Question regarding 'Event12, Kernel-WHEA' Error Log

[sangwooksohn]

I have question regarding Kernel-WHEA (EventID 12) error log which seems to
appear everytime Windows boots up.

I am currently running Windows Vista Ultimate x64 on AMD Opteron 280 box
with 4GB RAM. And I see following error log in Event Viewer;

Event12, Kernel-WHEA
Machine Check Event reported is a fatal Bus or Interconnect timeout error.
Memory Hierarchy Level: 3
Participation: 0
Request Type: 3
Memory/IO: 0
Address: 30422400


Same error log (different numbers in the address section, though) appears
everytime Windows boots up, but it does not seem to reappear while Windows is
on. In other words, this specific error gets logged only during the bootup
process.

Anyway, I did some research and tried to find out what it meant and how to
fix it, but only things I found were WHEA meaning 'Windows Hardware Error
Architecture' and Event12, Kernel-WHEA meaning 'Bus timeout machine check
exception'.

So I ran AMD's MCAT (Machine Check Analysis Tool) to dig little more
information, and here what MCAT says:

Event Source 0
Processor Number : 0
Bank Number : 0
Time Stamp (0x): 01C83B55 755106C4
Error Status (0x): D4384000 00000833
Error Address (0x): 00000000 01D03580
Error Misc. (0x): 00000000 00000000
Single bit errors:
Correctable ECC error
Error address valid in MCi_ADDR
Error reporting enabled
Second error
Error valid
Bus Error Code:
Participation processor: Local node originated the request (SRC)
Time-out: Request did not time out
Memory transaction type: Data read (DRD)
I/O: DRAM memory access (MEM)
Cache level: Generic (LG)
Data Cache Error MC0:
System line fill error into data cache
Syndrome: 0x70


I ran my system manufacturer provided diagnostic tool to see if I get any
CPU/RAM related error, but all test had passed without any error. I do not
have any problem in 'Device Manager' and even clean installation of Windows
Vista didn't resolve the issue.

This error log is driving me nuts for quite some time and since the my
computer manufacturer does not provide any support other than the one running
Windows XP, I am not comfortable to ask them to fix my issue unless I know
specifically which part of my hardware (such as CPU or RAM) has gone wrong.

So pelase, I desperately need any input to either resolve this annoying
error, or at least a hint what part of my hardware is gone wrong.

Thanks in advance

 

[Dwarf]

Hi sangwooksohn,

If you have overclocked your machine, try running it at its normal speed.
Also, are all your memory modules identical? Try booting your machine with
only one stick of memory in the first slot to see if that makes any
difference. Repeat with the other sticks and slots. Try different sticks,
slots and combinations.

 

[sangwooksohn]

Thanks for your comment.
I did run some more tests. And here's what I've found out:

Swapping CPU1 to CPU2: same error message
Repositioning DIMMs with each other: same error message
Removing DIMM's for CPU1 (with rest of 2GB installed for CPU2): no error log
Removing DIMM's for CPU2 (with rest of 2GB installed for CPU1): same error
message.


So it seems that neither CPU's nor DIMM's is causing the error.

Would it be due to faulty motherboard?

 

[TomB]

Without going to to excrutiating detail, I am guessing you have a faulty
driver on your system. Using Vista's built in driver verifier (verifier.exe),
run a test on the third party drivers on your system and you should be able
to determine the offending software.

Good Luck!
Memphis Tom

 

[TomB]

I wanted to add a couple notes to my previous post. First of all the previous
suggestions in this thread are wrong. Most memory errors and cpu errors will
NOT result in this error and they will definitely occur throughout the
operation of the computer and not just during startup.

Your problem is a timing or synchronization error between the OS and a
specific hardware device on the PCI bus (most likely). Some factors for you
consider to once you've determined which device is involved are PCI latency,
USB speed, and network file synchronization (BITS client configuration). The
above problems probably constitute 90 percent of the problems.