Fixing a memory problem?

[Terry Pinnell]

I've been getting an increasing number of crashes, which I now think
indicate some sort of memory failure. The messages are about not finding
addresses, 'exceptions', invalid instructions, etc.

How can I set about methodically isolating the cause and, more important,
repairing it please?

This is a MESH 'GTS Xtreme', Quad Core Q9450 2.66 GHz, 4 GB DDR2 667 MHz.
I can provide more info if that will help.

 

[glee]

I've been getting an increasing number of crashes, which I now think
indicate some sort of memory failure. The messages are about not
finding
addresses, 'exceptions', invalid instructions, etc.

How can I set about methodically isolating the cause and, more
important,
repairing it please?

This is a MESH 'GTS Xtreme', Quad Core Q9450 2.66 GHz, 4 GB DDR2 667
MHz.
I can provide more info if that will help.

As Dave mentioned in his reply, the best software tool to test memory is
Memtest86+ - Advanced Memory Diagnostic Tool
http://www.memtest.org/

However, your error messages may very well have nothing to do with
faulty RAM. Many error messages that mention the things you posted are
due to faulty device drivers, not faulty RAM. You should post the exact
error messages you have received, and when they occur.

 

[Tecknomage]

Have a look here, Terry: http://www.memtest.org/

Dave, Devon!

Question for Terry: Do you have POST Memory Test enabled in your BIOS
*and* allow POST to display (hide system logo boot screen)?

POST = Power On Self Test = BIOS run testing BEFORE boot to OS

The POST memory test is a basic hardware level test run by your BIOS
*if* you enable it. You should have this enabled (you can [Esc] to
end early) and in your case allow it to run completely through.

As for memory test utilities, like the one Dave linked to, are boot CD
that run in DOS Mode. That is because you cannot test memory while
Windows OS is loaded. BUT you cannot test memory that IS being used
by the utility, but DOS Mode use a much smaller portion of memory than
Windows. Having said that, these memory test utilities *do work* and
use multiple test patterns.

By contrast, the BIOS Memory test routine does test your full system
RAM since it runs on the BIOS, but it is a very basic test.


As to your errors. The MAY not be memory problems, you may want to
look at the following:

http://msdn.microsoft.com/en-us/performance/cc825801.aspx

http://msdn.microsoft.com/en-us/library/ms681381.aspx

http://msdn.microsoft.com/en-us/performance/default.aspx

http://msdn.microsoft.com/en-us/library/aa362823(VS.85).aspx

http://techsupt.winbatch.com/TS/T000001056F74.html

Consider bookmarking these links for the future.

Your problem could be your WinXP Configuration with your MESH system.

 

[Paul]

I've been getting an increasing number of crashes, which I now think
indicate some sort of memory failure. The messages are about not finding
addresses, 'exceptions', invalid instructions, etc.

How can I set about methodically isolating the cause and, more important,
repairing it please?

This is a MESH 'GTS Xtreme', Quad Core Q9450 2.66 GHz, 4 GB DDR2 667 MHz.
I can provide more info if that will help.

To do a thorough memory test with memtest86+, you need to
reconfigure the memory a bit. Memtest86+ can't test the first
1MB of memory or so (if you cared). You'd put the memory in
single channel mode, run the test, swap the DIMMs in the same
channel and run the test again. That would cover all locations.

Channel 0 Channel 1 Normal installation in dual channel mode
2GB_#1 2GB_#2 Two DIMMs are accessed alternately.
--- --- Less than perfect memtest86+ coverage
Low memory not covered.

These would be the two test cases to run with memtest86+, for perfect coverage.
You can use either Channel 0 or Channel 1 (or, as advised in your
motherboard manual). If you have more DIMMs than this, just test the DIMMs
two at a time, and put the device under test, in the higher memory slot.
To prevent going crazy, figuring out which slot is which, swapping the pair of
DIMMs covers all odds.

Channel 0 Channel 1 Single channel mode, #1 lowmem not all tested
2GB_#1 ---
2GB_#2 ---

Channel 0 Channel 1 Single channel mode, swap the DIMMs, test.
2GB_#2 ---
2GB_#1 ---

It's not really that important, but helps cover the area of memory
used by the BIOS.

Memtest86+ honors the E820 reserved memory area, which would be
things like BIOS space.

Another test, of value for indicating some kind of problem, is
Prime95. Memtest86+ is good for "stuck-at" faults, where a memory
bit refuses to set to one of the possible values. Prime95 is
better at noting transient problems (flaky memory timing),
tends to heat up the hardware, which is a form of "margin testing".
(Select the stress test option when running the program. You
don't have to "Join GIMPS".)

http://www.mersenne.org/freesoft

Prime95 only covers a limited amount of memory, but gives it a good
thrashing. It does a math computation with a known answer, which is
how it detects problems. Test coverage includes processor, Northbridge,
and (a portion of) system memory. Any computer should be able to pass
this test, for as long as you want. I stop at four hours, because I've
got better things to do. If it's error free for four hours, I'm
generally satisfied with that. There are tougher tests than Prime95,
but I haven't been keeping track of new tool introductions.

Paul

 

[Terry Pinnell]

Thanks all.

I downloaded Memtest86 and burned it to DVD, and it finished the first
pass with no errors in about an hour and a half. The documentation seems
to imply that it's unlikely to find any on subsequent passes.

Of course, that still leaves me trying to isolate cause of these crashes.
Symptoms are corrupt windows, spreading as I activate each in turn. I was
back to suspecting graphics card (nVidia 8800GT). But I'll study the links
Tecknomage suggested tomorrow.

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

I'm still curious why I couldn't reboot from the USB flash drive. After
more reading I think it might be because those instructions were
inadequate, and you're supposed to reformat and do other complicated stuff
to the flash drive?

I'd like to be able to reboot from one, partly for my own education but
also as preparation for future needs. Some simple utility would be ideal
to prove it. Not a full XP Pro reboot!

 

[Paul]

Thanks all.

I downloaded Memtest86 and burned it to DVD, and it finished the first
pass with no errors in about an hour and a half. The documentation seems
to imply that it's unlikely to find any on subsequent passes.

Of course, that still leaves me trying to isolate cause of these crashes.
Symptoms are corrupt windows, spreading as I activate each in turn. I was
back to suspecting graphics card (nVidia 8800GT). But I'll study the links
Tecknomage suggested tomorrow.

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

I'm still curious why I couldn't reboot from the USB flash drive. After
more reading I think it might be because those instructions were
inadequate, and you're supposed to reformat and do other complicated stuff
to the flash drive?

I'd like to be able to reboot from one, partly for my own education but
also as preparation for future needs. Some simple utility would be ideal
to prove it. Not a full XP Pro reboot!

With regard to your comments about USB flash, yes, some OSes support
booting from a USB flash. For example, I have a Knoppix distro on an
8GB USB flash, and that boots from the USB. That Live distro is also
"presistent", in that a 4GB section is dedicated to storing the
home directory, from one session to the next.

In Windows, one problem with USB booting in general, is the USB
bus gets disconnected, half way through the boot process. Or
so I'm told (I've never wasted time trying it). There is a
recipe, that involves changing some registry settings having
to do with "boot bus extenders". As far as I know, what that
does, is change the way the USB bus is handled, so it doesn't
get disconnected during the boot. Presumably, Microsoft had
no reason to set up USB devices that way, because of the
activation issues implied by a (portable) installation
of Windows. In other words, Microsoft didn't set this
up, because it would help you violate the "one OS one computer"
rule. And activation is intended to stop that. Facilitating
USB boot, would only mean more phone calls from users.

http://www.techspot.com/vb/topic116114.html

So, yes, you *could* set up a Windows install, to
work from a USB device. And as long as that installation
isn't moved, then the activation logic should continue
to be happy. But if you were whipping that flash drive
around, from one computer to another, I don't see good
things happening then. "ET will be phoning home", if you
do that. You'd have to re-activate, perhaps by phoning
Microsoft.

Paul

 

[glee]

Thanks all.

I downloaded Memtest86 and burned it to DVD, and it finished the first
pass with no errors in about an hour and a half. The documentation
seems
to imply that it's unlikely to find any on subsequent passes.

Of course, that still leaves me trying to isolate cause of these
crashes.
Symptoms are corrupt windows, spreading as I activate each in turn. I
was
back to suspecting graphics card (nVidia 8800GT). But I'll study the
links
Tecknomage suggested tomorrow.

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

I'm still curious why I couldn't reboot from the USB flash drive.
After
more reading I think it might be because those instructions were
inadequate, and you're supposed to reformat and do other complicated
stuff
to the flash drive?

I'd like to be able to reboot from one, partly for my own education
but
also as preparation for future needs. Some simple utility would be
ideal
to prove it. Not a full XP Pro reboot!

If you would provide us with the actual verbatim errors you are
receiving when you experience these "crashes", we could provide more
specific links and troubleshooting than the rather broad collection
Tecknomage gave you.

This is the first mention I see in this thread about booting from a USB
flash drive. Exactly what were you trying to boot from a flash drive?
Only some utilities and operating systems can do that, and must have a
setup routine for the flash drive specifically. Obviously, you also
need to set your boot order to boot from USB in BIOS setup. Were you
trying to put Memtest86+ on a flash drive, or some other utility? How
did you go about setting up the flash drive?

 

[Terry Pinnell]

If you would provide us with the actual verbatim errors you are
receiving when you experience these "crashes", we could provide more
specific links and troubleshooting than the rather broad collection
Tecknomage gave you.

This is the first mention I see in this thread about booting from a USB
flash drive. Exactly what were you trying to boot from a flash drive?
Only some utilities and operating systems can do that, and must have a
setup routine for the flash drive specifically. Obviously, you also
need to set your boot order to boot from USB in BIOS setup. Were you
trying to put Memtest86+ on a flash drive, or some other utility? How
did you go about setting up the flash drive?

Thanks Glen. You're right, I didn't explain my introduction of that
side-issue, sorry!

At the risk of telling you more than you want here's more background: ;-)

I originally found a program called mtinst.exe here:
http://www.ehow.co.uk/how_7321200_run-memory-test-windows-xp.html
and then the Resources link 'Microsoft: Download Windows Memory
Diagnostic' here:
http://oca.microsoft.com/en/mtinst.exe

But, after changing my BIOS boot priorities to

Removable
CD-ROM
Hard drive
Disabled

my PC still rebooted to Windows as usual. I then copied it to a CF card
and tried that, with same result. I now think the instructions were
inadequate, because further research shows that there's more necessary
than merely copying the downloaded file to the flash drive.

Anyway, curiosity apart, it's no longer my focus. Memtest86 is obviously
well-regarded and it worked for me, so I'm going to postpone any
experiments with flash drive booting. Possibly indefinitely!

Turning to symptoms, I wish I could be more precise. But when I get these
crashes the situation quickly becomes very confusing! And, of course,
there may be more than one issue.

The messages are about addresses, exceptions, invalid instructions, etc.
These sometimes arise in the course of trying to close windows that are
'corrupted' visually (spreading as I activate each in turn). But not
always!

The messages seemed vaguely indicative of 'memory' problems. But now,
after that green light from Memtest86, and a comment elsewhere referring
me to
http://www.semiaccurate.com/2010/07/11/why-nvidias-chips-are-defective/
I'm tending to suspect my nVidia 512MB 8800GT graphics card now.

When (I'd like to say 'If', but I fear it's only a matter of time) I get
the next crash, I'll try to be more methodical in recording the symptoms
and actions I take.

 

[Terry Pinnell]

With regard to your comments about USB flash, yes, some OSes support
booting from a USB flash. For example, I have a Knoppix distro on an
8GB USB flash, and that boots from the USB. That Live distro is also
"presistent", in that a 4GB section is dedicated to storing the
home directory, from one session to the next.

In Windows, one problem with USB booting in general, is the USB
bus gets disconnected, half way through the boot process. Or
so I'm told (I've never wasted time trying it). There is a
recipe, that involves changing some registry settings having
to do with "boot bus extenders". As far as I know, what that
does, is change the way the USB bus is handled, so it doesn't
get disconnected during the boot. Presumably, Microsoft had
no reason to set up USB devices that way, because of the
activation issues implied by a (portable) installation
of Windows. In other words, Microsoft didn't set this
up, because it would help you violate the "one OS one computer"
rule. And activation is intended to stop that. Facilitating
USB boot, would only mean more phone calls from users.

http://www.techspot.com/vb/topic116114.html

So, yes, you *could* set up a Windows install, to
work from a USB device. And as long as that installation
isn't moved, then the activation logic should continue
to be happy. But if you were whipping that flash drive
around, from one computer to another, I don't see good
things happening then. "ET will be phoning home", if you
do that. You'd have to re-activate, perhaps by phoning
Microsoft.

Thanks Paul. As you see from my reply to Glen, in view of your and others'
feedback I'm going to put my curiosity about this side-issue on hold!

 

[glee]

replies interspersed inline....

Thanks Glen. You're right, I didn't explain my introduction of that
side-issue, sorry!

At the risk of telling you more than you want here's more background:
;-)

I originally found a program called mtinst.exe here:
http://www.ehow.co.uk/how_7321200_run-memory-test-windows-xp.html
and then the Resources link 'Microsoft: Download Windows Memory
Diagnostic' here:
http://oca.microsoft.com/en/mtinst.exe

Half the problem is that the instructions on the site you went to are
totally incorrect. The file they had you download is an installer, to
install the bootable memory test onto a floppy disk or a CD, which it
will make bootable when it creates the disk. The other half of the
problem is that the Microsoft memory test utility does not allow for USB
stick installation. It only has options to make a floppy or a CD. Here
is the web page the bozo who wrote the ehow.com page SHOULD have
directed you to for instructions and download:
http://oca.microsoft.com/en/windiag.asp

Note that thie Microsoft utility cannot test more than 4GB of installed
RAM.

But, after changing my BIOS boot priorities to

Removable
CD-ROM
Hard drive
Disabled

my PC still rebooted to Windows as usual. I then copied it to a CF
card
and tried that, with same result. I now think the instructions were
inadequate, because further research shows that there's more necessary
than merely copying the downloaded file to the flash drive.

Exactly....see my explanation above.

Anyway, curiosity apart, it's no longer my focus. Memtest86 is
obviously
well-regarded and it worked for me, so I'm going to postpone any
experiments with flash drive booting. Possibly indefinitely!

Memtest86+ has an installer for creating a bootable USB flash drive, as
well as an installer to make a bootable floppy, and an ISO file to make
a bootable CD or DVD.
http://www.memtest.org/#downiso

If you download and then run the Auto-installer for USB Key (Win
9x/2k/xp/7), it will make a bootable USB flash drive with Memtest86+ on
it.

Turning to symptoms, I wish I could be more precise. But when I get
these
crashes the situation quickly becomes very confusing! And, of course,
there may be more than one issue.

The messages are about addresses, exceptions, invalid instructions,
etc.
These sometimes arise in the course of trying to close windows that
are
'corrupted' visually (spreading as I activate each in turn). But not
always!

The messages seemed vaguely indicative of 'memory' problems. But now,
after that green light from Memtest86, and a comment elsewhere
referring
me to
http://www.semiaccurate.com/2010/07/11/why-nvidias-chips-are-defective/
I'm tending to suspect my nVidia 512MB 8800GT graphics card now.

When (I'd like to say 'If', but I fear it's only a matter of time) I
get
the next crash, I'll try to be more methodical in recording the
symptoms
and actions I take.

It depends on the exact errors, but I suspect the nVidia graphics is NOT
the problem. A hardware fault like that would more likely be giving you
freezes, shutdowns and blue screens of death, than the kind of popup
errors you indicate. What you describe has more the earmarks of a
malware infection.

 

[Terry Pinnell]

replies interspersed inline....




Half the problem is that the instructions on the site you went to are
totally incorrect. The file they had you download is an installer, to
install the bootable memory test onto a floppy disk or a CD, which it
will make bootable when it creates the disk. The other half of the
problem is that the Microsoft memory test utility does not allow for USB
stick installation. It only has options to make a floppy or a CD. Here
is the web page the bozo who wrote the ehow.com page SHOULD have
directed you to for instructions and download:
http://oca.microsoft.com/en/windiag.asp

Note that thie Microsoft utility cannot test more than 4GB of installed
RAM.




Exactly....see my explanation above.




Memtest86+ has an installer for creating a bootable USB flash drive, as
well as an installer to make a bootable floppy, and an ISO file to make
a bootable CD or DVD.
http://www.memtest.org/#downiso

If you download and then run the Auto-installer for USB Key (Win
9x/2k/xp/7), it will make a bootable USB flash drive with Memtest86+ on
it.




It depends on the exact errors, but I suspect the nVidia graphics is NOT
the problem. A hardware fault like that would more likely be giving you
freezes, shutdowns and blue screens of death, than the kind of popup
errors you indicate. What you describe has more the earmarks of a
malware infection.

Thanks Glen, much appreciate your follow-up.

I said I'd try to report in more detail when the next crash occurred,
which it did about 30 minutes ago.

I was using my video editor, Magix Movie Edit Pro. (I've had similar
crashes when that wasn't running.) A message window suddenly appeared
from another program that was currently inactive, my music player, Media
Monkey. (I've had similar crashes when that wasn't running.) MM was
minimised as an icon in the XP system tray. The message was:

Title was name of a music track
"Thread creation error. Not enough quota is available to process this
command."

As MM didn't have focus and I hadn't given it a command, that's puzzling
in itself.

I tried to take a screenshot with Prt Scr followed by opening IrfanView
ready to paste it. But IV wouldn't open, giving the message:
"Windows cannot access the specified device, path or file. You may not
have the appropriate permissions..."

These error messages could not be closed.

Moving them by dragging on one occasion resulted in several copies being
created!

Opening another program (to see if I still could) I saw the familiar
'corruption' I've mentioned, and which made me suspect the graphics card:
its title was transparent, so other items below gave a confusing
impression.

I then tried to open XP Pro's Task Mgr with Ctl+Alt+Del but failed, with
the message:
"taskmgr.exe - Application Error
The application failed to initialize properly (0xc000012d). Click on OK to
terminate the application."

Doing so failed to close it.

I was able to open Process Explorer from the tray where I'd placed it for
such a situation. No excessive CPU activity was evident. I couldn't close
PE.

I was able to restart via the Start button.

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

I scan my OS drive nightly with Avira AntiVir Personal (Free), and my
backup drive weekly. Occasionally I run Malawarebytes and will do so again
shortly. But I'm inclined to think it's not malware.

Hope the above (from tediously handwritten notes!) might help to home in
on the likely cause.

 

[glee]

Thanks Glen, much appreciate your follow-up.

I said I'd try to report in more detail when the next crash occurred,
which it did about 30 minutes ago.

I was using my video editor, Magix Movie Edit Pro. (I've had similar
crashes when that wasn't running.) A message window suddenly appeared
from another program that was currently inactive, my music player,
Media
Monkey. (I've had similar crashes when that wasn't running.) MM was
minimised as an icon in the XP system tray. The message was:

Title was name of a music track
"Thread creation error. Not enough quota is available to process this
command."

As MM didn't have focus and I hadn't given it a command, that's
puzzling
in itself.

I tried to take a screenshot with Prt Scr followed by opening
IrfanView
ready to paste it. But IV wouldn't open, giving the message:
"Windows cannot access the specified device, path or file. You may not
have the appropriate permissions..."

These error messages could not be closed.

Moving them by dragging on one occasion resulted in several copies
being
created!

Opening another program (to see if I still could) I saw the familiar
'corruption' I've mentioned, and which made me suspect the graphics
card:
its title was transparent, so other items below gave a confusing
impression.

I then tried to open XP Pro's Task Mgr with Ctl+Alt+Del but failed,
with
the message:
"taskmgr.exe - Application Error
The application failed to initialize properly (0xc000012d). Click on
OK to
terminate the application."

Doing so failed to close it.

I was able to open Process Explorer from the tray where I'd placed it
for
such a situation. No excessive CPU activity was evident. I couldn't
close
PE.

I was able to restart via the Start button.

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

I scan my OS drive nightly with Avira AntiVir Personal (Free), and my
backup drive weekly. Occasionally I run Malawarebytes and will do so
again
shortly. But I'm inclined to think it's not malware.

Hope the above (from tediously handwritten notes!) might help to home
in
on the likely cause.

Have you tried what I suggested in your previous thread?
Click Start> Run, type sfc.exe /scannow
Click OK
Insert your XP CD when prompted.

How To Use Sfc.exe To Repair System Files
http://www.bleepingcomputer.com/forums/topic43051.html

 

[Terry Pinnell]

Have you tried what I suggested in your previous thread?
Click Start> Run, type sfc.exe /scannow
Click OK
Insert your XP CD when prompted.

How To Use Sfc.exe To Repair System Files
http://www.bleepingcomputer.com/forums/topic43051.html

Yes, that was one of my early steps. I have \i366 on my HD. SFC did
apparently replace one file, an intervention right at the end (it didn't
report its name), but that hasn't fixed the crashing.

I've also just had a suggestion that "You may be running a program that
has a Graphics Device Interface(GDI) "memory leak". That chimes with me,
because when I saw those messages and their rapid consequences, I thought
it looked like maybe I was running out of RAM or paging resources. Out of
my depth here though, so would welcome any other thoughts on how to
isolate it.

 

[Terry Pinnell]

I've also just had a suggestion that "You may be running a program that
has a Graphics Device Interface(GDI) "memory leak". That chimes with me,
because when I saw those messages and their rapid consequences, I thought
it looked like maybe I was running out of RAM or paging resources. Out of
my depth here though, so would welcome any other thoughts on how to
isolate it.

Here is a composite of screenshots taken just now of GDIview and PE
displays:

http://dl.dropbox.com/u/4019461/GDI+PE-1.jpg

Does that offer any clues please?

In case it's relevant, my PC has not been rebooted for about 30 hours. A
few major applications have been on overnight. Like my video editor
Videodeluxe.exe.

 

[glee]

Here is a composite of screenshots taken just now of GDIview and PE
displays:

http://dl.dropbox.com/u/4019461/GDI+PE-1.jpg

Does that offer any clues please?

In case it's relevant, my PC has not been rebooted for about 30 hours.
A
few major applications have been on overnight. Like my video editor
Videodeluxe.exe.

I wouldn't be surprised if your video editing apps, being open so long,
were using increasing amount of virtual memory. You have a lot of
resource-intensive apps open at the same time and for long periods, and
have a lot of background processes also, may running at Windows startup
from the looks of it.

When you have all these apps open for a long period, before you start
seeing the issues, open Windows Task Manager, click the Processes tab.
Put a check in the box to "Show processes for all users" at the bottom.
Click the View menu> Select Columns.
Put a check in the box for "Virtual Memory Size" and click OK.
Click the column heading named "VM Size" twice, to sort processes by VM
Size with the largest at top.
Minimize the Task Manager.

When the issue arises, bring Task Manager up and see how much Virtual
Memory is being used by the highest listed users, and what processes
they are.

Has you system ever NOT had this issue, since you've been running all
these apps simultaneously for over 24 hours, with all those background
processes? I suspect you may need to close some of these apps more
frequently than you are, to release some of the virtual memory
periodically. Video editors and some other multimedia apps, as well as
some browsers, will sometimes keep using more and more till the system
bogs down in odd ways.

 

[Terry Pinnell]

I wouldn't be surprised if your video editing apps, being open so long,
were using increasing amount of virtual memory. You have a lot of
resource-intensive apps open at the same time and for long periods, and
have a lot of background processes also, may running at Windows startup
from the looks of it.

When you have all these apps open for a long period, before you start
seeing the issues, open Windows Task Manager, click the Processes tab.
Put a check in the box to "Show processes for all users" at the bottom.
Click the View menu> Select Columns.
Put a check in the box for "Virtual Memory Size" and click OK.
Click the column heading named "VM Size" twice, to sort processes by VM
Size with the largest at top.
Minimize the Task Manager.

When the issue arises, bring Task Manager up and see how much Virtual
Memory is being used by the highest listed users, and what processes
they are.

Has you system ever NOT had this issue, since you've been running all
these apps simultaneously for over 24 hours, with all those background
processes? I suspect you may need to close some of these apps more
frequently than you are, to release some of the virtual memory
periodically. Video editors and some other multimedia apps, as well as
some browsers, will sometimes keep using more and more till the system
bogs down in odd ways.

Thanks Glen, appreciate your sticking with me on this.

I had "Show processes for all users" checked, but not the "Virtual Memory
Size" column, which I've now added, and will monitor as you suggest.

Yes, this is relatively recent, couple of weeks. With a *roughly* similar
workload/workstyle, I previously didn't get this sort of crash.

I'd dearly love to establish that this is a rectifiable software or user
issue and NOT need a replacement graphics card!

 

[Terry Pinnell]

Thanks Glen, appreciate your sticking with me on this.

I had "Show processes for all users" checked, but not the "Virtual Memory
Size" column, which I've now added, and will monitor as you suggest.

Yes, this is relatively recent, couple of weeks. With a *roughly* similar
workload/workstyle, I previously didn't get this sort of crash.

I'd dearly love to establish that this is a rectifiable software or user
issue and NOT need a replacement graphics card!

As part of this monitoring and detective work I've been watching the 'PF
Usage' number under the Performance tab of Task Manager. One thing that
puzzles me is why there's such a large difference between that number and
the total of the 'Mem Usage' column under the Processes tab?

Right now for example, PF usage = 918 MB, while Mem usage = 505 MB.(I
copied the data into Excel to get that sum.)

I'm guessing this has an obvious explanation?

 

[glee]

snip
As part of this monitoring and detective work I've been watching the
'PF
Usage' number under the Performance tab of Task Manager. One thing
that
puzzles me is why there's such a large difference between that number
and
the total of the 'Mem Usage' column under the Processes tab?

Right now for example, PF usage = 918 MB, while Mem usage = 505 MB.(I
copied the data into Excel to get that sum.)

I'm guessing this has an obvious explanation?

Sorry for not getting back sooner, but this really has nothing to do
with your problem.

PF Usage stands for Page File Usage, which is not Memory Usage, so there
really is no correlation.
Here, read this:
Explanation of Pagefile Usage as reported in the Task Manager -
http://blogs.technet.com/b/perfguru...le-usage-as-reported-in-the-task-manager.aspx