Motherboard or Memory Error?

Hi.

So I was surfing, downloading and listening to music on my pc when all of a sudden it hung. I restarted it and it just kept on restarting, without getting to the log-in page. I tried logging on in safe mode, but that didn't work either. I was ready to format the hd, but I changed the setting from quick POST to disabled and I saw the memory read as 524, when I had 1Gig, so I tried to further isolate the problem.

Proc: Sempron 2400
Mobo: KT600A
RAM: OCZ Platinum PC3200

1. Removed both sticks from their slots, and the pc beeped.
2. Inserting a stick into a slot allowed me to log on to Windows again.
3. I tested each individual stick in all three slots and ran memtest 3 times per stick/slot combo. I also surfed heavily (opening multiple windows) using both IE and Firefox. And I ran flash streaming video from yahoo and google.
- Firefox and IE would constantly shut down and report an error but they didn't shut down at the same time. Firefox would always die first and then followed by IE a few minutes later. Watching streaming video usually restarts the pc.
- Each memtest check had at least one instance where no errors were detected, but both Firefox and IE still shut down.
- Inserting a stick into DIM 1 resulted in more browser errors, more frequent restarts. With stick A in DIM 1, the pc wouldn't even enter the log in screen.
- Inserting both sticks in DIM 1 and DIM 3 (my normal config) the POST still showed 524MB RAM.
- I can usually surf longer, keep programs running longer with a stick in DIM 2. which was unused til now.
- Sometimes, not often, start up programs like Zone Alarm and Ewido would report an error and shut down.
- Even at idle, the pc still restarts, sometimes it took as long as an hour, sometimes as soon as 20 minutes.
- I downloaded and reinstalled all drivers (VIA, Catalyst) funny enough, the computer never restarted while I was downloading and installing.
4. I checked Event Viewer and these seem to be the most common errors:
- The COM+ Event System detected a bad return code during its internal processing. HRESULT was 80070422 from line 44 of d:\qxp_slp\com\com1x\src\events\tier1\eventsystemobj.cpp.
- Volume Shadow Copy Service error: Unexpected error calling routine CoCreateInstance. hr = 0x80040206.
- Faulting application firefox.exe, version 1.8.20060.25382, faulting module , version 0.0.0.0, fault address 0x00000000. (replace firefox with mplayerc.exe or bb2k7.exe or ...)
5. I checked the heat and it was steady at 40 to 41 degrees.

I hope you can help me isolate the problem. I'm leaning towards the mobo, but I can't totally write off the memory until I can test them in another box. As I have no other computer right now, I can't do it, but I'm going to a friend tomorrow to test the memory.
If there are other tests you can suggest, please let me know.
And I don't think its the PSU, because I run a total wattage of less than 300 on a 500 watt PSU.
Thanks.

Hi m4rk6 and welcome to Dev Shed.

Can you download and run memtest which will test the validity of your RAM.

Post back and let us know how it goes.

Sooo. your saying you had memory test failures with both sticks during testing, or what?

Is this one of the OCZ dual-channel memory kits or what? Are you running any modfied timings? Have you tried a failsafe mode setting change in the bios? What bios version fo you have?

WHO made the MB? KT600A is a via northbridge chipset, it might also be the mb model, but what company is it?

Thanks for replying.

Yeah. I tested each stick individually on each DIM slot, running memtest 3 times on each stick/slot combo, stick A in slot1, slot 2, slot 3, repeat for stick B. At least one memtest check on a stick/slot combo showed no errors. I let memtest run for about 10 minutes per test.

OCZ platinum is dual channel. Here's the link if you need it: OCZ
I use the 2x512.
I don't OC, I just used optimized defaults in the Bios. The Bios is Award.
I tried loading fail safe defaults and Windows XP just hung at the loading screen...Black screen with a white bar running at the bottom. I let it stay for 5 minutes before restarting and trying to run in safe mode. It seems to stop at Mup.sys.
I restored the bios to optimized default and I was able to get XP to load.

Let me rephrase my question, you said that at least one memory test passed for each stick...

Sooo. your saying you had (at least one) memory test failures with both (each) sticks during testing?

Hmm, not much of a rephrase.

The fact you couldn't get it to boot with failsafe settings also raises a HUGE red flag: either you didn't have something set right in your bios for your hardware with failsafe that is with optimized (likely since your using lower quality memory, probably designed only for speed, not reliability) or something is fubar on your motherboard (honestly... not as likely)

Yeah, I had at least one memory test failure with each stick during testing.
This is wierd though, I inserted both sticks into DIM 2 and 3, now the computer is running fine... its been running for 9 hours straight with no restarts, no errors, no nothing. I'm scratching my head, what the hell is going on? :P

Could be a faulty DIMM slot on the motherboard ?

Maybe.
I haven't tried using DIM 1 yet. I'm too scared. :P Guess I'll have to, hopefully it'll prove to be the one causing all the errors.
Umm, OCZ is considered lower quality memory? I know its 2 years old, I've had my computer for over 2 years actually, but at that time it was considered pretty good, as devhardware reviews said: LINK
Heh, not trying to be argumentative, just really wanna know your opinion on it.

My opinion on it, now, be aware, my girlfriend says I have a hardware fetish, I am that picky about my hardware...

It is neither registered nor ECC It is fast memory, but it has a very low mean error time (how often you will get an error because of the component) I would expect one memory error every 3 months due to those two DIMMs. If you are lucky, the error will be caught, and the app with the error will crash, and no further damage, if you are unlucky, the error will be undetected and will be in data that is written to disk, maybe causing corruption of a file, maybe causing your account program to misplace a number somewhere that won't be detected, and can lead to a major error that won't be detected but will affect a business severely.

(I am using numbers straight from OCZ in my estimations as well http://www.ocztechnology.com/displaypage.php?name=memory_types&psupp=1 they dismiss such failure chances as unimportant and minor, but when you run any sort of high demand or high reliability server it adds up.)

In the case of my personal server, if I was running memory with the error rate OCZ admits to, I would have an error every month and a half due to memory alone, add in other factors, such as CPU, HDD, etc. and I would be expecting a failure a month. Right now, I haven't had a single error, application shutdown, system restart, etc. for almost 110 days. I have 4GB of ECC Infineon memory, and while it is slower then your OCZ, I like knowing that my server, running windows XP, is very stable despite near constant 20-30% CPU loads and constant HDD load.

Sure, one crash or data error every three months does not seem like much, but it is enough to annoy me, it is enough to cause corruption/loss of files, forcing re-installs/recoveries, causing downtime and expense, in time to recreate the data or repair the damaged systems, or even just software downtime.