Memory errors, crashing - Asus P4C800-E deluxe

  • Thread starter Thread starter JustMe
  • Start date Start date
J

JustMe

Greetings, I hope someone can help me as I cannot figure this problem out.

My PC intermittently (every day or so, but sometimes after an hour,
sometimes after a few days) crashes, usually during video playback. I use
Zoom media player which will automatically play back the next file in a
directory. Usually the crash occurs following the end of one file's playback
when I'd expect it to simply start on the next file.

When it crashes I get the same memory error in Windows:
***STOP:0x0000007F (0x00000000,0x00000000,0x00000000,0x00000000)
UNEXPECTED_KERNEL_MODE_TRAP Beginning dump of physical memory

Because this is a memory error, I downloaded the Microsoft Windows Memory
Diagnostic tool (http://oca.microsoft.com/en/windiag.asp) to find the fault.

If I run this tool with the Motherboard Bios set to your default settings
(which sets Advanced>Onboard Devices Configuration>Onboard Promise
Controller>Operating Mode>RAID), not all hard disks are identified upon
boot, but the tool says that all memory passes all tests (the tool runs
before the OS is loaded, from a bootable floppy disk).

However, as soon as I change the Operating Mode of the Onboard Promise
Controller to IDE (not the default setting), all hard disks become
identified upon boot, BUT the memory fails all tests. This is the only
configuration that enables me to properly boot and use all drives, though.

I don't know whether the faults are connected however the errors I receive
are memory errors and, it would seem that the only way I can set the Bios to
successfully boot with all drives seen, is to set the Promise Controller to
IDE (I don't use a RAID array) and doing this leads to memory faults being
identified by the Windows Memory Diagnostic tool.

I have tried using different (known good) memory and still get the same
intermittent crashes with the same memory fault.

Motherboard: Asus P4C800-E Deluxe
Bios: 1018
Revision: 2
Video: Asus V9520 Magic NVidia
Driver: 44.03
CPU: Intel Pentium 4 3.066GHz Prescott (800MHz) 1MB cache
Memory: 2 x Geil 512MB DDR PC3200 400MHz running in dual channel mode
OS: Windows 2000

Please can anyone help me to identify the cause and suggest a resolution?
 
JustMe said:
Greetings, I hope someone can help me as I cannot figure this problem out.

My PC intermittently (every day or so, but sometimes after an hour,
sometimes after a few days) crashes, usually during video playback. I use
Zoom media player which will automatically play back the next file in a
directory. Usually the crash occurs following the end of one file's playback
when I'd expect it to simply start on the next file.

When it crashes I get the same memory error in Windows:
***STOP:0x0000007F (0x00000000,0x00000000,0x00000000,0x00000000)
UNEXPECTED_KERNEL_MODE_TRAP Beginning dump of physical memory

Because this is a memory error, I downloaded the Microsoft Windows Memory
Diagnostic tool (http://oca.microsoft.com/en/windiag.asp) to find the fault.

If I run this tool with the Motherboard Bios set to your default settings
(which sets Advanced>Onboard Devices Configuration>Onboard Promise
Controller>Operating Mode>RAID), not all hard disks are identified upon
boot, but the tool says that all memory passes all tests (the tool runs
before the OS is loaded, from a bootable floppy disk).

However, as soon as I change the Operating Mode of the Onboard Promise
Controller to IDE (not the default setting), all hard disks become
identified upon boot, BUT the memory fails all tests. This is the only
configuration that enables me to properly boot and use all drives, though.

I don't know whether the faults are connected however the errors I receive
are memory errors and, it would seem that the only way I can set the Bios to
successfully boot with all drives seen, is to set the Promise Controller to
IDE (I don't use a RAID array) and doing this leads to memory faults being
identified by the Windows Memory Diagnostic tool.

I have tried using different (known good) memory and still get the same
intermittent crashes with the same memory fault.

Motherboard: Asus P4C800-E Deluxe
Bios: 1018
Revision: 2
Video: Asus V9520 Magic NVidia
Driver: 44.03
CPU: Intel Pentium 4 3.066GHz Prescott (800MHz) 1MB cache
Memory: 2 x Geil 512MB DDR PC3200 400MHz running in dual channel mode
OS: Windows 2000

Please can anyone help me to identify the cause and suggest a resolution?
Click to expand...

First, I would get a copy of memtest86 from memtest.org . That program
will format a blank floppy for you. The floppy has its own boot loader,
so no OS is needed. Using the floppy route for testing, means the
code doesn't go near any IDE or SATA controllers. (Set the floppy
to be first in the boot order.)

Test the memory with the Promise controller set in both of the
options you mentioned above.

My suspicion is you have a bad Promise chip, and need to RMA.
But perhaps memtest86 will give a different set of symptoms
and a different conclusion as to where the fault lies. (I'm trying
to remember now, what the fault was with the 20378. I thought
some people were getting interrupt storms from faulty chips,
but I don't remember where I read that.)

For the memory, a Vdimm setting of at least 2.6V is needed at
DDR400. You can use 2.75V on just about any memory without
any consequences to speak of. Sometimes the memory just needs
a little more voltage to work well.

HTH,
Paul
 
Paul said:
resolution?

First, I would get a copy of memtest86 from memtest.org . That program
will format a blank floppy for you. The floppy has its own boot loader,
so no OS is needed. Using the floppy route for testing, means the
code doesn't go near any IDE or SATA controllers. (Set the floppy
to be first in the boot order.)

Test the memory with the Promise controller set in both of the
options you mentioned above.

My suspicion is you have a bad Promise chip, and need to RMA.
But perhaps memtest86 will give a different set of symptoms
and a different conclusion as to where the fault lies. (I'm trying
to remember now, what the fault was with the 20378. I thought
some people were getting interrupt storms from faulty chips,
but I don't remember where I read that.)

For the memory, a Vdimm setting of at least 2.6V is needed at
DDR400. You can use 2.75V on just about any memory without
any consequences to speak of. Sometimes the memory just needs
a little more voltage to work well.

HTH,
Paul
Click to expand...

Hi Paul,

Thanks for your help.

I did as you advised and tried Memtest. It found no fault with the memory
with the controller in IDE mode - contradicting the Microsoft test. I reran
the MS test under the same conditions and it still reported errors :o/

I'm not familiar with the faults you describe above but the only memory
setting I could find was under Voltages where there is a VCore st to 1.73
volts. This can be switched to "Disabled" but I see no means to increment
the voltage nor enter my own setting.

Could a low setting here be the cause of the faults I am experiencing and,
if so, how/where do I adjust this - I cannot see a means in the Bios, beyond
switching between a predefined voltage and "Disabled". Or am I looking at
completely the wrong setting?
 
JustMe said:
Controller

Hi Paul,

Thanks for your help.

I did as you advised and tried Memtest. It found no fault with the memory
with the controller in IDE mode - contradicting the Microsoft test. I reran
the MS test under the same conditions and it still reported errors :o/

I'm not familiar with the faults you describe above but the only memory
setting I could find was under Voltages where there is a VCore st to 1.73
volts. This can be switched to "Disabled" but I see no means to increment
the voltage nor enter my own setting.

Could a low setting here be the cause of the faults I am experiencing and,
if so, how/where do I adjust this - I cannot see a means in the Bios, beyond
switching between a predefined voltage and "Disabled". Or am I looking at
completely the wrong setting?
Click to expand...

Whoops - posted before complete.
I've found the setting and am now trying at 2.75 volts. Same questions
apply.
Cheers.
 
JustMe said:
in

Whoops - posted before complete.
I've found the setting and am now trying at 2.75 volts. Same questions
apply.
Cheers.
Click to expand...
Still getting the same problem :o(
What else would you suggest I check?
 
"JustMe" said:
Hi Paul,

Thanks for your help.

I did as you advised and tried Memtest. It found no fault with
the memory with the controller in IDE mode - contradicting the
Microsoft test. I reran the MS test under the same conditions
and it still reported errors :o/

I'm not familiar with the faults you describe above but the
only memory setting I could find was under Voltages where
there is a VCore st to 1.73 volts. This can be switched to
"Disabled" but I see no means to increment the voltage nor
enter my own setting.

Could a low setting here be the cause of the faults I am
experiencing and, if so, how/where do I adjust this - I cannot
see a means in the Bios, beyond switching between a predefined
voltage and "Disabled". Or am I looking at completely the
wrong setting?
Click to expand...

OK. The memory test results tell me your memory is not at fault,
so you don't need any extra voltage. (It would only be a waste.)
While it is not a complete test, I would say your processor Vcore
doesn't need to be adjusted either, because to complete the
memtest86 program, the processor has to be stable as well.

I would say your problem is disk related. When comparing the
potential reliability of the Promise chip, to those rotating
disk drive mechanisms, it is more likely you have a disk problem.

Do you have some kind of fancy "swap file" setup ? Maybe the disk
with the system swap on it is actually at fault ? (Like an errored
sector in the swap file area.) If this was my system, I would
back up the data on each drive, and then find a disk test that
includes write operations to the entire disk. The manufacturer's
test utilities aren't as a general rule, very thorough, but
would be a good place to start (they are free, after all).
A non-destructive test does nothing but read operations, and
doesn't test the ability to write any data. It is also possible
a simple surface scan for read errors, will give you some ideas
as to how healthy the disk is.

To test the disk controllers, you would need a disk utility that
works properly, whether the disk is on the Southbridge or
the disk is on the Promise controller. Run the test twice,
once on one controller and again on the other controller. If
the results diverge wildly (pass on one interface, fail
badly on the other), it may be time to RMA the motherboard under
warranty.

If I had to guess, you have a bad disk. You could also have a
bad power supply. Some disks are pretty sensitive to low
voltage from the PSU, so take a look at +5V and +12V using
Asus Probe, while the system is running something like
the Prime95 torture test (mersenne.org). Most power supplies
regulate to 5%, so if your +12V is, say, less than +11V, or
the +5V is less than +4.7V as measured by Asus Probe, you
might want to try another power supply and repeat your disk
tests. (I am allowing more than a 5% error here, because
the monitor chip has some potential for measurement error as
well, and I wouldn't want to condemn the power supply, if the
monitor chip is reading slightly on the low side. A cheap
digital multimeter can also be used to verify the voltages.)

Testing disks is no fun, but it is better to catch a bad disk
problem, before you lose some data.

HTH,
Paul
 
STOP 7f does not strictly indicate a memory fault (that is one of several
possibilities).

Try a google on "STOP 0x0000007f" (without the quotes).

Note that you indicate that the first parameter to the stop is 0x00000000,
not 0x00000008 (which indicates double fault - a somewhat different problem)
as mentioned in many of the googled articles.

There is some good news on this page:

http://www.microsoft.com/resources/...Windows/XP/all/reskit/en-us/prmd_stp_ukdq.asp

It says if Parameter 1 is Zero ==> Divide by zero error which could be a
driver fault, some other hardware fault (EG the device controllers you
mention before) or a number of other things. Quote "0x00000000, or a divide
by zero error, occurs when a divide (DIV) instruction is run and the divisor
is 0. Memory corruption, other hardware failures, or software problems can
cause this message. ". This makes me think stuffed driver, stuffed
controller, or stuffed up bios / firmware / settings.

Have you tried clearing the BIOS using the reset procedure? This would be my
first step. Also check the RAM is running at the correct voltage (usually
default + 0.1v - this will do no harm to any ram).

The fact you have issues with the promise controller tends to indicate
either a BIOS, Controller Firmware, or Electronic fault problem. If you can,
can you use a controller other than the promise and disable it?

- Tim
 
Hi Tim,

Thanks for the advice.

I'm starting to think that the Promise Controller isn't the issue. The
reason I linked that to a potential fault initially, was because the MS
Memory Test utility wouldn't pass the memory with the controller set to IDE.
But the utility suggested by Paul (above) passed it just fine, set to IDE.

However your talk of drivers has set me thinking.

I can now predictably cause a crash. Here's how:

Boot into Windows 2000.

Start viewing video using Zoom player.

My video card is set to "Clone". This has the effect of delivering a
full-screen output of whatever video window is being played at the time, via
the video card's composite or S-Video outputs.

Normally I can then minimise Zoom, leave it running in the background with
the video happily rendered on the TV screen. However I am finding that if I
restore the Zoom player to the front, my PC crashes. For a while I linked
this to Zoom starting to play a new file, as I would tend to bring zoom back
for the start of playing back the next file, usually to adjust the volume.

So I'm going to try reinstalling the drivers for the video card and see
where that gets me...

Thanks to you both - BRB!
 
Well, that does make me think that the problem lies around the graphics
drivers / zoom software (I am not familiar with it).

As a general rule, if memtest86 does not report errors after 5 full runs
then your memory is ok.

It is always possible that the graphics card has faulty video ram, so if you
have an alternative graphics card - known good - then give that a blast.

- Tim
 
Well, I've worked backwards. I was getting crashes with video and also when
I disabled the USB ADSL modem, I found.

So I ripped out the only non-essential card (Compro DTV card which has been
a pain anyway), uninstalled it and its drivers, uninstalled USB modem
removed drivers and uninstalled/reinstalled video drivers.

Since then it's been OK with no crashes, although I'll give it a couple of
weeks before I'm certain that the problems have been eliminated, but it's
been good so far.

Thanks for advice so far - I'm sure I'll be back!

Cheers.
 
Back
Top