Death of an AGOIA?

[jna]

1) Watercooled AGOIA on 8K3A was stable at 1750mhz, 1.75V, for 6 months. Diode temps approached 40C under load.

2) Replaced 2x256MB Samsung PC2100 (oc'd to 2700) with 2x512MB Crucial PC2700. Memtest for 12 hours passed.

3) Replaced watercooling with SVC GC68 for a weekend trip. Load temps approached 50C (diode, not thermistor.)

4) Installed extra 80MM fan in 2 5.25 bays (to cool hard drives). 7volted TTake GF4 cooler.

4) Computer was not used for anything stressful over the weekend, just a spare.

5) Returned home, and C&C Generals began crashing at random.

6) Prime95 fails after 5 seconds or so, overclocked. (Temps around 50C).

7) Memtest passed for 12 hours again.

8) Increased voltage to 1.8, Prime95 still fails. Decreased voltage to 1.725 (decreased heat to 48C), Prime95 still fails after a few seconds.

9) Decreased FSB to stock. Load temps 45C. Prime95 still fails after a few seconds.

This is the same version of Prime95 that the same CPU used to pass with flying colors - I think it's the latest one.

The CPU was not initially stable at 1.75V. I ran it at 1.8V for about a month, then it "burned in".

It has been treated very gently, except for the 20% overclock. It's not even unlocked. But now, it won't run stably, at stock, with aircooling. Is there anything I could possibly be doing wrong, or is the chip gone?

 

[Duvie]

We are talking about a tbird 266fsb right?? What is the exact speed???

It does sound like possibly a damaged cpu...test it but turning off the cache in bios and see if it fails....Also the "burn in" statement says it all to me....There is no such credible evidence and we must hash this out once every 2 weeks here in the forums that there increased benefit from running a chip at a certain speed to gain stability at a voltage it couldn't run prior....The real word for it is "burn out"


I don't think temps are the factor....What about power issues???

 

[TerryMathews]

Well, in a short period of time, you replaced known good RAM with other RAM and replaced water cooling with air, causing two problems :

a) air cooling is less effective than water

b) with all the fans you added, you're drawing more current.

Bottom line, put everything back to your known good configuration and see if it's stable then. Athlons require a lot of juice to be stable.

 

[jna]

Originally posted by: Duvie
We are talking about a tbird 266fsb right?? What is the exact speed???

Palamino. The stock speed is 1400mhz. Overclocked speed is 1743mhz.

It does sound like possibly a damaged cpu...test it but turning off the cache in bios and see if it fails....Also the "burn in" statement says it all to me....There is no such credible evidence and we must hash this out once every 2 weeks here in the forums that there increased benefit from running a chip at a certain speed to gain stability at a voltage it couldn't run prior....The real word for it is "burn out"

Until August, it needed 1.8V to work. From August to March, it needed 1.75V to work. No other variables changed during this time period. It must have been the magic electron faeries. Every time a bell rings, a magic electron fairy gets it's wings, and turns a 1600+ into a real 2100+.

Turns out neither temps or power were a factor. I replaced the Crucial with 1x256MB Samsung and everything works fine. It's been running Prime95 for a few hours now. Looks like some Crucial is getting shipped back. It's funny that memtest86 didn't detect any errors, but replacing the memory made everything better. Looks like I need a new criteria for memory stability. (Yes, timings were set to SPD).

 

[jna]

Originally posted by: TerryMathews
Well, in a short period of time, you replaced known good RAM with other RAM

Bingo!

 

[TerryMathews]

Furthermore, lots of people have been having trouble with Crucial 512MB DIMMs, 'specially in NForce2 boards. Something is amiss there, I have a feeling it has something to do with the Vdimm or SPD chip, but who really knows.