22

I heard that if you leave a computer on 24/7 at 100% utilization, the CPU is expected to last around 3-4 years.

Is there any truth to this claim?

Udi Meiri
  • 321
  • 1
  • 2
  • 5
  • 3
    What often happens after about that time is the thermal goop connecting the CPU to its heatsink degrades. This manifests as crashing after being on for a while, but it can be easily fixed. – pjc50 Feb 23 '12 at 15:52

10 Answers10

35

Usage of a CPU does result in wear at the atomic/electronic level.

The actual lifespan of the silicon transistors of a consumer CPU is typically in the range of 20-30 years before there is a failure, not 3-4years. It is asssumed by then that the item would be obsolete.

How does Intel/AMD know if there's no way to test a chip for 30years? It is tested under load under higher than normal conditions (heat voltage, clock) and the variables and failure data are then extrapolated and calculated backwards for typical use.

Of course there are other failure points such as the chip packaging wires and things of that nature, but low decades would be a fair assumption.

3-4 years is more of the practical obsolescence of the product due to Moore's law and all that.

Source: Masters degree in EE, and learned in a course on chip failure design.

raytseng
  • 351
  • 3
  • 2
  • in business environment most hardware is usually being in use for 5 - 10 (or even more )years before it's decommissioned, so the actual "obsolescence" takes slightly longer. I am talking about servers here, not workstations. – Petr Apr 18 '15 at 21:53
10

Your powersupply, CPU fan and hard disk will all die long before your CPU. They're the three parts likely to fail. CPUs aren't that well known for failing.

Also, it's a very very very rare computer that is used 100% 24x7.

I have recently decommissioned a server that was nine years old and was perfectly usable, if slow by today's standards.

David Pashley
  • 23,151
  • 2
  • 41
  • 71
  • 1
    We recently turned of an old P2 400mhz machine recently. It had been running fairly heavy stuff for about two years before it was replaced. Many people run things like Seti and Folding@Home for years without any issues. – Ryaner Sep 13 '09 at 21:23
  • You can take the CPU fan out of the equation by using passive CPU cooling (e.g., Scythe Orochi). – Dmitri Nesteruk Oct 26 '14 at 08:06
  • 2
    I don't think it's rare to find a 24/7 box on "serverfault" :P – Petr Apr 18 '15 at 21:54
  • 3
    A computer that's on 24/7, no. A computer that's on and using it's CPU 100%, yes. – David Pashley Jul 13 '15 at 06:32
  • 2
    A lot of people run BOINC or Folding@Home, so 100% CPU 24/7 is not rare at all. – Atario Jun 14 '16 at 01:14
9

Unless it overheats, and modern CPUs have thermal cut-offs that will try not allow that (by simply shutting off or by throttling their performance when the heat hits certain limits), a CPU is unlikely to die prematurely.

Of course if you have mucked about with the clock and voltages you might harm the CPU in ways that will give it an early death, but that is a known risk you take in such cases.

If running a CPU at full load 24/7, its fan and other cooling arrangements are going to be under as much stress as they will ever be so this could have an impact. If the fan fails then the CPU may overheat so fast that any thermal cut-off might have time to trip before it is too late as the temperature rockets up.

David Spillett
  • 22,534
  • 42
  • 66
6

Typically the CPU is one of the most reliable parts of the PC. You stand a better chance of a failure of the CPU's fan, the system powersupply, the hard drive, or the motherboard before the CPU fails.

That being said, I recently had a string of CPU "failures" a few months back. We were running Intel Celerons in a production environment for a period of several years. I'm going to guess around 3-4 years.

I say "failures" in quotes because none of them ever actually DIED to the point where they system wouldn't boot.

One started doing "bad math", one started running REALLY HOT (so we took it out before it died), and one would boot up perfectly normal except that the screen display would eventually be whacked out and display random characters, etc. That one was the most interesting one as I've never seen anything quite like it in all my career. At first I blamed the mobo and we replaced it with an identical one. It wasn't the mobo... it was definitely the CPU as we put that same CPU in three identical mobos and got the same problem in all three. I have no explanation for it.

So, while its unlikely to experience CPU failure... it CAN and DOES happen. Just remember, in most cases... (at least in my experience) its usually one of the LAST things to go.

I have seen countless old systems still running perfectly fine after years and years and years.

KPWINC
  • 11,274
  • 3
  • 36
  • 44
4

The few times I've seen processors that had a rated life, it was 100,000 hours. This works out to 11.4 years of continuous operation. I've noticed this same number appearing on lots of other components that one would not expect to wear out (such as LEDs), so I think it's just the number they put when you expect something to last, for practical purposes, forever.

I've never known a previously working CPU to fail during normal operation.

David Schwartz
  • 31,215
  • 2
  • 53
  • 82
4

Apart from overheating or physical damage, I've never known a CPU die. Take good care of it, and it should last until well after the machine is redundant.

Dentrasi
  • 3,672
  • 23
  • 19
2

To answer your question: The "useful lifespan" of a CPU is about 3-4 years. My guess on the "technical lifespan under full load" is closer to 8-10 years if nothing else breaks and takes the CPU with it.

The reasons:

High Performance Computing Clusters usually keep the computers for 3-4 years. They remove the old computers because they are not competitive any more - or not reliable enough. And I suspect someone said "CPU" last 3-4 years meaning "Computer".

At work, we have one special purpose computing cluster that is 7-8 years old. It´s starting to get problematic for mechanical failures.

The CPU itself very rarely breaks. Power Supply, Hard Disk, Fans, Mainboard, Memory, Graphics Cards are the parts that break first usually.

In a computing cluster, you want to get as much horsepower as your restraints allow. The restraints are space, cooling, and running costs. After 4 years, computers dont pull their weight any more. Today: 16 cores and 64 GB RAM on 1 HE and 400ish Watts vs 4 cores and 16 GB RAM back then. Out with the old! Costs four times as much per instruction cycle!

Posipiet
  • 644
  • 5
  • 4
1

It will last until you take it out of the motherboard and touch it without being grounded. Static Shock!

BLAKE
  • 706
  • 9
  • 25
1

A candle that burns twice as bright burns for half as long; and you over clockers burn your CPU's so very, very brightly.

Fred
  • 11
  • 1
0

I am not an expert on CPUbut after lots of research on web about this area i believe that it is the most unlikely to encounter a CPU failure if you handle your computer and CPU installation properly. Keep an eye on the temperature of CPU and the CPU fan. It is said that if CPU fan fails it could lead to permanent damage to the processor

Kenny Siu

user170409
  • 1