a pagefile larger than 4095MB. With PAE
enabled it will allow for the creation of a larger pagefile and it will
keep it in one fragment. I'm learning new things here so at this point
I am unsure how Windows XP will actually use such a large pagefile. From
the information here it is hard to say exactly but it does confirm what
jorgen said, Server 2003 SP1 can in fact use a larger pagefile with the
use of PAE. Windows XP does not use PAE in the very same way as Server
2003 so as it concerns XP I'm unsure of how it affects pagefile size and
usage, suffice to say at this point that it does permit the cration of a
pagefile larger than 4095MB.

How to overcome the 4,095 MB paging file size limit in Windows
http://support.microsoft.com/kb/237740

In any case, for your Windows XP box with 4GB of installed memory you
almost certainly do not need anywheres that big a pagefile. You can let
Windows manage it and see how big a file it actually uses. Or you can
use perfmon to monitor and log its usage then make a decision based on
the actual data obtained.

John

 

[John John]

I don't know if it is impossible. The rumours say that Windows 2003 SP1
can go beyond that in a single file. Otherwise you can create several
files to make one big

Indeed, it does so with PAE, it is confirmed here:

How to overcome the 4,095 MB paging file size limit in Windows
http://support.microsoft.com/kb/237740

Thanks for the tip.

John

 

[db ´¯`·.. >]

well, it is likely because
you are trying to create
a multi gigabyte pagefile
when windows is actually
using a small portion of
it.

unless you are calculating
the rotational gravitational
pull of the moon on a minute
by minute basis, then simplistically
your o.s. and the off-the-shelf
software will never call for more
that 300-600 megabytes of pagefile,
even if you have a 2, 4 or 16
gigabyte of pagefile.

in some cases, when there
is a lot of ram, the pagefile
is not even required.

you might want to monitor
the amount of pagefile you
actually use on a regular basis

third party programs like the one
from are quick and helpful:

http://www.amsn.ro/

in addition, you may want
to review a switch
called PAE.

for more info:

http://search.microsoft.com/results.aspx?mkt=en-US&form=MSHOME&setlang=en-US&q=large+ram+systems

--

<)))º>·´¯`·.¸. , . .·´¯`·.. ><)))º>`·.¸¸.·´¯`·.¸.·´¯`·...¸><)))º>

..

 

[Colin Barnhorst]

x86 Windows supports up to 16 page files of 4GB. x86 with PAE and x64
Windows support 16TB page files.

 

[John John]

Yes, indeed. Here is the official word:

Maximum paging file size

Maximum paging file size is limited as follows:

x86 x64 IA-64
Max. size pagefile 4 gigabytes 16 terabytes 32 terabytes
Max. No. pagefiles 16 16 16
Total paefile size 64 gigabytes 256 terabytes 512 terabytes

Note: When the Physical Address Extension (PAE) option is enabled for
an x86-based processor, you can set the paging file size to a maximum of
16 terabytes (TB).

APPLIES TO

• Microsoft Windows Server 2003, Standard x64 Edition
• Microsoft Windows Server 2003, Datacenter x64 Edition
• Microsoft Windows Server 2003, Enterprise x64 Edition
• Microsoft Windows Server 2003, Datacenter Edition for Itanium-Based
systems
• Microsoft Windows Server 2003, Enterprise Edition for Itanium-based
Systems
• Microsoft Windows Server 2003, Standard Edition (32-bit x86)
• Microsoft Windows Server 2003, Datacenter Edition (32-bit x86)
• Microsoft Windows Server 2003, Enterprise Edition (32-bit x86)
• Microsoft Windows Server 2003, Web Edition
• Microsoft Windows XP Professional x64 Edition
• Microsoft Windows XP for Itanium-based Systems Version 2003
• Microsoft Windows XP Professional 64-Bit Edition (Itanium)
• Microsoft Windows XP Professional
• Microsoft Windows XP Home Edition
• Microsoft Windows XP Tablet PC Edition
• Microsoft Windows 2000 Server
• Microsoft Windows 2000 Advanced Server
• Microsoft Windows 2000 Professional Edition
• Microsoft Windows 2000 Datacenter Server

http://support.microsoft.com/kb/254649

John

 

[John John]

third party programs like the one
from are quick and helpful:

http://www.amsn.ro/

That is not quick and helpful unless you think that quickly bugging up a
computer is helpful! These programs are nothing but snake oil and they
actually harm memory management and cause problems were none previously
existed. Absolute crap is the only way to describe these "optimizers"
and the one you propose is no different, useless trash to say the very
least!

John