The cause of shortened CPU life

[clicknext]

What exactly causes short CPU life when overclocking? Is it the raised voltage, the higher clock speed, or the heat created from those two? I want to know which one of those specifically is the reason.

 

[ReiAyanami]

both

but for pentium 4's its the voltage that does it
for AMD's its more of a heat problem

watch out for migrating electrons

 

[clicknext]

I have a AthlonXP. If I raise the voltage from 1.75 to 1.9 and put a better heatsink on, should it be fine? Right now I'm using the retail HSF, and when at 1.9v and overclocked to 1.75ghz, it goes up to 58-59 after an hour of 100% load.

 

[CrazySaint]

Originally posted by: clicknext
I have a AthlonXP. If I raise the voltage from 1.75 to 1.9 and put a better heatsink on, should it be fine? Right now I'm using the retail HSF, and when at 1.9v and overclocked to 1.75ghz, it goes up to 58-59 after an hour of 100% load.

The stock HSFs that come with Athlons are fine for stock settings, but I would definitely get a new HSF for overclocking.

 

[touchmyichi]

Most cpus have 10 years of life on them at default. Its not like your going to be using it for 10 years lol. So you can drop a few years off. I mean the most u ever use it for is 2-3 rite (and besides this means keeping it on for 10 years straight lol). So don't worry. OC happy.

 

[clicknext]

lol, actually, I've been using my oldest computer for almost 8 years now. Well, not me, my mom, because she just types stuff. Old shitty 286, lol.

 

[EKAtBzboyz]

Originally posted by: clicknext
I have a AthlonXP. If I raise the voltage from 1.75 to 1.9 and put a better heatsink on, should it be fine? Right now I'm using the retail HSF, and when at 1.9v and overclocked to 1.75ghz, it goes up to 58-59 after an hour of 100% load.

its based on .18micron fab, same as tbirds
running tbirds at 1.85 was pretty much normal for any overclocker, most went far above that (2.1v even)
im pretty sure it would still cause the same wear and tear as it caused on tbirds, so it will be fine for the most part

cpu life will be shortened, but even raising it .05 above would theoretically shorten it

 

[TerryMathews]

Originally posted by: EKAtBzboyz
cpu life will be shortened, but even raising it .05 above would theoretically shorten it

So then by your logic some AMD chips come from the factory set with a longer life than others? I can't imagine that being true. Would think that AMD/Intel would have to disclose that your XP2800 or P4 3.0 might not run as long as a newer XP1700 or P4 2.26...

 

[clicknext]

So basically, CPU life is shortened only if raised voltage causes higher temperature. If I have good cooling, then my processor won't explode with a chunk hitting me between the eyes.

 

[EKAtBzboyz]

Originally posted by: TerryMathews

Originally posted by: EKAtBzboyz
cpu life will be shortened, but even raising it .05 above would theoretically shorten it

So then by your logic some AMD chips come from the factory set with a longer life than others? I can't imagine that being true. Would think that AMD/Intel would have to disclose that your XP2800 or P4 3.0 might not run as long as a newer XP1700 or P4 2.26...

Probably every chip has a life different than others. I was saying that if the voltage is increased by any amount other than the "recommended" voltage for that specific fab process and what its been tested at, it will have an affect on the life of the cpu, whether it be great or small.

I dont mean that because both tbreds and northwoods are .13 micron cpu's, they should use the same amount of voltage, and that since tbred uses more voltage it will have a shorter lifespan. Even the C1 steppings are using 0.025v more than the B0 steppings for northwoods. It all depends how teh cpu was made, what it was made to run at(voltage wise).

clicknext: Not true necessarily, especially for the northwoods since we've has "guinea pigs" already do the testing. Ive seen many with really good cooling (w/c or w/c+TEC) and ye who ventures past 1.7-1.75v, more often than not, comes back with an expensive keychain lol

 

[CaptnKirk]

It comes down to one thing - Thermal stress, which is HEAT.

If you raise the voltage, the temperature is elevated. Put on a BIG heatsink, you can get rid of more of the heat, but the core temperature has still been raised,
even if you continue to cool it. You can expect the projected life span of the CPU to drop by 25%, which would be about 2 1/2 - 3 years.

In 7 years it will be obsolete anyway, and if you go for 5 years it will be a pretty slow computer compared to the new stuff that will be out by then.
How long ago was the P-2 the 'HOT TICKET' ?

 

[clicknext]

lol, lots of my friends still use P2's. My girlfriend uses a Pentium Pro and tries to play cs, lol.

Anyway, so voltage affects CPU life even if cooling is good, but some can be raised higher than others, like AthlonXPs can be raised higher than P4.

 

[BostonTeaParty]

There are three main wearout mechanisms for CMOS integrated circuits: electromigration, hot carrier degradation, and oxide breakdown. Each of the three has a different dependence on temperature, voltage, and clock frequency. The sensitivity of an integrated circuit (IC) to each is dependent more than anything on the particular design and fabrication process of the IC. Without access to Intel and AMD's reliability testing data, I couldn't tell you much about the relative importance of the three aforementioned failure mechanisms for their CPU's. However, I can tell you how temperature, voltage, and frequency changes affect each of the mechanisms.

The lifetime of an IC is shortened by electromigration in proportion to the square of the voltage and exponentially with increasing temperature. It is independent of clock frequency.

The lifetime of an IC is shortened by oxide breakdown exponentially with increasing voltage and exponentially with increasing temperature. It is independent of clock frequency.

The lifetime of an IC is shortened by hot carrier degradation in proportion to frequency and exponentially with increasing voltage. The temperature effect on hot electron degradation varies depending on transistor geometries and voltages--let's basically say it's independent of temperature.