FAQ: Computer Burn-in: Investigated - Overclock.net

S.M. · 09-03-08

Computer Burn-in myth: Investigated. (Not officially defined)

Many people have read or researched computer enthusiasts say things such as "To get your computer to overclock higher and more stable you need to burn it in!" This is completely misrepresented, and for the most part untrue. But not necissarily false.

The thought of burning in a somewhat unstable overclock for a long period of time is just rediculous. Parts of a computer are in a solid state and are as they are. They do not change for the better under stress like a human muscle, or like a freshly rebuilt motor.

I can see how people who do excercise frequently, or have experience with combustion engines might have transferred success with those to success with computers. But for the most part it's internet hogwash.

A word every computer enthusiast should know is electromigration.

Electromigration: Electromigration is the transport of material caused by the gradual movement of the ions in a conductor due to the momentum transfer between conducting electrons and diffusing metal atoms. The effect is important in applications where high direct current densities are used, such as in microelectronics and related structures. As the structure size in electronics such as integrated circuits (ICs) decreases, the practical significance of this effect increases.

TL;DR = Electromigration is a process at which electronic components transfer electricity more efficiently with use. It happens at all operating time, on any electronic item.

This is the closest thing to "burn-in" you can get with a computer. The only way to effectively speed-up electromigration is to LOWER your clock frequency, and raise voltage. By doing so for about about week will show some improvements. Unfortunately, electromigration takes so long to work, that the only imrpovement you should see is a higher clock frequency reached at stock voltages. Any overclock success at higher frequencies is on a "remain to be seen" level, and un-gauranteed.

Fun stuff I discovered when researching CPU burn in.

Refutable overclocking resource: "A system loses it's overclocking ability when operated at highly overclocked settings of high HTT/FSB and high Vcore for extended periods of time"

Globally accepted definition for Burn-in: "operating at highly overclocked settings for extended periods of time"

Hmmm?

Many people who do in fact burn-in their computers, will soon come to realize that in months time, they're computers will become more unstable then their initial overclock health status. This does not necissarily mean completely unstable. Side effects come on slowly, like cancer. First it will be an occasional BSOD. Intermittent gaming lag, or freezes. And it will soon turn into failing a computer stress test and lowering their overclock. These effects can take much longer than months as well.
It could take years. But it will eventually happen. This doesn't happen to everyone. But fair estimate of more than 50% of computer overclocking enthusiasts have had this happen to them, in some form or another. Again, these are the people that made an unstable overclock stable AFTER computer burn-in. Which isn't that many in general.

dralb · 09-03-08

Quote:

This is the closest thing to "burn-in" you can get with a computer. The only way to effectively speed-up electromigration is to LOWER your clock frequency, and raise voltage. By doing so for about about week will show some improvements. Unfortunately, electromigration takes so long to work, that the only imrpovement you should see is a higher clock frequency reached at stock voltages. Any overclock success at higher frequencies is on a "remain to be seen" level, and un-gauranteed.

Why lower the clock? Why would you want to increase electro-migration?

jarble · 09-03-08

humm great thought. Cant you also burn in for a new heatsink or thermal paste ?

S.M. · 09-03-08

Quote:

Originally Posted by dralb

Why lower the clock? Why would you want to increase electro-migration?

Lowering the clock allows you to lower the stress on the computer part, while passing more electricity through it. Speeding up the effects of electromigration is simply making that electronic part more efficient at transferring electricity. It's like making your CPU process a river for awhile. Then making it process a creek. It will do it easier and more efficiently.

Quote:

Originally Posted by jarble

humm great thought. Cant you also burn in for a new heatsink or thermal paste ?

Yes, but thats a totally different playing field. You don't need to overclock or change voltages for that. Curing thermal paste is not all about running your stressing your CPU. It's of two, or maybe even three parts. Just having AS5 out of the application tube starts the curing process, but the transfer of heat does slightly move AS5 particles better into groves of both the CPU and heatsink, allowing for better heat transfer. This is irrelevant now that AS5 is becoming obsolete, and new thermal pastes do not need time to cure.

EDIT: Now that I think back on it, I believe the way AS5 cured was the action of cooling down and heating up. Allowing the particles to shrink and expand, shrink and expand, shrink and expand until they wedged themselves well in the minute cracks of the heatsink and CPU. I may be wrong though.

jarble · 09-03-08

Quote:

Originally Posted by S.M.

CPU Burn-in myth: Defined.

"To get your computer to overclock higher and more stable you need to burn it in!" This is completely misrepresented, and for the most part untrue. But not necessarily false.

during a burn in problems become apparent where they might not show up normally

S.M. · 09-03-08

Quote:

Originally Posted by jarble

during a burn in problems become apparent where they might not show up normally

Overclock stress testing and burn in are not one in the same.

Usually when someone fails an overclock stress test, they change settings until they don't fail a stress test.

Burn in, is when someone fails a stress test. And continues to stress that configuration in other forms to where it doesn't give as many errors, or doesn't error at all. This is usually done with Memtest errors, and not usually prime or orthos errors.

theCanadian · 09-03-08

Oh wow... I define burn in as running it for a while to make sure parts don't fail, your paste is curing, stability. I think that it follows logically that if you make your computer do something that it doesn't want to do, your going to have problems down the road.

S.M. · 09-03-08

Quote:

Originally Posted by theCanadian

Oh wow... I define burn in as running it for a while to make sure parts don't fail, your paste is curing, stability. I think that it follows logically that if you make your computer do something that it doesn't want to do, your going to have problems down the road.

That would be stress testing/Health monitoring.

And that's what I thought too.

TheSubtleKnife · 09-15-08

Electromigration is a BAD thing...its what kills chips.

Quote:

Electromigration is the transport of material caused by the gradual movement of the ions in a conductor due to the momentum transfer between conducting electrons and diffusing metal atoms. The effect is important in applications where high direct current densities are used, such as in microelectronics and related structures. As the structure size in electronics such as integrated circuits (ICs) decreases, the practical significance of this effect increases.

That just means that the electrical traces get broken/shorts because of the ion movement within the chip, and this happens at any voltage, but is increased when you put more voltage through (due to more electron flow).
You wonder why the 45nm's die so fast with 1.4+v? because the gaps between the electrical traces and such are smaller and closer together, raising the chances of a short/circut break.
One particular example was back in the P4 days where if you put like 1.8v through a northy, you'd get SNDS ( Sudden Northwood Death Syndrome):

Quote:

When core voltage (Vcore) was increased past 1.7 V, the processor would slowly become more unstable over time, before dying and becoming totally unusable. This became known as Sudden Northwood Death Syndrome, which is caused by electromigration.[12]

tl;dr... Electromigration is bad

S.M. · 09-15-08

Quote:

Originally Posted by TheSubtleKnife

Electromigration is a BAD thing...its what kills chips.

That just means that the electrical traces get broken/shorts because of the ion movement within the chip, and this happens at any voltage, but is increased when you put more voltage through (due to more electron flow).
You wonder why the 45nm's die so fast with 1.4+v? because the gaps between the electrical traces and such are smaller and closer together, raising the chances of a short/circut break.
One particular example was back in the P4 days where if you put like 1.8v through a northy, you'd get SNDS ( Sudden Northwood Death Syndrome):

tl;dr... Electromigration is bad

That's exactly what I said.

But it's a fact, if you raise the vcore without raising the clock speed for extended periods of time. You will see a higher stable overclock at stock voltages and possibly higher.

It DOES put extreme strain on a CPU. And it does deteriorate the CPU. But for the effects of electromigration to kill a CPU takes years. Your argument of raising vcore on a northy to 1.8 is irrelevant. Because that isn't in the same vicinity as the effects of electromigration for stability benefits. That's just overloading the chip. A northy just can't process or transfer 1.8v effectively or at all, causing the electrons to find a route to take, usually jumping from traces to who knows where.

		Overclock.net - Overclocking.net > Intel > Intel - General
FAQ: Computer Burn-in: Investigated

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)