Could This Be a Memory Problem?

[rookcifer]

Hardware:

motherboard: Gigabyte MA770T-UD3
CPU: Phenom II 545 (RB-C2)
RAM: GSkill F3-12800CL9D-4GBNQ (4GB)
I am running everything at stock speed and voltages.

I run linux and am getting some Machine Check Exception errors in my logs. Here are a couple of examples:

HARDWARE ERROR. This is *NOT* a software problem!
Please contact your hardware vendor
MCE 0
CPU 1 0 data cache 
ADDR 687c1480 
TIME 1297143901 Mon Feb  7 23:45:01 2011
  Data cache ECC error (syndrome 0)
       bit46 = corrected ecc error
       bit62 = error overflow (multiple errors)
  memory/cache error 'data read mem transaction, data transaction, level 2'
STATUS d4004000f1000136 MCGSTATUS 0
MCGCAP 106 APICID 1 SOCKETID 0 
CPUID Vendor AMD Family 16 Model 4

HARDWARE ERROR. This is *NOT* a software problem!
Please contact your hardware vendor
MCE 1
CPU 1 2 bus unit 
ADDR b3889480 
TIME 1297143901 Mon Feb  7 23:45:01 2011
  L2 cache ECC error
  Bus or cache array error
       bit40 = error found by scrub
       bit46 = corrected ecc error
  memory/cache error 'generic error mem transaction, generic transaction, level 2'
STATUS 940041000000010a MCGSTATUS 0
MCGCAP 106 APICID 1 SOCKETID 0 
CPUID Vendor AMD Family 16 Model 4

I am not usually seeing any instability issues from these errors. However, about once every couple days I will get a hard lockup that requires a reboot. I have posted to the AMD forums with no response. I have posted to various enthusiast forums and most people say to RMA the CPU and motherboard. However, I am past my 30 days at Newegg for that (I have had this machine for 31 days), and I don't feel like waiting months for AMD and Gigabyte to give me new parts.

It seems the errors are related to the L2 cache, but from my reading, these errors can be a result of a bad CPU, RAM, motherboard and PSU. So, it's not automatically a CPU problem. Does anyone have any insight?

[GSKILL TECH]

How do you have the memory configured in BIOS? Have you tested one module at a time?

Refer to this thread:

http://www.gskill.us/forum/showthrea...7226#post37226

Thank you
GSKILL TECH

[rookcifer]

I have the memory set as follows:

1600Mhz, 9-9-9-24 @ 1.5v

I am still getting Machine Check Exceptions in my logs. I even had a hard lockup earlier today. And, yes, I have tested both modules individually and I still get the errors. I have tested them at 1300MHZ and still get the errors.

Looking through dmesg, I also found the following error (this happens on boot):

pnp 00:02: disabling [mem 0x00000000-0x00000fff window] because it overlaps 0000:00:00.0 BAR 3 [mem 0x00000000-0x1fffffff 64bit]
[    0.435604] pnp 00:02: disabling [mem 0x00000000-0x00000fff window disabled] because it overlaps 0000:01:00.0 BAR 6 [mem 0x00000000-0x0001ffff pref]
[    0.437211] pnp 00:0c: disabling [mem 0x000cd400-0x000cffff] because it overlaps 0000:00:00.0 BAR 3 [mem 0x00000000-0x1fffffff 64bit]
[    0.437214] pnp 00:0c: disabling [mem 0x000f0000-0x000f7fff] because it overlaps 0000:00:00.0 BAR 3 [mem 0x00000000-0x1fffffff 64bit]
[    0.437216] pnp 00:0c: disabling [mem 0x000f8000-0x000fbfff] because it overlaps 0000:00:00.0 BAR 3 [mem 0x00000000-0x1fffffff 64bit]
[    0.437219] pnp 00:0c: disabling [mem 0x000fc000-0x000fffff] because it overlaps 0000:00:00.0 BAR 3 [mem 0x00000000-0x1fffffff 64bit]
[    0.437221] pnp 00:0c: disabling [mem 0x00000000-0x0009ffff] because it overlaps 0000:00:00.0 BAR 3 [mem 0x00000000-0x1fffffff 64bit]
[    0.437224] pnp 00:0c: disabling [mem 0x00100000-0xcfdeffff] because it overlaps 0000:00:00.0 BAR 3 [mem 0x00000000-0x1fffffff 64bit]

I have googled the error and found others with it. However, I can't tell if it is hardware related or not (some hint that it might be a kernel software bug).

I am going to run Memtest over night tonight and see if I can find any errors. I think this could be a CPU problem, but I want to rule memory out first.

[Tradesman]

Let us know...if problems continue could you post your system voltage settings

[rookcifer]

OK, I ran Memtest for over 9 hours and didn't see a single error. I guess now I can run some sort of memory stress test, and if it passes, I will know it's my CPU.