How does CPU degradation work?

[Hans Power]

Hello!

 

I'd like to know how CPU degredation manifests itself on my overclocked CPU (Phenom II X4 965 3,4Ghz stock). Mind you that I overclock not as much as a hobby but more out of the actual need for more performance in gaming and I don't have a comfortable budget, yet to upgrade to what I want. So longevity is a concern.

Now I had my CPU running at 4Ghz for about a year and it ran absolutely stable. But then I suddenly got crahes when viewing Youtube videos or during gaming at random and it crashed after about 5 minutes in an OCCT test. After a while I figured out that my CPU couldn't take the overclock anymore so I reduced the clock speeds to 3,7Ghz although it actually can still handle 3,9Ghz just fine (was running 30min of OCCT Test to make sure of that).

Now, I wonder if should go with those 3,9Ghz or for 3,7Ghz and give the CPU some more breathing room. Now, for that I'd like to know how CPU degredation actually works so that I can determine the risk of my overclocks a little bit better. I'd also apreciate it if you could describe your personal experiences with this if you have any - that would also help me a lot to make an educated guess for my own CPU. Thanks!

[ZetZet]

Well everything degrades over time, especially things that heat up and cool down over and over again. I think after a while CPU's can't handle higher voltages, not sure how it works though.

[Hans Power]

Well everything degrades over time, especially things that heat up and cool down over and over again. I think after a while CPU's can't handle higher voltages, not sure how it works though.

 

Yeah, that's as far as my knowledge goes as well. Unfortunately that's not enough to make an educated guess about longevity.

[i_build_nanosuits]

should give this a read, it's based on old intel but all the info's still applies the same for AMD CPU's..

 

http://www.anandtech.com/show/2468/6

 

Your CPU BTW is also a 0.045 micron SOI (45nm) just like the intel cpu discussed in this brilliant arcticle.

[Hans Power]

should give this a read, it's based on old intel but all the info's still applies the same for AMD CPU's..

 

http://www.anandtech.com/show/2468/6

 

Your CPU BTW is also a 0.045 micron SOI (45nm) just like the intel cpu discussed in this brilliant arcticle.

 

Yes, the article is pretty awesome, thank you! I'm not sure if I understood it all - I'm not a native speaker after all and that's pretty heavy stuff but I'll give it another read and soldier my way through it. But the message is pretty clear - overclocking has a very measurable impact to longevity. I will go the middle of the road then and go for 3,8Ghz right now - running the CPU on it's maximum doesn't seem like a good idea.

[i_build_nanosuits]

Yes, the article is pretty awesome, thank you! I'm not sure if I understood it all - I'm not a native speaker after all and that's pretty heavy stuff but I'll give it another read and soldier my way through it. But the message is pretty clear - overclocking has a very measurable impact to longevity. I will go the middle of the road then and go for 3,8Ghz right now - running the CPU on it's maximum doesn't seem like a good idea.

you got it all...the closer to the max safe voltage you get the greater the degradation and it's exponential, this means it goes faster and faster as you go...and temperatures is the other key factor here, keep it as low as possible...as shown in this graph:

 

[Hans Power]

you got it all...the closer to the max safe voltage you get the greater the degradation and it's exponential, this means it goes faster and faster as you go...and temperatures is the other key factor here, keep it as low as possible...as shown in this graph:

 

 

Yeah, temperatures are actually a thing in my system. Under full loads core temperatures go up to 65°C while AMD recommends 62° under full loads (doesn't matter if I go 3,7 or 3,9Ghz, btw - temperature don't differenciate much after a certain overclock). Hmm, I actually really don't want to buy a new cooler cause that is supposed to be a part of a later upgrade and I'm not sure if I can fit anything I want (Corsair H60 or a Scythe ashura) for the new system in my current case (Thermaltake matrix).

[Admire]

I would really be interested in a huge long term study of this with different CPUs, temps and so on.

 

A friend of mine has one of those enthusiast CPUs that has run very well for about 11 years on a stock cooler. Im rather impressed with that to be honest.

[harrynowl]

Well I degraded the FX8320 I'm using currently. It used to do 4.8 without a hitch now module 3 is a dog and won't boot unless I chuck 1.5v at it.

[i_build_nanosuits]

Well I degraded the FX8320 I'm using currently. It used to do 4.8 without a hitch now module 3 is a dog and won't boot unless I chuck 1.5v at it.

The more watts a CPU use the faster it tend to degrade i believe...some intel atom chips in tablets will last a life time i'm sure...but their screen will die first.

Many FX bulldozers are dead already.

[harrynowl]

The more watts a CPU use the faster it tend to degrade i believe...some intel atom chips in tablets will last a life time i'm sure...but their screen will die first.

Many FX bulldozers are dead already.

Well with a sample size of around 50 CPUs at the shop there are more non working AMD CPUs. For Zambezi chips to be dying within 3 years of release is a bit poor

I'd imagine the vcore / tdp / heat output is responsible as you suggested.

I just need it to hang on (It defaults to 1.45v in BIOS. Ouch.) until I've got my bike (why I sold 2600K) As I'll be moving to ITX and probably something like 4690K if we hear nothing about new broad well things.