Symtomns of CPU degradation from overclocking?

[Dave3000]

I just wanted to know if anyone here has experienced their CPU degrading from overclocking especially with too much vcore. Does the CPU just die and result in an inoperable system, or do you get mysterious errors, crashes, bsod, corrupted installations of programs or corrupted files, while using your system? Would the user be in a stressful stage when the CPU nears it's death or is degraded but system still functioning because of these problems I was asking about? Is it possible that installing Windows can go wrong behind the scenes, such as corrupting files during installation without user's awareness, because the CPU is almost degraded to death from overclocking, especially with high voltages?

 

[FalseChristian]

I've got my i5 2500K overclocked to a reasonable level. It's at 4.5GHz using only 1.34v at full load so mine won't degrade at all and will last me for years to come.

You'll get crashes, BSOD, restarts and all sorts of nifty things when you're CPU is dieing.

 

[2is]

If you're installing windows on an unstable system, yes, it's certainly possible for the installation to be corrupted.

Also, sometimes "CPU degradation" can actually be motherboard degradation. My previous system with a Q6600 was overclocked to 3.2GHz the entire 4+ years I had it. Towards the later part of its life I had to periodically increase voltage to maintain stability. I thought it was the CPU until I moved that CPU into another system. I was able to hit 3.4GHz at a lower voltage than I was using for 3.2GHz on the previous board.

 

[Ben90]

I've got my i5 2500K overclocked to a reasonable level. It's at 4.5GHz using only 1.34v at full load so mine won't degrade at all and will last me for years to come.

What settings do you use? My motherboard has this pesky setting that forces it to "Use Physics". As such, even when running stock, my processor is still internally degrading due to many factors.

 

[BrightCandle]

That sounds like an unstable CPU overclock to me.

CPU degredation for me has meant a CPU that was 24/7 stable loosing stability and the voltage having to be increased and then the frequency being decreased in order to keep it stable. My i7 920 went from 4.0 Ghz down to 3.6 Ghz over its 2.5 year lifetime while maintaining the same voltage.

But in xtremesystems and such I have heard of people popping their CPUs very quickly with large amounts of voltage (1.4V) probably to blame.

 

[JimmiG]

It will behave exactly like an unstable overclock - BSODs, application crashes, reboots, silent data corruption etc. and it will fail stress tests that it used to pass at the same settings.

Increasing the voltage and/or reducing the overclock will fix it. However if the CPU has already started degrading, chances are it will eventually become flaky again at the new settings, forcing you to repeat the process. Eventually it won't even work reliably at stock clocks/voltages.

This is very rare with moderate overclocks, in my experience. It's mostly an issue with extreme overclocking. A friend had it happen to his Duron 600 MHz @ 950 MHz - I believe he took the overclock down by ~100 MHz and used it for another year or two.
I've had it happen to a Radeon X800XT GPU that wasn't even overclocked, resulting in visual artifacts. I underclocked it a little bit, and used it for another year, before upgrading to a faster video card.

The first thing to check when a previously stable overclock becomes flaky is that all the fans are spinning and that the heatsinks aren't clogged with dust.

 

[ShintaiDK]

You can also get permanent transistor damage that no voltage/underclock will fix.

 

[PrincessFrosty]

It will start with artifacts and errors in calculations which may simply go unnoticed as largely irrelevant mistakes in calculation or it may lead to application crash or in the case of the OS a BSOD.

If you run something like prime95 where you're calculating known values it can test for hardware error by seeing if a calculation error is made anywhere, if that's occurring you need to back down your clocks/voltage.

Generally speaking higher vcore will degrade the CPU faster, when that occurs you'll require higher vcore to sustain the same overclocks, so you might start finding you need to up the vcore to return stability to an otherwise previously stable overclock. Or if you're at the limit of what the CPU can handle you'll have to bite the bullet and just lower the clocks to maintain stability

I've never seen a CPU completely burn out, they tend to go through a slow degradation of needing to lower voltages and clocks to return stability. Note that more aggressive overclocks and vcores cause faster permenent degradation, and that degradation tends to accelerate over time, it will take a while for a new CPU to start to see permanent degradation but after that you'll find it degrades faster with time/usage.

The typical lifespan for even heavily overclocked components is fine though, by the time it's a serious issue you'll have long since replaced the hardware, or the cost of an equivalent replacement would be very small.

 

[cytg111]

That is one tall order, I've had cpu's Id have to back down to get stable, but the reason? Cant say that for sure, cause I didnt record thermals before/after, maybe it was the ram, maybe it was the psu, harddrive (from oc'ed bus) maybe maybe. But off the bat, then yes, I've had oc'ed cpu's i've had to back down to get stable (after the obvious tasks, cleaning out dust etc.)

 

[cytg111]

It will start with artifacts and errors in calculations which may simply go unnoticed as largely irrelevant mistakes in calculation or it may lead to application crash or in the case of the OS a BSOD.

- And while artifacts and errors is a sign, it could as well be a sign of bad software. As a reference point it would have to be the same software, same drivers(especially for games), same instance of the OS(patches etc) to make the comparison. Not easy to pin down. In the end it comes down to some intuitive statistics that might well be wrong.(read: I hate it what that happens).

 

[john3850]

Twice my CPU degradation was from the mb and not the cpu.
Both times it took over 3yrs to show up once on a p3 and once on c2d.
What I throught was CPU degradationon on 500e turned out to be worn mb.
With a new mb that 500e fc ran 163x5 or 815 Ghz ram speed limitted.
My first c2d also ran normal again when I tried it in new mb before I gave away.
Motherboards do seem to go down hill fast or maybe the water cooling helps prolong the cpu life.

 

[aigomorla]

i have killed... 7xC2D's + 4xC2Q + 2 i7 Gulftowns from too much voltage.

First u get random blue screens..
Then your system wont post.
Then you need to reset your system.. if ur lucky.. it will post at default.
If not lucky,, it wont post at default,, you need to supply another chip raise voltage @ stock clocks and try to post.
Then if u ever lose your bios settings u need to repeat.

The chip will get worse and worse over time requiring more and more voltage to get what it used to get until it just wont work anymore.

It will behave exactly like an unstable overclock - BSODs, application crashes, reboots, silent data corruption etc. and it will fail stress tests that it used to pass at the same settings.

Increasing the voltage and/or reducing the overclock will fix it. However if the CPU has already started degrading, chances are it will eventually become flaky again at the new settings, forcing you to repeat the process. Eventually it won't even work reliably at stock clocks/voltages.

This is very rare with moderate overclocks, in my experience. It's mostly an issue with extreme overclocking. A friend had it happen to his Duron 600 MHz @ 950 MHz - I believe he took the overclock down by ~100 MHz and used it for another year or two.

^ this guy beat me to it.. yes its like he says.... it doesnt really happen at low clocks or low voltage with excellent cooling.
You can cheat degredation by either lowering the temps on your cpu, or reducing voltage.

Use my statement to fill in the gaps he's missed, but what he says is the start of chip degradation.

 

[2is]

Jesus man! 13 processors?! All of which are considered to be quite robust lol.

I know who not to take OC advise from!

 

[aigomorla]

Jesus man! 13 processors?! All of which are considered to be quite robust lol.

I know who not to take OC advise from!

lol...

i killed all but the gulftown on purpose tho due to NDA and testing out what the extreme limits were on said cpu's.
The gulftowns... mah.. the first was a OOPS... cuz i was too comfortable to i7 bloomfield... the second one was more of a WTF... cuz i wasnt trying to kill it.
Havent killed the 3rd or 4th or even 5th and 6th, cuz i know where to stop.

i was a real NDA holder for a while... so that meant.. no i didnt sell my ES's... and yes they were used in my lab or destroyed when testing was completed.
I even sent back intel a few dead CPU's cuz they wanted to see exactly how i destroyed them... lol... they actually like people who are testers to send back dead cpu's.

You cant tell people what the limits are unless u yourself have been at those limits, and have reproduced enough results to label it not a fluke.

the chips i have yet to kill are:

1. i7 Bloomfield.. OMG freaken TANKS dude.. 1.5vcore 2 yrs... still going hard... this chip i cant seem to kill unless i do something REALLY stupid. :O
(i7 920 owners can contest to how robust this chip truely is... one of intel's greatest cpu's in my book... next to the Yonah processor which saved intel.)

2. c2q kentsfield... i swear this is the big brother of the bloomfield... another near impossible chip to kill.

 

[GLeeM]

My i7 920 went from 4.0 Ghz down to 3.6 Ghz over its 2.5 year lifetime while maintaining the same voltage.

What motherboard did you have? And how much vcore?

1. i7 Bloomfield.. OMG freaken TANKS dude.. 1.5vcore 2 yrs... still going hard... this chip i cant seem to kill unless i do something REALLY stupid. :O (i7 920 owners can contest to how robust this chip truely is... one of intel's greatest cpu's in my book...

I've had my i7 920, in a P6T Deluxe V2, at 187x21=3927 (or higher) @ 1.264vcore for more than four years at 100% load running DC. I would guess average temp for 4 years at 80C!
A year ago I had a blue screen so I lowered my BCLK (bus speed) from 190 to 187 and also lowered the GPU overclock.

especially with high voltages?

What CPU is this?
And how high voltage?
If voltage is not that high ... what motherboard?

 

[2is]

lol...

i killed all but the gulftown on purpose tho due to NDA and testing out what the extreme limits were on said cpu's.
The gulftowns... mah.. the first was a OOPS... cuz i was too comfortable to i7 bloomfield... the second one was more of a WTF... cuz i wasnt trying to kill it.
Havent killed the 3rd or 4th or even 5th and 6th, cuz i know where to stop.

i was a real NDA holder for a while... so that meant.. no i didnt sell my ES's... and yes they were used in my lab or destroyed when testing was completed.
I even sent back intel a few dead CPU's cuz they wanted to see exactly how i destroyed them... lol... they actually like people who are testers to send back dead cpu's.

You cant tell people what the limits are unless u yourself have been at those limits, and have reproduced enough results to label it not a fluke.

the chips i have yet to kill are:

1. i7 Bloomfield.. OMG freaken TANKS dude.. 1.5vcore 2 yrs... still going hard... this chip i cant seem to kill unless i do something REALLY stupid. :O
(i7 920 owners can contest to how robust this chip truely is... one of intel's greatest cpu's in my book... next to the Yonah processor which saved intel.)

2. c2q kentsfield... i swear this is the big brother of the bloomfield... another near impossible chip to kill.

Ahh, well that makes a little more sense. Sounds like a fun job!

 

[bononos]

From my limited experience, the few oc'd cpus didn't die straightaway, but became less stable (bsod/reset) at oc'd settings which were previously stable. I had to dial back the settings and then went back to stock settings because I didn't need the hassle of the possibility of crashes. One earlier oc'd system was running fat32 boot disk which had odd errors after crashing and I had to reinstall. It could've been the mb contributing to the crashes but I don't know for sure.