"John Navas" <(E-Mail Removed)> wrote in message
news:(E-Mail Removed)...
> To change the Windows XP Paging File size, search Help and Support for
> "Change the size of the virtual memory paging file"
>
> My advice to those genuinely interested in setting an optimum Windows XP
> Paging File size (Virtual Memory) is to start with:
> Initial size (MB): 2 [minimum]
> Maximum size (MB): 2048 [safe large limit]
> Run a _heavy_ mix of however you use your computer for a period of at
> least a few hours (not just a few minutes).
> If and when Windows needs more virtual memory you will see popup
> messages as that memory gets allocated.
> Then go to Virtual Memory again:
> Initial size (MB): round Current value up to next power of two
> (example: change current value of 326 to 512)
> Maximum size (MB): double the Initial value you just set
> (1024 in above example)
> Lastly, download, install and run Sysinternal PageDefrag
> <http://technet.microsoft.com/en-us/sysinternals/bb897426.aspx>
>
> From time to time check Virtual Memory to see if your Initial size has
> been exceeded.
>
> --
> Best regards, You are cordially invited to participate in the
> John official ThinkPad group news:comp.sys.laptops.thinkpad

Do what you want, but I have never had a problem allowing
Win98/ME/XP/Vista/7 manage its own page file.

SC Tom

<see below>
---
Leonard Grey
Errare humanum est

SC Tom wrote:
>
> "John Navas" <(E-Mail Removed)> wrote in message
> news:(E-Mail Removed)...
>> To change the Windows XP Paging File size, search Help and Support for
>> "Change the size of the virtual memory paging file"
>>
>> My advice to those genuinely interested in setting an optimum Windows XP
>> Paging File size (Virtual Memory) is to start with:
>> Initial size (MB): 2 [minimum]
>> Maximum size (MB): 2048 [safe large limit]
>> Run a _heavy_ mix of however you use your computer for a period of at
>> least a few hours (not just a few minutes).
>> If and when Windows needs more virtual memory you will see popup
>> messages as that memory gets allocated.
>> Then go to Virtual Memory again:
>> Initial size (MB): round Current value up to next power of two
>> (example: change current value of 326 to 512)
>> Maximum size (MB): double the Initial value you just set
>> (1024 in above example)
>> Lastly, download, install and run Sysinternal PageDefrag
>> <http://technet.microsoft.com/en-us/sysinternals/bb897426.aspx>
>>
>> From time to time check Virtual Memory to see if your Initial size has
>> been exceeded.
>>
>> --
>> Best regards, You are cordially invited to participate in the
>> John official ThinkPad group news:comp.sys.laptops.thinkpad
>
> Do what you want, but I have never had a problem allowing
> Win98/ME/XP/Vista/7 manage its own page file.
>
> SC Tom
>

I'm with Tom. The optimal value for virtual memory is "System Managed Size."

LG

**See my comments and questions below**

>On Wed, 28 Oct 2009 14:29:46 -0400, Leonard Grey <(E-Mail Removed)> wrote:

><see below>
>---
>Leonard Grey
>Errare humanum est
>
>SC Tom wrote:
>>
>> "John Navas" <(E-Mail Removed)> wrote in message
>> news:(E-Mail Removed)...
>>> To change the Windows XP Paging File size, search Help and Support for
>>> "Change the size of the virtual memory paging file"
>>>
>>> My advice to those genuinely interested in setting an optimum Windows XP
>>> Paging File size (Virtual Memory) is to start with:
>>> Initial size (MB): 2 [minimum]
>>> Maximum size (MB): 2048 [safe large limit]
>>> Run a _heavy_ mix of however you use your computer for a period of at
>>> least a few hours (not just a few minutes).
>>> If and when Windows needs more virtual memory you will see popup
>>> messages as that memory gets allocated.
>>> Then go to Virtual Memory again:
>>> Initial size (MB): round Current value up to next power of two
>>> (example: change current value of 326 to 512)
>>> Maximum size (MB): double the Initial value you just set
>>> (1024 in above example)
>>> Lastly, download, install and run Sysinternal PageDefrag
>>> <http://technet.microsoft.com/en-us/sysinternals/bb897426.aspx>
>>>
>>> From time to time check Virtual Memory to see if your Initial size has
>>> been exceeded.
>>>
>>> --
>>> Best regards, You are cordially invited to participate in the
>>> John official ThinkPad group news:comp.sys.laptops.thinkpad
>>
>> Do what you want, but I have never had a problem allowing
>> Win98/ME/XP/Vista/7 manage its own page file.
>>
>> SC Tom
>>
>
>I'm with Tom. The optimal value for virtual memory is "System Managed Size."
>
>LG

There seems to be little or no concensus AFAICT.

For example, on a 32 bit WinXP Pro computer with 4 gig of RAM, any
setting other than No Page File has resulted, in my case, with an
intermittent event ID 51 "An error was detected on device
\Device\Harddisk4\D during a paging operation." The error message is
not limited to just one drive; it may refer, randomly, to any
installed hard drive or USB drive.

I also find the following post reasonable (from a 2008 MS TechNet
thread):

>>cut here for text of thread<<

Here is a copy of an issue regarding Microsoft's Windows XP Memory
Manager that might make it clear to you why you would want to kill the
page file. In addition to this excellent reason, another reason would
be that kliling the page file would increase WIndows XP's performance
and reduce the ewar and tear on your hard drive thus expanding its
expecrted life span for you as well. Here is the quote:


To all concerned,

This is Brian Dowding writing to you regarding a Microsoft Windows
product design flaw. SRZ041029003929 is my old case number regarding
this issue. This email, I believe, is much more clear and concise.


The Problem:
------------
The algorhythms designed to assist manage memory resources in Windows
Environments based on NT code in 32-bit environments have essentially
remained unchanged in the last 11 years and were not likely made with
the realistic expectation that an average home user would have more
than 1GB of physical RAM in its lifecycle of usage in Windows
products.

Here-in lies the current problem with memory management in Windows
NT/2000 Pro & Advanced Server/XP Home, Pro, Tablet, and Media Center
Editions/ as well as any other version of Windows that is 32-bit and
based on NT code.

In a 32-bit environment, memory addressing is limited to a 4GB space.
This applies to the sum of Physical RAM installed PLUS the size of the
page file in place.

When Windows is set to manage the size of the page file automatically,
then the page file is calculated by a simple algorhythm of "Installed
RAM Size" multiplied by 1.5 equals the "page file minimum size";
"Installed RAM size" multiplied by 3 equals the "page file maximum
size".

As a security feature, if one has the page file set manually or turned
off AND the settings are drastically different enough from what the
automatic management would have set it to, Windows can pay lip service
to the users desired settings while managing the page file on its own
instead. This can be demonstrated by setting Windows to delete the
page file on system shutdown, turning off the page file altogether,
rebooting, and using a third party tool like "Space Monger" to analyze
the contents of the C:\ or primary partition on which Windows is
installed, one will sometimes find a page file has been recreated by
Windows anyway, and their settings telling Windows not to use one
intact.

None of this is a problem IF the user has up to or less than 1GB of
RAM. Once you install more than 1GB of RAM, this becomes a serious
problem to system stability.

The following are examples of the algorhythm at work in a 1GB, 2GB,
3GB, and 4GB installed RAM environments:



A) A system with 1GB of RAM installed:

installed RAM = 1GB
page file minimum = 1.5GB
total memory space used with a minimum sized page file = 2.5GB
page file maximum = 3GB
total memory space used with a maximum sized page file = 4GB
does this exceed the 4GB limitation of 32-bit addressing? = No



B) A system with 2GB of RAM installed:

installed RAM = 2GB
page file minimum = 3GB
total memory space used with a minimum sized page file = 5GB
page file maximum = 6GB
total memory space used with a maximum sized page file = 8GB
does this exceed the 4GB limitation of 32-bit addressing? = Yes



C) A system with 3GB of RAM installed:

installed RAM = 3GB
page file minimum = 4.5GB
total memory space used with a minimum sized page file = 7.5GB
page file maximum = 9GB
total memory space used with a maximum sized page file = 12GB
does this exceed the 4GB limitation of 32-bit addressing? = Yes



D) A system with 4GB of RAM installed:

installed RAM = 4GB
page file minimum = 6GB
total memory space used with a minimum sized page file = 10GB
page file maximum = 12GB
total memory space used with a maximum sized page file = 16GB
does this exceed the 4GB limitation of 32-bit addressing? = Yes



Now, in theory, of course, turning off the page file should be a
solution for those with 4GB of RAM. Also, in theory, manual settings
adjusting for the page file size so that the sum of the installed RAM
and page file sizes do not exceed 4GB should solve this. However, in
practice this is not a reliable remedy as Windows can and has been
known to pay lip service to the manual settings while ignoring them
and managing a hidden page file under the hood, so to speak, which
reintroduces system instability. This is because such manual
adjustments would create a significant difference between what the end
user sets as a page file versus what Windows automatic settings would
have been. Additionally, the larger the installed base of RAM, the
larger the gap becomes.

The /3GB switch has been suggested in the past. This obviously does
not address the issue as it simply extends single application resource
allotment to RAM maximums from 2GB to 3GB. Despite this, 4GB remains
the total addressable limit in a 32-bit environment. The problem
remains the same whether or not one uses the /3GB switch.


How Can This Problem Be Solved:
-------------------------------
I propose a simple and elegant solution. The existing algorhythm is:

---------------------------------------------
Minimum Page File Size = Installed RAM x 1.5
Maximum Page File Size = Installed RAM x 3
---------------------------------------------

It should be replaced by the following algorhythm:

---------------------------------------------
Page File Size = 4GB minus Installed RAM Size
---------------------------------------------

This would address all issues.

I would like to reiterate that this problem only affects 32-bit
Windows environments with an installed base of RAM over 1GB. Thus,
for example, this affects nearly 95% of the worldwide server base -
not just home users with over 1GB of RAM.

How Can This Be Exploited:
--------------------------

One example would be a someone that managed to distribute a simple
single change in the registry that turns the page file on for Windows
to manage. Most servers, for example again, have 4GB installed RAM
with no page file - and is 32-bit. Having the page file on makes
Windows try to manage 16GB of memory space when it can only manage
4GB.

Just imagine the widespread system crashes - and who would check for a
working page file? Technically this would not even be a virus as just
telling Windows to work as designed is the culprit.

Conclusion
----------
In short, this is a design flaw.

I would like to see a HOTFIX developed, and I wouldn't mind seeing my
name appear on the HOTFIXs official title.

Thanks for reviewing this submission,

Brian Dowding

>>stop cutting here<<

I would genuinely be interesting in knowing what is the correct
answer.

Duncan

(E-Mail Removed) wrote:
> **See my comments and questions below**
>
>> On Wed, 28 Oct 2009 14:29:46 -0400, Leonard Grey <(E-Mail Removed)> wrote:
>
>> <see below>
>> ---
>> Leonard Grey
>> Errare humanum est
>>
>> SC Tom wrote:
>>> "John Navas" <(E-Mail Removed)> wrote in message
>>> news:(E-Mail Removed)...
>>>> To change the Windows XP Paging File size, search Help and Support for
>>>> "Change the size of the virtual memory paging file"
>>>>
>>>> My advice to those genuinely interested in setting an optimum Windows XP
>>>> Paging File size (Virtual Memory) is to start with:
>>>> Initial size (MB): 2 [minimum]
>>>> Maximum size (MB): 2048 [safe large limit]
>>>> Run a _heavy_ mix of however you use your computer for a period of at
>>>> least a few hours (not just a few minutes).
>>>> If and when Windows needs more virtual memory you will see popup
>>>> messages as that memory gets allocated.
>>>> Then go to Virtual Memory again:
>>>> Initial size (MB): round Current value up to next power of two
>>>> (example: change current value of 326 to 512)
>>>> Maximum size (MB): double the Initial value you just set
>>>> (1024 in above example)
>>>> Lastly, download, install and run Sysinternal PageDefrag
>>>> <http://technet.microsoft.com/en-us/sysinternals/bb897426.aspx>
>>>>
>>>> From time to time check Virtual Memory to see if your Initial size has
>>>> been exceeded.
>>>>
>>>> --
>>>> Best regards, You are cordially invited to participate in the
>>>> John official ThinkPad group news:comp.sys.laptops.thinkpad
>>> Do what you want, but I have never had a problem allowing
>>> Win98/ME/XP/Vista/7 manage its own page file.
>>>
>>> SC Tom
>>>
>> I'm with Tom. The optimal value for virtual memory is "System Managed Size."
>>
>> LG
>
> There seems to be little or no concensus AFAICT.
>
> For example, on a 32 bit WinXP Pro computer with 4 gig of RAM, any
> setting other than No Page File has resulted, in my case, with an
> intermittent event ID 51 "An error was detected on device
> \Device\Harddisk4\D during a paging operation." The error message is
> not limited to just one drive; it may refer, randomly, to any
> installed hard drive or USB drive.
>
> I also find the following post reasonable (from a 2008 MS TechNet
> thread):
>
>>> cut here for text of thread<<
>
> Here is a copy of an issue regarding Microsoft's Windows XP Memory
> Manager that might make it clear to you why you would want to kill the
> page file. In addition to this excellent reason, another reason would
> be that kliling the page file would increase WIndows XP's performance
> and reduce the ewar and tear on your hard drive thus expanding its
> expecrted life span for you as well. Here is the quote:
>
>
> To all concerned,
>
> This is Brian Dowding writing to you regarding a Microsoft Windows
> product design flaw. SRZ041029003929 is my old case number regarding
> this issue. This email, I believe, is much more clear and concise.
>
>
> The Problem:
> ------------
> The algorhythms designed to assist manage memory resources in Windows
> Environments based on NT code in 32-bit environments have essentially
> remained unchanged in the last 11 years and were not likely made with
> the realistic expectation that an average home user would have more
> than 1GB of physical RAM in its lifecycle of usage in Windows
> products.
>
> Here-in lies the current problem with memory management in Windows
> NT/2000 Pro & Advanced Server/XP Home, Pro, Tablet, and Media Center
> Editions/ as well as any other version of Windows that is 32-bit and
> based on NT code.
>
> In a 32-bit environment, memory addressing is limited to a 4GB space.
> This applies to the sum of Physical RAM installed PLUS the size of the
> page file in place.
>
> When Windows is set to manage the size of the page file automatically,
> then the page file is calculated by a simple algorhythm of "Installed
> RAM Size" multiplied by 1.5 equals the "page file minimum size";
> "Installed RAM size" multiplied by 3 equals the "page file maximum
> size".
>
> As a security feature, if one has the page file set manually or turned
> off AND the settings are drastically different enough from what the
> automatic management would have set it to, Windows can pay lip service
> to the users desired settings while managing the page file on its own
> instead. This can be demonstrated by setting Windows to delete the
> page file on system shutdown, turning off the page file altogether,
> rebooting, and using a third party tool like "Space Monger" to analyze
> the contents of the C:\ or primary partition on which Windows is
> installed, one will sometimes find a page file has been recreated by
> Windows anyway, and their settings telling Windows not to use one
> intact.
>
> None of this is a problem IF the user has up to or less than 1GB of
> RAM. Once you install more than 1GB of RAM, this becomes a serious
> problem to system stability.
>
> The following are examples of the algorhythm at work in a 1GB, 2GB,
> 3GB, and 4GB installed RAM environments:
>
>
>
> A) A system with 1GB of RAM installed:
>
> installed RAM = 1GB
> page file minimum = 1.5GB
> total memory space used with a minimum sized page file = 2.5GB
> page file maximum = 3GB
> total memory space used with a maximum sized page file = 4GB
> does this exceed the 4GB limitation of 32-bit addressing? = No
>
>
>
> B) A system with 2GB of RAM installed:
>
> installed RAM = 2GB
> page file minimum = 3GB
> total memory space used with a minimum sized page file = 5GB
> page file maximum = 6GB
> total memory space used with a maximum sized page file = 8GB
> does this exceed the 4GB limitation of 32-bit addressing? = Yes
>
>
>
> C) A system with 3GB of RAM installed:
>
> installed RAM = 3GB
> page file minimum = 4.5GB
> total memory space used with a minimum sized page file = 7.5GB
> page file maximum = 9GB
> total memory space used with a maximum sized page file = 12GB
> does this exceed the 4GB limitation of 32-bit addressing? = Yes
>
>
>
> D) A system with 4GB of RAM installed:
>
> installed RAM = 4GB
> page file minimum = 6GB
> total memory space used with a minimum sized page file = 10GB
> page file maximum = 12GB
> total memory space used with a maximum sized page file = 16GB
> does this exceed the 4GB limitation of 32-bit addressing? = Yes
>
>
>
> Now, in theory, of course, turning off the page file should be a
> solution for those with 4GB of RAM. Also, in theory, manual settings
> adjusting for the page file size so that the sum of the installed RAM
> and page file sizes do not exceed 4GB should solve this. However, in
> practice this is not a reliable remedy as Windows can and has been
> known to pay lip service to the manual settings while ignoring them
> and managing a hidden page file under the hood, so to speak, which
> reintroduces system instability. This is because such manual
> adjustments would create a significant difference between what the end
> user sets as a page file versus what Windows automatic settings would
> have been. Additionally, the larger the installed base of RAM, the
> larger the gap becomes.
>
> The /3GB switch has been suggested in the past. This obviously does
> not address the issue as it simply extends single application resource
> allotment to RAM maximums from 2GB to 3GB. Despite this, 4GB remains
> the total addressable limit in a 32-bit environment. The problem
> remains the same whether or not one uses the /3GB switch.
>
>
> How Can This Problem Be Solved:
> -------------------------------
> I propose a simple and elegant solution. The existing algorhythm is:
>
> ---------------------------------------------
> Minimum Page File Size = Installed RAM x 1.5
> Maximum Page File Size = Installed RAM x 3
> ---------------------------------------------
>
> It should be replaced by the following algorhythm:
>
> ---------------------------------------------
> Page File Size = 4GB minus Installed RAM Size
> ---------------------------------------------
>
> This would address all issues.
>
> I would like to reiterate that this problem only affects 32-bit
> Windows environments with an installed base of RAM over 1GB. Thus,
> for example, this affects nearly 95% of the worldwide server base -
> not just home users with over 1GB of RAM.
>
> How Can This Be Exploited:
> --------------------------
>
> One example would be a someone that managed to distribute a simple
> single change in the registry that turns the page file on for Windows
> to manage. Most servers, for example again, have 4GB installed RAM
> with no page file - and is 32-bit. Having the page file on makes
> Windows try to manage 16GB of memory space when it can only manage
> 4GB.
>
> Just imagine the widespread system crashes - and who would check for a
> working page file? Technically this would not even be a virus as just
> telling Windows to work as designed is the culprit.
>
> Conclusion
> ----------
> In short, this is a design flaw.
>
> I would like to see a HOTFIX developed, and I wouldn't mind seeing my
> name appear on the HOTFIXs official title.
>
> Thanks for reviewing this submission,
>
> Brian Dowding
>
>>> stop cutting here<<
>
> I would genuinely be interesting in knowing what is the correct
> answer.

32-bit NT operating systems support up to 16 paging files, with the
normal kernel the maximum size of each paging file is 4,095MB, for a
total paging area of 64GB. The use of the PAE kernel allows the 32-bit
operating system to create paging files of 16 terabytes each, for a
whopping total paging area of 256 terabytes!

Remember that each and every process on 32-bit operating systems has its
own private address space with up to 4GB of available space. The
operating system needs to reserve space for its own use so it divvies up
the 4GB space 50/50 and reserves 2GB for its use and makes the other 2GB
available to the process. The process sees the whole 4GB that is
available to it but the system makes its 2GB share of the space visible
to the process so the process can only use the remaining 2GB of
available space. The use of 4-gigabyte tuning (4GT) (aka /3GB switch)
changes the allotment of the available space and limits the operating
system space to 1GB and makes the remaining 3GB available to the
process. In theory if you were running 100 heavy duty applications each
consuming its full available private space you would need a paging area
of 200GB, 300GB if the applications were making full use of the
4-gigabyte tuning /3GB mechanism.

John