Stability issues...

[Deanodriver]

Right, my parents' machine has been having some rebooting problems for a while. It'll just randomly reboot for no reason, and the folks are getting sick of it (to the stage where they've mentioned bringing in a professional tech).

Now, the machine is:

Athlon 64 3500+ AM2 (my PC also has a 3500+, but it's a 939)
Asus M2NPV-MX
1GB Corsair DDR2-667
80GB WD HDD
Liteon DVD-RW
genericish (Usicase) $60 case and PSU
XP Home

Now, the thing is, the problems also happened (more often though) with their old 1.2 Duron system (which did have bulging capacitors on the motherboard), and they've mentioned they upgraded this machine to fix that.

I'm thinking it's to do with the power up that end of the house, but does anyone have any ideas? Shall I swap out the RAM or PSU? Shall I invest in a UPS for it? Could something else be causing it?

I've run memtest86+ on it, and it came up with 128 errors, and, worryingly, the RAM was running at DDR736 (it's 667MHz RAM, I've turned all overclocking options off, and specified that it's 667 RAM in the BIOS).

What's more, the PSU has '400W' written on it, but its serial number has 330 in it, and it's only got a 16A 12v line. It'll be replaced eventually, but should it matter that much, considering it's got onboard everything?

Thanks in advance,

-- Dean

[NooNoo]

If the ram has errors after you have set it correctly in the bios, that would be the first thing to replace...but run memtest with one stick in, then the other and then together - it might just be a mismatch.

The psu sounds dodgey... but then I know many a dodgey looking psu functions perfectly fine.

You have turned off the automatic restart haven't you?

[Deanodriver]

If the ram has errors after you have set it correctly in the bios, that would be the first thing to replace...but run memtest with one stick in, then the other and then together - it might just be a mismatch.

The psu sounds dodgey... but then I know many a dodgey looking psu functions perfectly fine.

You have turned off the automatic restart haven't you?

I'll do the RAM switcheroo tomorrow, then. I did purchase it as a kit, though (Corsair Value Select 2x512MB)

Well, I plan to replace its PSU eventually, but if it's urgent, I'll replace it sooner.

Automatic restart? I don't think I have...

[NooNoo]

turn it off and you get to see the bsod.... did you check the event viewer?

[slgrieb]

Memory seems like the most likely culprit, but I think I would also pull the heatsink/fan and verify that there is a decent thermal interface in place. If the CPU uses an AMD provided heatsink/fan, there should be a nice uniform coat of slightly sticky grey goop on both surfaces. If the coating has obvious gaps or is missing, I'd say the thermal compound that bonds the CPU to the heatsink was either missing or improperly applied. Try cleaning off the current gunk and using Arctic Silver. Follow the Mfg directions very closely.

[Deanodriver]

I've swapped the DIMMS (same sticks, different slots), and although I haven't run memtest on it yet, it seems to be OK.

HSF seems to be fine, it's just the stock one, but it keeps it nice and cool

[NooNoo]

So run memtest and hope it's sorted!

[Deanodriver]

Did, and now there's only 53 errors, rather than 128.

I wonder if putting the RAM in single channel will help?

[Kodiak]

I have found it only takes 1 error. If you have 2 stick test one at a time and find the bad one and replace it. One error over time will corrupt your whole system.

[Deanodriver]

I've put a new PSU in, which helped a little.

Now, I finally decided to run the RAM (667 RAM) at 533 (well, 551 according to it, better than 736), to see if that helps it at all. Memtest turned out 0 errors, so here's hoping