Is "degradation" real or a myth?

[tigersty1e]

degradation is something that is almost impossible to measure on a large scale.

a bios flash alone could change the voltage your cpu gets. temperature changes due to dust could cause instability. windows update that causes instability. power supply would degrade first and give off dirty power. the list goes on and on.

you may need more voltage than previouslya after 3 years, but there are so many variables at work that it would be hard to say the cpu degraded.

 

[aaksheytalwar]

If you dont run at stock, eventually it will degrade. Period

 

[T_Yamamoto]

If you dont run at stock, eventually it will degrade. Period

+1

 

[aaksheytalwar]

However, it may still last a very long time, especially with a mild OC So unless you OC like mad or let temps go out of control or plan to pass the PC on to your kids, stay cool bro

 

[Selenium_Glow]

There is no such thing as big or small electrons. Nor can you send them to the motherboard to avoid material degrading.

I think what he meant is this ->
Big Electron = Holes (positive charge carriers in semiconductors)
Small Electron = Electrons (negative charge carriers in semiconductors).

Ideally, influx of holes into the substrate can cause a larger channel distortion than influx of electrons... or something like that. I barely passed them FET classes in college. :|

 

[Ferzerp]

No, he was really making things up. All electrons are identical as far as we can tell with all our current knowledge.

Feynman even supposed that we could live in a universe that contains a single electron rapidly moving through space and time as all electrons appear exactly the same to us.

 

[truth_benchmark]

I have an AMD A8-3870K 3.0 GHz. Loading BIOS defaults, the core voltage is 1.4125 V

I overclocked it to 3.3 GHz while reducing the core voltage to 1.375 V with LLC set to 1/4. I'm using an aftermarket cooler. Core temperature never reaches 50 C while running games or encoding videos

To the experts here, should I be worried about degradation ?

 

[aaksheytalwar]

That isn't much of an oc. Prolly no prob for 1.5-2.5+ years at least, maybe even 3+ years

 

[Avalon]

My Core2 E8400 running 4Ghz @ 1.28V degraded after around 4 years. It now needs 1.29V to prime large FFT at that speed. I just dropped it down to 3.8 @ 1.21V instead.

 

[LOL_Wut_Axel]

That isn't much of an oc. Prolly no prob for 1.5-2.5+ years at least, maybe even 3+ years

Very unlikely to have problems for 5 years if not more, actually.

 

[exar333]

Thanks for that.

Same for me. Had a 90nm Winchester that ran at 2.5ghz for years and then would barely run at 2.2ghz, and then further fell to 2.0ghz after that. Saw similar issues with my 2160 that I ran at about 3.0ghz for a while, and then had to throttle it back.

 

[Smartazz]

I've never seen a chip die from degradation, but I had a Q6600 that required more voltage as it aged. Could have been the motherboard's VRMs, the CPU or likely both. I'm fairly confident that my modest 2500K overclock should last me beyond its usefulness.

 

[aaksheytalwar]

I've never seen a chip die from degradation, but I had a Q6600 that required more voltage as it aged. Could have been the motherboard's VRMs, the CPU or likely both. I'm fairly confident that my modest 2500K overclock should last me beyond its usefulness.

If eventually it needs high volts even for stock speed, like 7-10 years down, that is literally dieing

 

[Martimus]

I have written this a few times, but I will repeat it here again. The more current that travels through a transistor, the higher its breakdown region becomes. The higher its breakdown region becomes, the higher the voltage you need to supply it to make it act like a switch. So you will need to increase the voltage to get it to work, which will increase the current load, which will increase the rate at which the breakdown region creeps upward.

Now I haven't worked in the field in nearly 8 years now, so my terminology may be off, but I tested thousands of components and the increase in the breakdown region was repeatable for all components as you increase the current flowing through it. This makes logical sense as well if you think of the physics behind the movement of electrons through doped silicon.

 

[aaksheytalwar]

I have written this a few times, but I will repeat it here again. The more current that travels through a transistor, the higher its breakdown region becomes. The higher its breakdown region becomes, the higher the voltage you need to supply it to make it act like a switch. So you will need to increase the voltage to get it to work, which will increase the current load, which will increase the rate at which the breakdown region creeps upward.

Now I haven't worked in the field in nearly 8 years now, so my terminology may be off, but I tested thousands of components and the increase in the breakdown region was repeatable for all components as you increase the current flowing through it. This makes logical sense as well if you think of the physics behind the movement of electrons through doped silicon.

I said the same thing in layman's language

 

[Martimus]

I think what he meant is this ->
Big Electron = Holes (positive charge carriers in semiconductors)
Small Electron = Electrons (negative charge carriers in semiconductors).

Ideally, influx of holes into the substrate can cause a larger channel distortion than influx of electrons... or something like that. I barely passed them FET classes in college. :|

I think you mean Proton instead of "big electron".

Also, a hole doesn't actually have a positive charge, but instead is an atom with a missing electron meaning that it will accept an electron. At least, this is the way I remember it, although it has been a while for me.

 

[Martimus]

If you dont run at stock, eventually it will degrade. Period

Actually, the chip will degrade at stock too, just at a slower rate.

 

[soccerballtux]

I think what he meant is this ->
Big Electron = Holes (positive charge carriers in semiconductors)
Small Electron = Electrons (negative charge carriers in semiconductors).

Ideally, influx of holes into the substrate can cause a larger channel distortion than influx of electrons... or something like that. I barely passed them FET classes in college. :|

No, he was really making things up. All electrons are identical as far as we can tell with all our current knowledge.

Feynman even supposed that we could live in a universe that contains a single electron rapidly moving through space and time as all electrons appear exactly the same to us.

+1 to the first statement

I was being facetious

 

[nyker96]

Even chip degradation is just minor, maybe a 4ghz now doing 3,8 etc, but that takes like 4-5 yrs. so I wouldn't worry about it too much.

 

[2is]

Actually, the chip will degrade at stock too, just at a slower rate.

Well it doesn't count if you'll die before the chip

 

[aaksheytalwar]

Even chip degradation is just minor, maybe a 4ghz now doing 3,8 etc, but that takes like 4-5 yrs. so I wouldn't worry about it too much.

A 3.4 stock at 4Ghz will take 3-5 years. But a 2.5 at stock will degrade from 4 in under 2 years max.

 

[OVerLoRDI]

Yep. My 6950s are a great example. They have led a really rough life, spending most of it at 100% load bitcoining. They are watercooled, so temperatures haven't been an issue, but over time I have had to lower the clock speed.

1015mhz @ 1.3v initially
980mhz @ 1.3v 1 year later.

So I have lost 35mhz. This kind of degradation is exacerbated by my overclocking and near constant load on these chips.

 

[truth_benchmark]

For overclocking that will be used for 24/7, my self-imposed rule is to use the highest stable overclock speed without increasing core voltage

Anybody here also doing this?

 

[aaksheytalwar]

I don't do exactly this. But I don't cross limits either. My current 2600k is at 1.27V and my new 3770k will hopefully be btw 1.15-1.2V

 

[Smartazz]

Yep. My 6950s are a great example. They have led a really rough life, spending most of it at 100% load bitcoining. They are watercooled, so temperatures haven't been an issue, but over time I have had to lower the clock speed.

1015mhz @ 1.3v initially
980mhz @ 1.3v 1 year later.

So I have lost 35mhz. This kind of degradation is exacerbated by my overclocking and near constant load on these chips.

What kind of temps were they sitting at?