What kills the CPU?

[AMDXP]

This is a real dumb question for Overclocking....

But what really kills the CPU

Is it the voltage? Heat or the Speed increase?

 

[mikelz85]

usually it will be voltage/heat, causing electron migration. As far as a quick death goes, I'd imagine very high vcore to be the #1 cause. Though, you don't really here about too many people killing their CPUs.

 

[AMDXP]

So just upping FSB wont hurt it correct?

 

[Ice Czar]

well you asked for it
normally its just buried in the back of the ESD & Electromigration Rant in the PSU Power FAQs thread
>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
Electromigration is the mechanisim that typically first degrades and then kills IC Chips (Integrated Circuits)

Semiconductor Electromigration In-Depth @ DWPG.com < the totally understandable link (if somewhat dated these days in particulars)

What is electromigration?

Harris Semiconductor Lexicon of technical terms puts it this way:

"Motion of ions of a metal conductor (such as aluminum) in response to the passage of high current through it. Such motion can lead to the formation of "voids" in the conductor, which can grow to a size where the conductor is unable to pass current. Electromigration is aggravated at high temperature and high current density and therefore is a reliability "wear-out" process. Electromigration is minimized by limiting current densities and by adding metal impurities such as copper or titanium to the aluminum."

Electromigration is an effect that occurs when an extremely dense electron flow knocks off atoms within the wire and moves them, leaving a gap at one end and high stress at the other. In a chip, the formation of such a void will cause an open circuit and result in a failure. At the other end, the increase of stresses can cause fracture of the insulator around the wire and shorting.

Electromigration and Voids @ Cornell

Electromigration Simulator @ MIT
processing temperature -- increasing the processing temperature will increase the initial tensile stress present in the line due to thermal mismatch, thereby leading to earlier failures. To neglect the thermal stress, set the process temperature the same as the test temperature.
test temperature -- increasing the test temperature will dramatically reduce the calculation time since diffusivity is follows the temperature by an Arrhenius relationship.

the Arrhenius equation

roughly translating to this rule of thumb
Each 10°C (18°F) temperature rise reduces component life by 50%
Conversely, each 10°C (18°F) temperature reduction increases component life by 100%.

however there is more to it
http://www.triquint.com/company/quality/faqs/faq_07.cfm (caution heavy wading)
"Electromigration mechanisms are accelerated by current density as well as temperature. The general relationship is sometimes referred to as Black's Equation. Just as with the Arrhenius equation, we can observe the electromigration effects on lifetimes using a graphical approach."

so, both elevated temperature, and voltage, can cause voids to form in the circuits of any chip, a problem that becomes more and more important as the number of atoms that comprise the width of that circuit decrease, (that first link was written when the manufacturing scale was at 0.18 microns we are now at .09 microns (90nm) with some chips) in addition the clock rates that the power is cycled through is all that much higher as well, making the chips all that more suceptible to ESD, voltage irregularities and temperature.

and a PDF addendum

Componnt Reliability Turtorial
actually not bad reading if you skim through it

 

[BillR]

Wow, didn't take much to see that one coming from a mile away

 

[USMC2Hard4U]

What kills a CPU exactly?

Ever hear of a Heatsink called the "Thermoengine" Grrrrr

It ate 2 of my T-Birds...

 

[AMDXP]

So technically you can think of it like a car the faster you go the more fuel you burn (Power in this case maybe) and more miles get put on it faster. The hotter the car gets the more likely it is to stall out.... or over heat.

 

[Ice Czar]

power and heat are two sides of the same coin
electromigration in an exponential phenomena
its always happening but gets massively worse as the temperature and current increases
thus it takes just a split milisecond to fry an IC with static

here is an exponential curve

so its absolutely critical that both the current and the tempurature keep the level of wear in the flat at the bottom or the heel of the "hockey stick", as soon as you allow it to climb things will go to hell in a handbasket in quick order

edit > well that gif sux on a black background
but you can see the curve, and thats pretty representative of how steeply it would climb

under volting w\ high temperature or over volting with extremely low temperature can still see unacceptable electromigration, obviously the lower your temprature the more overvolting is possible, but the point is that its not a directly observable phenomena thus overclocking is an art form and a craft,m there is hard science behind it but even IC manufacturers have to "go see" what the particular level of integrity a given IC has, in fact overclocking itself is a form of testing a chip

again from the ESD & Electromigration faq

This board is filled every day with people who have developed RAM errors, data corruption problems (generally RAM) ect, Most of which can be traced to either poor power regulation (Transient Response\Load regulation - overshoots) of the PSU, or ESD

Latent defects caused by ESD in any IC (and they are just everywhere from HDDs to NIC, CPU, RAM ect) are massively underated as a cause of problems. If you have eliminated power fluctuation problems (PSU voltage regulation and power conditioning) and still experience a component failure, odds are that it was a latent defect, either from installation, or one that wasnt caught during manufacturing.
the membership displays a cavalier attitude towards this issue for 2 reasons, RMA's are pretty easy, and they rarely employ the same component for its fully rated lifespan, upgrading before the eventual premature failure becomes appearent.

But
a latent defect, not only effects the lifespan, it degrades the performance of the IC as well, and is often the difference between the "Golden Chip" benchmark leader, the norm, and "why cant I get the same OC as this guy? Ive got the same components"

 

[Empyrean]

edit > well that gif sux on a black background

Just highlight the pic and then it looks normal and you can read thetext.

 

[madmat]

It's also possible to blow holes in the insulating layers in stacked layers of IC's thus causing a short between layers which will lead to the death of the chip but usually when that happens the chip goes out rather noisily.

 

[Ice Czar]

It's also possible to blow holes in the insulating layers in stacked layers of IC's thus causing a short between layers which will lead to the death of the chip but usually when that happens the chip goes out rather noisily.

pretty sure thats what they mean by stress on the other end of the circuit and shorting

the atoms migrate to that end from the created voids and bust the insulator through deposition

a little google work will turn up some not so cool (pun intended) electron microscope movies
from places like Cornell, I used to link them but they kept changing

 

[Bullitt]

What kills a CPU exactly?

Ever hear of a Heatsink called the "Thermoengine" Grrrrr

It ate 2 of my T-Birds...

Wow, I gotta sorta agree and disagree. I've had that sucker on my 1.0ghz for the past 4 1/2 years Its been pretty good to me, but it probably sucks for everyone else

My only gripe was the retention clip is under so much tension, I have killed a motherboard using it when the screwdriver slipped and into the PCB it went. I'll have to chalk that up to noobishness, being it was only my 2nd socket-A build.

 

[[Spectre]]

This is a real dumb question for Overclocking....

But what really kills the CPU

Is it the voltage? Heat or the Speed increase?

Reptiles