Virtual memory problem? - PF usage grows, not freed when apps clos

[Guest]

I'm running XP w/ SP2 (on a Tablet PC in case that matters) w/ 512 MB
(currently - I'm upgrading memory soon, largely because of this problem :-(
).

The system is fine at startup - memory usage normal, and performance is
fine, so I don't think that my problem is an issue of background tasks loaded
at startup using up excessive amounts of memory.

After I've been using the system for a while, the system will slow to a
crawl, and becomes almost completely unresponsive – moderately high CPU use,
with some disk activity. When I use the Task Monitor to look at what's going
on, in the Performance tab I see that the Page File usage is 800 or 900 MB.

Fine, so I close any/ALL open apps. The PF usage only goes down to 725 to
850 MB, and the system is still extremely slow, with some disk access going
on, and high CPU use.

(None of the processes seem to be abnormally large by them selves, so I
don't think that I'm seeing the TCServe.exe memory leak.)

1) Is it normal that PF usage a) continues to grow *indefinitely* over time,
and b) is not freed when all apps are closed, resulting in continued degraded
performance after XP has been running for a day or so?

2) If not (hopefully! , what is the best (quickest) way to determine
whether I'm seeing a memory leak in an app, or some strange XP memory
management behavior?

Thanks for any help or suggestions!
Gregg

 

[Gerry Cornell]

On the Processes tab in Task Manager you may get more information if you
select View, Select Columns and check before Virtual Memory Size.

What programmes and versions are you using, which use significant amounts of
virtual memory? Any old programmes pre Windows 98 era?

Have you installed any "memory boosters"?

--

Hope this helps.

Gerry
~~~~
FCA
Stourport, England

Enquire, plan and execute
~~~~~~~~~~~~~~~~~~~


What anti-spyware arrangements?

 

[Guest]

On the Processes tab in Task Manager you may get more information if you
select View, Select Columns and check before Virtual Memory Size.

Cool - thanks for the tip. I'll try that, and see if that shows anthing
interesting.

What programmes and versions are you using, which use significant amounts of
virtual memory? Any old programmes pre Windows 98 era?

No old programs.

One thing I find interesting is that once the problem occurs, I can close
ALL applications, and be running ZERO apps, and the PF memory usage continues
to be excessive 725 - 850 MB.

Programs I use pretty consistently (and typcially simultaneously) include
Windows Explorer, MS Outlook, MS Word, and Firefox.

Have you installed any "memory boosters"?
Nope.

What anti-spyware arrangements?

Twice weekly scans using McAfee Anti-spyware. Also, since the system is fine
at startup and the problem only occurs after a day or so, I don't think it's
a spyware problem.

Me too . Thanks for the reply, and any ideas.

cheers,
Gregg

 

[Guest]

What programmes and versions are you using, which use significant amounts of
virtual memory? Any old programmes pre Windows 98 era?

In addition to the apps I mentioned, I also have running Zone Alarm (5.1),
Office Clipboard, Outlook system tray tool, Synaptics point device tool,
Wireless Connection monitor tool, LAN LAN connection monitor tool, Tablet &
pen settings tool.

And of course an assortment of background processes some from XP a few from
other apps -all of which appear to be small.

 

[Gerry Cornell]

You could try disabling each item in turn to see if you can identify which,
if any causes the problem.

You said "Fine, so I close any/ALL open apps. The PF usage only goes down to
725 to
850 MB, and the system is still extremely slow, with some disk access going
on, and high CPU use."

What Process ID is generating significant CPU usage once open applications
are closed.

--

Hope this helps.

Gerry
~~~~
FCA
Stourport, England

Enquire, plan and execute
~~~~~~~~~~~~~~~~~~~

 

[Guest]

What Process ID is generating significant CPU usage once open applications
are closed.

Yeah, good question. Unfortunately, the system is unresponsive while the CPU
activity happens, to the point that the task monitor won't load. So it's been
tough to find out. Because it prevents even the TM from running, I suspect a
system process, but can't be sure yet.

To clarify - the unexplained CPU activity is only present once the 'tipping
point' is reached, then exists for a while, then the CPU activity goes away.
But the large PF usage and slow performance remain.

I've been avoiding the problem by rebooting daily. I'll try keeping the TM
running, to see if that allows me to access it when the CPU ativity occurs.

Are there by any chance utilities that I could run after I close all the
apps to see by whom (by what app) the PF memory is being used? Would the PF
monitor app tell me this? I haven't taken the time to find and try that yet
(I just heard about it today).

 

[Gerry Cornell]

Let's try an other approach.

Download, install an run HijackThis an anti-spyware detection programme:
http://tomcoyote.com/hjt/

Post a copy of the HijackThis log to me at (e-mail address removed)
removing the obvious


--

Hope this helps.

Gerry
~~~~
FCA
Stourport, England

Enquire, plan and execute
~~~~~~~~~~~~~~~~~~~

 

[Guest]

OK, I'll do that when I get a moment - it can't hurt to rule this out.

However, as I've mentioned, since the system runs fine for a day or so after
booting, it doesn't seem likely (to me at least that the problem I
described results from a spyware issue. (Doesn't mean there might not be some
there somewhere though .

Does anyone have any other info related to the issue of memory use that
starts fine, grows steadily, and then eventually becomes excessive - and then
is not reclaimed when the apps are closed?

thanks!
Gregg

 

[Guest]

Let's try an other approach.

Download, install an run HijackThis an anti-spyware detection programme:
http://tomcoyote.com/hjt/

Post a copy of the HijackThis log to me at (e-mail address removed)
removing the obvious

OK, I've sent the log.

 

[Gerry Cornell]

What particular McAfee, Zone Alarm and Firefox programmes and precise
versions do you have installed? I ask these because these programmes have
been known to problematic at times in the past.

How does Roboforms feature as a user of virtual memory? Large or small? I am
not familiar with how it works in practice. When does accumulated
information get released?

Is there any pattern to the problem? Does the problem correspond to any
Scheduled Task?

How much RAM memory and what are your virtual memory settings?

--

Hope this helps.

Gerry
~~~~
FCA
Stourport, England

Enquire, plan and execute
~~~~~~~~~~~~~~~~~~~

 

[Guest]

I've filled in the answers after the questions below.

What particular McAfee, Zone Alarm and Firefox programmes and precise
versions do you have installed? I ask these because these programmes have
been known to problematic at times in the past.

Zone Alarm Security Suite 5.1.001
McAfee AntiSpyware 1.10.0163.0
Firefox 1.5

How does Roboforms feature as a user of virtual memory? Large or small? I am
not familiar with how it works in practice. When does accumulated
information get released?

Roboform 's VM use should be small. It stores short text strings to be
filled into forms in small files on disk - 1 per each site for which info is
saved, plus a general user file. As far as I know, it doesn't accumulate
additional info, esp of any significant size.

Is there any pattern to the problem? Does the problem correspond to any
Scheduled Task?

I typically run serval apps at once. It appeats that once the amount of PF
memory needed exceeds the available physical memory, the problem occurs.

I have not been able to connect the problem with any particular interactive
or scheduled task.

How much RAM memory and what are your virtual memory settings?

512 MB RAM.

Ah ... virtual memory paging file size is set to only 753 MB. This seems WAY
too small to me... (I've seen PF usage #s reported by TM of 900 MB - 1 GB.)

Do you agree? Do you recall off hand a better size/ratio, or shall I track
this down.

thanks,
Gregg

 

[Gerry Cornell]

Has the Tablet Memory leak fix been applied?
http://snipurl.com/kq68

--

Hope this helps.

Gerry
~~~~
FCA
Stourport, England

Enquire, plan and execute
~~~~~~~~~~~~~~~~~~~

 

[Gerry Cornell]

This is a leading Article on Virtual Memory:
http://aumha.org/win5/a/xpvm.htm

--

Hope this helps.

Gerry
~~~~
FCA
Stourport, England

Enquire, plan and execute
~~~~~~~~~~~~~~~~~~~

 

[Guest]

Thanks. Could a PF too small initially be causing at least part of the
problem - the behavior once the PF size gets larger than the current
(initial) size setting?

Of course I still wonder why the PF usage reported by TM doesn't go down
again (much) once all the apps are closed. ..

cheers,
Gregg

 

[Gerry Cornell]

Have you set a minimum and maximum or have you let Windows manage the
pagefile? Is the pagefile in the same partition as your Windows operating
system?

What is the answer to my suggestion about Tablet?

--

Hope this helps.

Gerry
~~~~
FCA
Stourport, England

Enquire, plan and execute
~~~~~~~~~~~~~~~~~~~

 

[Guest]

Have you set a minimum and maximum or have you let Windows manage the
pagefile?

I had not touched these setting, have left them as they came from the
factory, presumably from the Windows installation. Looking at the settings,
I see Custom size, Initial size = 756, Max size = 1512.

Is the pagefile in the same partition as your Windows operating
system?
Yes.

What is the answer to my suggestion about Tablet?

Sorry, I had missed that question. The answer is no, this patch has not been
installed - separately/manually at least. I'm not certain whether it has
been installed automatically. Is there a way to check whether this one item
has been installed?

By the way, this is the patch I referred to when I mentioned that the
TCserver process (itself at least) is not growing abnormally large. (it
reaches about 20 MB).

I spoke with the gentleman from whose blog I first heard about this memory
leak (http://blogs.msdn.com/jonathanh/archive/2005/07/05/435922.aspx) and he
said (as I understood it) that TCServer process size would grow if I was
having this problem. Is that correct?

So nope, I have not installed this patch. Do you recommend installing it?

thanks,
Gregg

 

[Gerry Cornell]

I would download and install the patch.

With regard to virtual memory remove the maximum so that there is no
maximum.


--

Hope this helps.

Gerry
~~~~
FCA
Stourport, England

Enquire, plan and execute
~~~~~~~~~~~~~~~~~~~

 

[Guest]

I would download and install the patch.

With regard to virtual memory remove the maximum so that there is no
maximum.

OK, I have installed the patch, and updated the VM settings.

(For now at least, I set the max to 3 GB - I wasn't comfortable with NO
limit . I also increased the minimum to 1.5 GB, given that this system has
'only' 512 of physical RAM at this point, and I tend to run 4 or 5 apps
simulataneously + WIN XP.)

I will now run without rebooting for a a few days (or until the problem
resurfaces , and see if the problem recurs. I'll post again if it does.
If it doesn't happen again, I'll try to remember to post an update in a few
days just to confirm

Thanks very much for the suggestions (and questions .

By the way, sorry to say that I have no idea what "FCA" stands for. If you
don't mind my asking do you work for (or contracted by) Microsoft?

regards,
Gregg

 

[deebs]

:




OK, I have installed the patch, and updated the VM settings.

(For now at least, I set the max to 3 GB - I wasn't comfortable with NO
limit . I also increased the minimum to 1.5 GB, given that this system has
'only' 512 of physical RAM at this point, and I tend to run 4 or 5 apps
simulataneously + WIN XP.)

I will now run without rebooting for a a few days (or until the problem
resurfaces , and see if the problem recurs. I'll post again if it does.
If it doesn't happen again, I'll try to remember to post an update in a few
days just to confirm

Thanks very much for the suggestions (and questions .

By the way, sorry to say that I have no idea what "FCA" stands for. If you
don't mind my asking do you work for (or contracted by) Microsoft?

regards,
Gregg

Ok, this is something I am interested in too.

But! first of all some background: a 4000+ AMD thingy, a Gig of RAM, 4
by 250 Gig SATAs.

I run Adobe Photoshop and Illustrator and these, of course, have their
own dedicated temporary memory management structure.

I used to have PF on a separate platter but now am trying this
arrangement for a week or so:
- PF on C: system managed by Windows (XP Pro SP2 & all recent updates)
- another PF on F: set to Max = Min = 1533MB (as recommended by Windows
for PF on C

I've recently been using Doug's PF Usage monitor utility and the
previous arrangement of one PF set to Max = Min = 2.5 Gig to compliment
empirical observations about how PF arrangement performs.

Previous arrangement showed 55 MB PF implying more RAM required
according to the utility.

This neer arrangement (system managed PF of C:, Max = Min = 1533 MB on
F is returning stuff like 28 MB, 25 MB of C: and F: page files.

At startup today it showed PFU at 8 MB and 12 MB on C: and F:

Conclusion?
It's too early to tell but the system is far zippier than previous
arrangement and looks good.

Even rotations in Photoshop seem to zip suggesting (at this stage) that
photoshop is managing well where the system directs temporary management
to PF on C: or F: (or both?) as required

Any more observers out there with tales to tell?

 

[Gerry Cornell]

Gregg

I n recent years there have been countless arguments over pagefile ( also
called virtual memory ) settings. Some believe that the user should leave
Windows to manage virtual memory. Others believe in setting just a minimum
and others in having a minimum and a maximum. Some will set the minimum and
maximum at the same figure. Some believe in having a partition dedicated to
hold only the pagefile, whilst others are opposed to that idea.

There are several pitfalls for the unwary, particularly if disk space is at
a premium. Setting a too small maximum can result in the system crashing if
the pagefile needs to expand beyond the maximum permitted. However, if there
is little extra disk space available the setting of a greater maximum may
make little difference. This can mean loss of data, which may or may not be
important. There will of course be low memory warnings before this point is
reached.

If you put the page file in it's own partition it is generally recognised
that this is best placed in the first partition on the second hard drive. If
you have a dedicated partition you should retain a small pagefile in the
same partition as the windows operating system to satisfy demands from the
system for virtual memory. For certain operations the system needs virtual
memory rather than ram memory.

It is desirable to try to use RAM memory rather than virtual memory. Reading
from and writing to ram memory is quicker than to and from virtual memory.
Therefore where the Commit Charge is regularly greater than the available
RAM memory it is desirable to add more memory up to the maximum allowed by
the motherboard. Setting a maximum size for the pagefile needs to be based
on peak demands, which can be monitored using Task Manager and / or a
specialised utility. However, it is also wise to have a generous cushion for
unexpected demands. However, as Alex Nichol commented in the article posted
earlier the greater the RAM the system is able to access the less will be
it's need for virtual memory. The great majority of informed users of XP now
agree that setting a maximum, which is a multiplier of RAM memory is not a
sensible approach.


F= Fellow and I am a member of this professional body:
http://www.ifac.org/About/MemberDetails.tmpl?MemID=5456514

You are the first person to ask in 5 years.

I do not work for Microsoft. Trying to help with computer problems is a
hobby much as others choose to do crosswords.

--

Regards.

Gerry
~~~~
FCA
Stourport, England

Enquire, plan and execute
~~~~~~~~~~~~~~~~~~~