• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

What really shortens the lifespan of a CPU/GPU?

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

Lone

New Member
Joined
Dec 17, 2015
Hi, so I have been wondering about this for a long while now. My question is which part is the one that really kills/shortens the lifespan of a CPU or a GPU, the voltage itself or the temperature?
Here is an example(This is not fact, I'm just making this up to try to make a point), Let's say I'm at 4.0GHz at [email protected] and another example is at 5.0GHz at [email protected], which factor is really the one that shortens the life? At a lower voltage but at very high temperature or at very high voltage and at lower temperature? I hope you guys see the point I'm trying to make here. If this is confusing, I'll try to rephrase it a bit, thanks.
 

ShrimpBrime

~MadHatDeLidder~
Joined
Apr 19, 2012
You'd want to be specific here. I mean 1.5v on Intel you wouldn't want to do daily. However AMD it may be OK.

If you push say 1.9-2.0v you are using LN2. This type of overclocking is harmful but the freezing temps help prevent instant damage.

So you could pick a particular processor and type of cooling used and we could tell you where the safe limits would be for daily use and such.

But there is nothing set in stone while overclocking. There is always slight risks here and there.
 
OP
L

Lone

New Member
Joined
Dec 17, 2015
I'm on an Intel i7 5930k on Asus Rampage V Extreme with Evga 980 ti Hybrid atm. I'm using fully EKWB custom water cooling on my CPU only since I'm okay with the hybrid's temperatures.

Here's the EK parts I picked:
CPU block: https://shop.ekwb.com/catalogsearch/result/?q=3831109800324
Radiator: https://shop.ekwb.com/ek-coolstream-xe-360-triple
Pump: https://shop.ekwb.com/ek-xtop-revo-d5-pwm-plexi-incl-pump
Reservoir: https://shop.ekwb.com/ek-res-x3-250

The radiator will have 6 of these fans on it: https://shop.ekwb.com/ek-vardar-f3-120-1850rpm
 

Tech Tweaker

Member
Joined
Dec 14, 2010
Well, I know that 1.7v on an AMD CPU (socket 939) with air cooling can kill the CPU. I know because I killed two of them doing that, I also had temps around 65°C (so, a bad combination). By all accounts, the system should have shut down due to thermal protection, but it didn't. I DON'T recommend trying that, two was enough for me.

Generally, don't go past 1.55v for air cooling with AMD is the recommended maximum I believe. Probably want to keep it to 1.45v or below for normal 24/7 operation (for sockets 754 through AM3 (AM3+ might be different, I have no experience with it)).

For Intel I believe the recommended max is 1.45v for LGA775, 1.35-1.4v for LGA1156, 1.4v for LGA1366 (which tends not to be an issue, as most of the one's I've tested hit a thermal wall long before they get close to needing 1.4v), LGA1155 is somewhere around 1.35-1.4v I believe, LGA1150 is around 1.3-1.35v if I remember correctly. LGA1151 no idea, LGA2011 no idea, LGA2011-3 no idea.

These are all numbers for air cooling, water cooling setups might be a little different.

Correct me if I'm wrong, in case I'm off on some of these.
 

I.M.O.G.

Glorious Leader
Joined
Nov 12, 2002
Location
Rootstown, OH
Voltage really does it. Temperature can aggravate the issue under normal conditions, but its the voltage that really kills.

For example under LN2, you can have a CPU thats kept under -100C but dies due to voltage being too high.

That said, under good conditions a processor/GPU can last a decade without issue - its typically other components that would fail first. Even if you cut that lifespan in half, few people want to be using the same processor 5 years from now... So I take this into consideration when people worry about shortening the lifespan of a chip through overclocking. Worst case scenario if you are pretty brutal, you may take years off your chips life... But unless you are really treating it poorly, it isn't going to die on you within a reasonable usable timespan (1-5 years in my book).
 

Mr.Scott

Beamed Me Up!
Joined
Jun 9, 2013
Voltage really does it. Temperature can aggravate the issue under normal conditions, but its the voltage that really kills.

For example under LN2, you can have a CPU thats kept under -100C but dies due to voltage being too high.

That said, under good conditions a processor/GPU can last a decade without issue - its typically other components that would fail first. Even if you cut that lifespan in half, few people want to be using the same processor 5 years from now... So I take this into consideration when people worry about shortening the lifespan of a chip through overclocking. Worst case scenario if you are pretty brutal, you may take years off your chips life... But unless you are really treating it poorly, it isn't going to die on you within a reasonable usable timespan (1-5 years in my book).
I tend to disagree with you on this.
Case being, I have many socket A processors that have seen years of 2.0v 24/7 on water, that still run to this day. That's 10 years of severe over voltage bro. As long as you keep them cool, they're fine.
Same with 939, I have processor's over 1.7v 24/7 still kicking.
Voltage if it's going to kill, will kill almost instantly. Heat is a slow death. If you keep them cool it's like life support.
 

Dolk

I once overclocked an Intel
Joined
Mar 3, 2008
IMOG is correct but Voltage is more of a secondary that is seen as a primary to an end user. The real answer comes down to silicon and interconnect materials.

As Mr. Scott has pointed out, an older Socket A and 939 were developed on larger MOSFET designs (90nm and larger). Because of this size, the voltage is able to move through the materials without damaging the lines. We have larger resistances and lower capacitance inside each FET and interconnect. At higher speeds, voltages, and lower temperatures, the materials are more adapt at absorbing the abuse for longer periods.

Bring into today's CPUs and its a different ball game. The CPUs range from 14nm to 45nm but really the FETs that will have a shorter lifespan are the FIN/Tri-Gate FETs. These FETs have a much lower resistance and a higher amount of capacitance due to the thinning materials. At greater voltages, faster speeds, and still the same temperature range as the older chips, these CPUs will give out quicker. Especially if there are a lot of unexpected ripples in the voltage line.

Now granted, creating a comparison of life span between these chips would require a lot of data. We would need multiple chips of the same family, and each would have to run in a different state for more than 5yrs. So the true answer could be left up to silicon processing at EACH fab plant (Every FAB plant, even all of Intel's, create and produce CPUs a bit differently).
 

JrClocker

AKA: JrMiyagi
Joined
Sep 25, 2015
Dolk is correct...voltage kills first.

All things being equal, electronics have a "bathtub" reliability model...they fail either really early or after many years. Running a part at its maximum rated temperature will shorten its lifespan (primarily due to electron migration (or electromigration)...higher temperatures make this easier).

As Dolk stated, the smaller the fabrication process, the larger the capacitance becomes. Additionally, the smaller the fabrication process, the closer the two sides of the capacitor are to each other (which makes the capacitor larger). Any dielectric (non-conducting material) is rated for its dielectric strength in terms of Volts per meter (V/m). Once you exceed the V/m rating of a dielectric, it will conduct electricity...and no longer be an insulator. As the process gets smaller, the capacitor dielectric gets thinner, and fails at a lower voltage.
 

Woomack

Benching Team Leader
Joined
Jan 2, 2005
At least on new chips it's always about the balance. If you keep reasonable temps and voltages then CPU won't degrade so fast. If you set too high voltage then it can be instantly damaged but if you keep it at too high temps for too long then it will damage too. Temperature isn't so big issue lately as most new series or throttle or shut down when temp is too high.
Let's say it looks like voltage may damage CPU instantly, temps in most cases can't.
 

t1nm4n

Member
Joined
Jun 12, 2005
Location
Texas
In short, everyone is correct, Voltage and heat are the real killers, you can't narrow it to one or we will all start fighting to the death. But really its What Dolk said, its the materials used and fabrication of said materials. In a perfect world and a perfect processor there would be no leakage of electicity or resistance to generate heat and degrade the materials it was made with, all CPU/GPUs are doomed to failure from the moment they were put into production, they are just ticking time bombs waiting to ruin your wonderful day of fragging.
 

wingman99

Member
Joined
Dec 10, 2003
Voltage mostly causes Degradation

Well more voltage makes heat that you can't see with CPU because you are dissipating the heat to quickly at upper level of the chip with great coolers or LN2, not at the atomic level of voltage and AMPS = power.

Think of it like this incandescent light bulb you can cool the bulb however the filament still runs hot.

Another example when you have a small wire and you cool the wire with LN2 you apply more voltage than is mathematically calculated for material also diameter it will still burn up.

Another example at the atom level of a IC chip you have degradation when the electron's tears away metal from transistors and tracers do to force of electron's, like going through a tunnel for the force of water rips the tunnel apart.

So cooling helps the material layers of the tunnel the electron is flowing through to help from melting material above a example of 110c, compared to the force of the electrons tearing with heat like a arch welder.

Cooling helps allot for changes in tolerance in tracers also transistors in a IC Chip, when you heat the conductive material it expands changing the resistance, so cooling make it more stable to it's original size. That is why cooling helps with overclocking.

Hopefully this is really simple to understand because electrical power seeks the path of least resistance and it's hard to visualize so I said tunnel.