Problems with Overclocked Opteron

[Deathcharge]

Hi

I built a system back in Dec 05 (5 months ago)

DFI LANPARTY UT nF4 SLI-DR eXpert
Opteron 146
Thermalright XP-90 with 92mm Panaflo fan
Crucial Ballistics PC4000 2GB @3-4-4-8 -- BL2KIT12864Z503
256MB Leadtek VIVO 7800GT
Antec SLK3000B
Seagate Barracuda 7200.8 250GB
BenQ's DW1640
ENERMAX 535watt ATX 12V V2.01/EPS 12V

after i built it i experimented with OC a bit, i was happy that it ran at 2.5 Ghz (1-1 with the memory) on STOCK voltage ie just increased the FSB to 250Mhz. It has been running happily ever since without a single crash or BSOD. no matter what i throw at it from DOOM3 to FEAR to prime 95 etc. my longest running time was 6 weeks when I was overseas, came back and it was still up

a week ago my problems started with several BSODs from doing simple tasks like unraring large files. I didnt think much of it really, until the crashes became more frequent and very annoying

i tried repairing windows XP but that didnt help, i tried installing a new instance of XP just in case the original was corrupted but that didnt help either.

I thought this could be the memory since there are several reports of the ballistics dying after few months (from excess voltage, which shouldn't be an issue here because its running on stock voltage).

I ran MemTest and it was stable with no errors when the memory ran at DDR 400 (ie FSB 200 Mhz) and DDR 500 (ie FSB 250 Mhz) so i thought that rules the memory out, and it must be the CPU.

Dropped the CPU down to 2.0 Ghz from 2.5 by changing the FSB and it is very stable priming happily and 3dmark05/06 ran fine where it BSOD before.

increased it to 2.4 and it errors and crashes, how bizarre it was as happy as larry before on 2.5 so whats going on? I haven't tried running it at 2.2, but wouldn't surprise me if that fails too.

I tried going back to 2.5 and I cant even boot into Windows!!! BSOD even before the user login screen.

any ideas? suggestions? comments? Am I missing something here? I would really like to run it at 2.4 at least , I know that there are people running this at 2.6-2.8 daily - stable if not 3.0Ghz. I am happy with 2.4-2.6 for daily use. Do you think I should give it more juice and see how it goes? should I RMA it? and on what grounds since it seems that it is stable at 2.0. how do i go about RMAing it, i am tempted to increase the voltage and really push the system to the limit so if it dies (which is very likely i can just RMA it while it is still under warranty).

why the sudden change whats happened? I can't check temps unfortunately because of the DFI bug. If i try running smart guardian it reports something like 70C and the computer shuts down instantly. I am sure it is not that high room temp in Melbourne is around 16C right now and this was running fine during the summer in 30C+ heat. CPU fan is spinning so its not that either.

I would really like to hear your comments so please let me know anything that might be helpful/useful

Cheers

Deathcharge

 

[DrMrLordX]

First off, don't RMA the chip.

Hard to say what else you should due if your onboard temp sensors are flipping out. If they were pretty high, bumping vcore might only make things worse in the long run. But if the temps are low, you might want to try boosting vcore. You can go up to about 1.45v safely, usually, though your chip seems a little sensitive. It would be worth your while to find a good way to check your temperatures.

As for the sudden change, it could be that your chip was sensitive and that electron migration got you in the end. This usually doens't happen at stock vcore, but stranger things have happened. Some dual-core owners lose cores or single cores go unstable, etc. It might be the on-chip memory controller that went bad.

Heck, it might even be the board itself that has issues, though I doubt that.

 

[Deathcharge]

should I be concerned that this sort of thing happened after 5 months? if it happened overnight sure its not stable and I would then run it at stock speed. I am a bit concerned that the chip might be on the way out and this is just an early warning sing. what do you think?

 

[taylor77]

I would check the power on 12v rails

 

[DrMrLordX]

Yes, checing your 12v rails from the PSU would help. Also, what kind of protection do you have against power surges? Got a power strip, surge protector, UPS, or something else?

 

[Deathcharge]

Originally posted by: DrMrLordX
Yes, checing your 12v rails from the PSU would help. Also, what kind of protection do you have against power surges? Got a power strip, surge protector, UPS, or something else?

No i dont which is a good point, i am going to go out today and buy a UPS or at least a surge protector

EDIT- what do i need to check the 12v rail on the PSU?

 

[customcoms]

To make sure its stable and the board/cpu etc. is being properly powered. Unfortuanatly for you, you will have to use a voltmeter as the DFI's onboard sensors for the rails are notorious for reporting undervoltage. I use MBM5 with the DFI settigns from dfi-street.com without any problems. Excuse the spelling, it is getting late and I was sailboat racing in over 35 knots of wind for 2 hours (racing for 6 hrs) today..very exhausted.