Intel Burn Test Fails Instantly

[Robert]

Here's the time line of events:

6 months ago - CPU @ 3.2 prime stable

Tried Intel Burn Test - Only stable at 2.8

At Christmas added an XFX RAID card - will no longer hold overclock - same WITHOUT the RAID card

Tried IBT at stock and it fails instantly


Any tips? I'd rather NOT have to replace anything but if I do I want to be certain of what the issue is

[Blaineoliver]

As time goes on, you need more voltage to keep the overclock as high.

[ConTRo13R]

i get the feeling of deja vu

[SiM]

As time goes on, you need more voltage to keep the overclock as high.

Got any sources or evidence to prove that claim?

[Blaineoliver]

Its far too late in the evening for me to do this, but show me any evidence that this is not the case? CPUs get hotter over time, cpus require more voltage over time, so overclocks, when system is strained more, need more volts.

[ConTRo13R]

Got any sources or evidence to prove that claim?

+1

I too fail to see you point in the cpu needing more voltage

[Agent]

Got any sources or evidence to prove that claim?

Wear and tear.

Over time, capacitors and other components wear. Their efficiency drops, thus often more voltage is needed to compensate for this.
It is normally more noticeable in overclocking due to things so often being run right on the edge.
Normal household appliances will normally never have this issue due to the *huge* tolerances they are often build with (why fit one of component X when you can fit 3 for only a few pence more and get lower failures?). Often end user circuits are self regulating too, which alter the 'real' voltage to get the 'required' one at the actual component.

[Agent]

Its far too late in the evening for me to do this, but show me any evidence that this is not the case? CPUs get hotter over time, cpus require more voltage over time, so overclocks, when system is strained more, need more volts.

It's not the CPU that requires more voltage, it will be the components before it.

Pumping more voltage though a CPU will only cause Electromigration to kick in sooner (http://en.wikipedia.org/wiki/Electromigration). I know of no evidence myself that suggests that more voltage helps slow / solve electromigration in chips.

On the other hand, there are many examples of how pumping more voltage through a chip than is needed will cause it to die quicker.

edit - http://www.csl.mete.metu.edu.tr/Elec...ation/emig.htm is nice too

[SiM]

Its far too late in the evening for me to do this, but show me any evidence that this is not the case? CPUs get hotter over time, cpus require more voltage over time, so overclocks, when system is strained more, need more volts.

I have never needed or known anyone to need to up voltage on a once stable overclock assuming cooling and everything else remains constant.

Wear and tear.

Over time, capacitors and other components wear. Their efficiency drops, thus often more voltage is needed to compensate for this.
It is normally more noticeable in overclocking due to things so often being run right on the edge.
Normal household appliances will normally never have this issue due to the *huge* tolerances they are often build with (why fit one of component X when you can fit 3 for only a few pence more and get lower failures?). Often end user circuits are self regulating too, which alter the 'real' voltage to get the 'required' one at the actual component.

Maybe that argument would be true if the CPU was being run at 4ghz with chilled water cooling or something, but I doubt that is the case for most of our overclocks. Also the speed binning process does not ensure that the stock speed is all the cpu is good for without subjecting it to extra wear and tear... And when most of us overclock we do not run it right on the edge 24/7. We might try to max it for fun, and then drop down to a more sensible setting. My Q6600 is on 3.3ghz (even though it does 3.6ghz+) for the last 7 months and I have not needed to adjust voltage...

Can anyone here say that a full Intel Burn Test stable machine 1 year ago is not stable now (with exact same setup)?

Robert, to me it sounds like the chip is burned out, maybe its only a matter of time until it completely dies.

Edit: didn't see agent's second post

[Jay]

I had a P4 2.8 at 3.4 for years then one day it was no longer stable and needed more vcore. I think it was the board that was getting old though.

[SiM]

I had a P4 2.8 at 3.4 for years then one day it was no longer stable and needed more vcore. I think it was the board that was getting old though.

How many years? It could have been the board/psu was unable to provide stable voltage - not that the cpu has got worn

[Agent]

Maybe that argument would be true if the CPU was being run at 4ghz with chilled water cooling or something, but I doubt that is the case for most of our overclocks. Also the speed binning process does not ensure that the stock speed is all the cpu is good for without subjecting it to extra wear and tear...

Did you even read my post? The CPU does wear and tear, and I've provided links explaining this, but I did not attribute it to required voltage increases though, the opposite!

Over time, capacitors and other components wear

It's not the CPU that requires more voltage, it will be the components before it.

[SiM]

Edit: didn't see agent's second post

^^^^

Your second post does explain it well

[Jay]

How many years? It could have been the board/psu was unable to provide stable voltage - not that the cpu has got worn

I said it was probably the board.

[SiM]

I said it was probably the board.

Yup, I was agreeing with you. But how old was it? I have never had something like this happen to me, but I have never run a system for more than 4 years...