Trying to troubleshoot an unstable FX-51 box

[kb3edk]

Hey everyone, I'm having some stability issues with the following setup:

Mobo: Asus SK8V (Socket 940)
CPU: Athlon FX-51 with Koolance waterblock
RAM: 2x512 OCZ Registered PC3200
SATA: 2x150GB Seagate 7200 RPM, RAID 0 on VIA Southbridge controller (VT8237)
IDE: Master 40GB Seagate 7200 RPM, Slave 120GB Maxtor 7200 RPM
PSU: Generic CompUSA 400W
GPU: Radeon 9800XT
Audio: onboard AC97
OS: Win2K Server SP4

I had run this stable for several weeks, with the FX-51 overclocked one multiplier step up to 2.4 (12x200) But I have lots of problems running Half-Life 2 and their support board discourages overclocking, so what I did was downclock the FX-51 back to stock 2.2 (11x200) in the BIOS and let it run a little while.

I went out to grab dinner, when I got back the machine had locked up. I rebooted and checked Event Viewer, the 40GB IDE drive had lost its controller. It made kind of a grinding noise, the drive is probably b0rk3d. Fortunately I didn't lose much important data there. But the problem is now the machine is chronically unstable. It pretty much always fails all the Prime95 torture tests within 15 minutes or less, I do Folding@Home for Team AnandTech and that runs poorly as well (in fact I don't even like doing DC with this box for now, because if it's doing bad math that means bad science).

So now I am trying to troubleshoot. First thing I did was run Memtest86, my RAM is at stock 2-3-2-6 timings and passes 100%. So that leaves the following:

1 - Replace power supply: even though this is a generic PSU I keep a close eye on all the rails and there is very little fluctuation on any of them... max is +- .01v every few minutes. However I may not have enough watts with 4 (now 3) HDDs and a hungry Radeon card so maybe I should upgrade to an Antec 500W ($100 or so)

2 - Replace mobo: I had tried some HT bus overclocking on this board when I first got it 8 months ago, it pretty much refused to POST if I put HT anywhere above 205 MHz. I can replace this with another SK8V or an SK8N (pretty much the only widely available Socket 940 boards) for $100 to $200. Also there might be some gremlins in the IDE controller since I lost a drive recently. I unplugged my other IDE drive just to be on the safe side.

3 - Replace CPU: Maybe my FX-51 is fried from running an extra multiplier for the last few months. It runs a little on the hot side under load (54 or so, at 1.6vcore), however last night I underclocked it down to 1.6 GHz @ 1.4vcore, it ran around 46 degrees under load and still failed Prime95. I suppose I could switch my generic thermal grease to AS5 but it may be already too late. I can get a replacement FX-51 or Opteron on eBay for around $200.

4 - Something else: I did a clean install of Win2K two weeks ago, in my opinion it is more stable than XP, I don't think I need to reinstall it. I could also bail entirely and switch to a Socket 939 setup but that will be at least $600 even if I get all used parts.

Does anyone have any more advice for me? Am I missing something totally obvious here? Is this the right order to try debugging things?

Thanks to anyone who can help,

-Adam in Philly

 

[KGB]

Adam,

Definitely get a new HIGH QUALITY power supply.
I would stick with name brands like Antec, Fortron, Enermax, PCP&C.
A lot of stability issues should be resolved with a good power supply.

 

[Malak]

Originally posted by: KGBMAN
Adam,

Definitely get a new HIGH QUALITY power supply.
I would stick with name brands like Antec, Fortron, Enermax, PCP&C.
A lot of stability issues should be resolved with a good power supply.

I concur. Try this first.

 

[rasczak]

you say it's not stable, and you hear a grinding noise from your hard drive? shouldn't you try that first?

try underclocking your cpu to see if it runs more stable, if so install a spare hdd and see if you get problems after a clean install of windows.

i'm not sure if your psu is the prob in this one.

edit** i just saw your post about installing win2k. maybe you should go back to xp. support for drivers is much better with this os imo.

 

[kb3edk]

Originally posted by: jpbelauskas
you say it's not stable, and you hear a grinding noise from your hard drive? shouldn't you try that first?

try underclocking your cpu to see if it runs more stable, if so install a spare hdd and see if you get problems after a clean install of windows.

i'm not sure if your psu is the prob in this one.

edit** i just saw your post about installing win2k. maybe you should go back to xp. support for drivers is much better with this os imo.

Clarification... I should have mentioned I am booting Windows from my SATA array. The grinding noise was from an IDE drive that I was just using for data storage. I pulled it from the system right away and just for safety, I pulled my other IDE drive so all I have left now is the SATA array. I wasn't having any system problems before my IDE drive died... can't tell at this point if that was the cause of the problem or (more likely) a symptom.

Also yes, I am underclocking right now at 1.6 GHz (stock is 2.2). System is stable at idle, i.e. no more BSODs. However it still fails Prime95 torture tests after a few minutes and even ordinary apps like web browsers randomly crash.

I will probably try replacing the PSU first, all the stores around here have Antec units in stock so I'll pick one up tonight. If that doesn't work I'll do a clean WinXP install next, then the mobo and CPU.

Thanks everyone for their advice so far -

-Adam in Philly