Hardware Failure - MTBF

[PeteSmith]

I'm curious as to which PC components last the longest, recently i suffered two failures on the same day, my RAID array degraded due to a failed disk on my NAS, and the Graphics Card in my desktop also failed. How unlucky am i? Anyway the HDD is still under warranty so that's going back for RMA. I'm now on my third graphics card in my desktop, no other components have been replaced, which does concern me given they are probably one of the most expensive components. I have had a couple of PSU's fail me in the past, never had a motherboard, CPU or optical drive fail. Couple of memory sticks have been replaced.

I'm trying to find some statistics to see which components have the shortest MTBF, or effective lifespan before going to silicon heaven. If you have any links i would appreciate it. Otherwise just personal experience would be welcomed. Assume factory settings, no over-clocking shenanigans.

Here is my personal experience:

1. Graphics Cards - Replaced 2 in 7 years, same PC, no other components replaced (No, i don't need a new PC ) ~ both died after 3 years
2. HDD's - Only ever had two outright fail but i have a few that are showing severe signs of age, average life ~3 years
3. PSU's - Had 3 fail on me and need replacing in various PC's over the last seven years, average life ~4 years
4. RAM - Maybe two sticks that caused no end of instability issues which i eventually binned , average life ~5 years

[Agent]

For me, graphic cards have been the kicker too. I'd say it was by far the component which has failed on me the most. This includes some £300+ cards that barely lasted a year.
In most cases it's been the GPU / Memory directly and not the supporting components.

Internal HDD's I've been okay with, minus issues with early gen SSD's. External HDD's have caused me a few issues though (2.5" desktop drives), with even ones that have never been moved and purely sat on my desk failing within 2 years.

I've had a few optical drives go on me, but for their cost, even after a few years it's not a concern. I also use them a lot.

I've had a few motherboards fail, but this was quite a long time ago. Recent ones seem to be a lot better made.

Never had a CPU failure.

Memory - I've lost track of how many, but it's normally been faulty on arrival, which is why I run memtest on every new system I build given the time it takes. I just wish you could patch the Windows kernel to avoid certain memory addresses.

[jim]

Every graphics I've had that has started to fail has gone to a new home and made a miraculous recovery, so not sure.

My HDDs have been incredibly reliable, not sure I've ever had a production failure (a few DOAs though). 3 years feels sensible though.

PSUs, I've had two failures IIRC. A budgety one lasted about 4 years, but when it went, it blew up my PC - dead mobo, dead CPU, dead RAM. My HX620 died after 2 years, but it went quietly and didn't drag anything down with it.

[MrJim]

In 25 years or so of using & building PCs, I've had one hard-disk failure (the good ol' IBM 'Deathstar'), one PSU (Enermax something or other), one motherboard failure (probably due to a cheap Chinese PSU...lesson learned there), and one pair of memory modules (DDR Mushkin 'Redline' series - 2.5v!). Not a single graphics card failure yet...touch wood!

[stevie lee]

current computer is 5 years old, only replaced graphics due to winning new one.
the radeon 4870 is still going strong in another computer.

old Athlon 2500+ computer was around 7 years old, used every day for 5-6 hours or so. graphics is what broke first (crysis make it melty) then the northbridge fried, bios corrupted, CRT monitor started making loud popping sounds. needed a new pair of pants after that

old old PIII 450 computer, that was 5 years old I think. Graphics card broke on that. ATI 9500, morrowind with a high texture mod I think, took ages to download the 25mb patch on dial up thinking back on it, 1 frame per 5 minutes should've alerted me that something was up.


so for me its been mainly been graphics cards failing (due to me running stuff they weren't designed to ) after 5 years.


and I just looked at the SMART for my C drive. power on time is 621 days 16 hours. so nearly 15000 hours from the hitachi 7k1000b drives
id199 UltraDMA CRC Error Count is 15 and that was from when I interrupted a defrag by pressing the reset button.


only my current CPU has been overclocked for about a week from 2.8 to 3.2ghz, every other component i've ever had hasn't been overclocked. so maybe that's the key to all this, just run it at spec. and not any faster and it'll last forever?

[DanceswithUnix]

only my current CPU has been overclocked for about a week from 2.8 to 3.2ghz, every other component i've ever had hasn't been overclocked. so maybe that's the key to all this, just run it at spec. and not any faster and it'll last forever?

Sort of. Component death is often heat related. Either stuff cooks due to not enough cooling, or the cooling isn't clever enough so the component cycles a lot from hot to cold. The thermal expansion and contraction leads to fatigue and failure.

Occasionally, like the IBM DeskStar, they mess the design up in some subtle way.

[kalniel]

For me it depends if you're talking about complete failure, or partial.

I was about to jump in an say 'case!' - but actually I've know of several cases which have had front panel connector issues soon after purchase.
Likewise motherboard - I've not had an outright failure, but I've had a couple where the Ethernet port or USB connector goes.
Hard drives I've had several failures over the years, but recently they've been quite good *touch wood*.
Graphics cards, again failures after quite a bit of use from one of the manufactures, the other has tended to be obsoleted before failure.
I think I have had a CPU failure, but only way back in the day, and it was bringing an old CPU out of retirement so not sure it counts - it had been in a dodgy m/bo at one point.
PSU.. I've not had a power failure, but I've had fans degrade to the point they're unbearably noisy.
Ram.. I've had errors in very old out of retirement sticks, rare though.

That leaves only a few bits and bobs really - ethernet/wifi cards are one - I don't know how, perhaps were over-engineered for business use. And SSDs - it's probably still too early to say as I've only had one.

[Marcvs]

For me:

1 Mobo Failure - Asus P5K DLX Was working when i put it away - tried it a couple years later = dead
6 HDD Failures - Samsung Spinpoints & Seagate Barracudas mostly (though 90% of the HDDs I buy are these so not saying much)
2 Ram Failures - Crucial DDR2 and OCZ DDR2
0 CPU Failures
0 PSU Failures
0 GPU Failures

[Pob255]

I've had a psu (cheap) go and take out a hard drive, cpu, ram & sound card but that was a long time back (1999ish)

Last major failure was cpu&motherboard, not sure if the cpu took the motherboard with it or the motherboard took the cpu with it but they both died. 1st gen i5 750
Although my mouse recently died.

I have had some graphics cards go on me, still remember my old ati 9700pro that cooked it's own vram

At work the most common faults atm are PSU by a long way followed by motherboard then hard drive, optical drive and mice are regular too but that's down to mindless user abuse (eg sicking chewing gum into optical drives or pulling the buttons off the mice)
However 6 years ago hard drives where a far more common fault equalling psu's, but I think that was more when we were having some very hot summers.

So I'd say environment does play a big part as to what fails when and the pace of tech change makes the whole question very variable.

[PeteSmith]

The votes are in, i did a quick tally based on the Hexus forum readers response and here are the official results:

Starting with the most unreliable:

HDD (internal) - 11
Motherboard - 7.5 (Because Pob wasn't sure which took the other out)
GPU and PSU - 7
RAM - 6
Optical Drives and Mice - 2
Monitor (CRT), NIC (Wired/Wireless), Case, External HDD's - 1
CPU - 0.5 (Because Pob wasn't sure which took the other out)

I think the above sample size is fairly reflective of the real world hardware failure rates, not too sure i agree with the external HDD's, i'm using two of those as bookends but i haven't counted those as i didn't mention it to start. I'm still none the wiser about the MTBF.

[Bonebreaker777]

Hardware failures hmm.

10+ HDD failures (old Samsund PATA, Maxtor 40GB killed 4x from the same batch or so, some DoA).

MoBo 2x killed, one was some ABIT replacement I got after I killed some{ ASUS Deluxe socket 939 MoBo just before it (high chance of humidity corrosion, after two weeks being away even my mirror finished case was rusty)

Some memory modules not working correctly (I think they were some 500Mhz DDR)

Enermax PSU (my old 350w was working fine but my new shiny 550w couldn't carry my MSI GPU)

Graphics cards killed some when added DIY cooling over the naked chip and chipped it

No CPUs killed or optical drives

[CAT-THE-FIFTH]

HDDs are the worse for me - I don't know whether it is down to the fact I use SFF PCs.

I had a few sticks of DDR2 go on me and one stick of DDR3.

Two Logitech G5 mice went down to a design fault,but Logitech replaced them.

Replaced two PSUs - they had not gone but were getting quite old.

One graphics card I had did go kaput.

[Pob255]

The votes are in, i did a quick tally based on the Hexus forum readers response and here are the official results:

Starting with the most unreliable:

HDD (internal) - 11
Motherboard - 7.5 (Because Pob wasn't sure which took the other out)
GPU and PSU - 7
RAM - 6
Optical Drives and Mice - 2
Monitor (CRT), NIC (Wired/Wireless), Case, External HDD's - 1
CPU - 0.5 (Because Pob wasn't sure which took the other out)

I think the above sample size is fairly reflective of the real world hardware failure rates, not too sure i agree with the external HDD's, i'm using two of those as bookends but i haven't counted those as i didn't mention it to start. I'm still none the wiser about the MTBF.

Personally I don't think you can generalise that much, there's far too many variables involved.

I'm pretty sure out in the "real world" psu failure is higher because there's a lot more low end low quality psu's being used out there then there are by hexus members.

It's like saying the Malaria is the number one killer, (I'm pretty sure it is globally) but it's not very high up on the list in the UK.

[kalniel]

Personally I don't think you can generalise that much, there's far too many variables involved.

I'm pretty sure out in the "real world" psu failure is higher because there's a lot more low end low quality psu's being used out there then there are by hexus members.

It's like saying the Malaria is the number one killer, (I'm pretty sure it is globally) but it's not very high up on the list in the UK.

There's no reason to believe he is generalising. As the OP is an Hexus member, things that apply to the Hexus member experience are probably more relevant than world-wide. Another example might be graphics cards - they could be terribly unreliable, however a hexite might replacement them due to obsolescence rather than hardware failure, making reliability less important.

[shaithis]

I've seen it all fail (with the exception of CPUs....don't ever remember seeing a dead CPU that hadn't been broken by someone mishandling it)....but that has been over 25 years or so!

HDDs are the worst by far AFAIC.....although not that surprising as they are mechanical.
Next on my personal list would be motherboards.......one of the first things that go due to the stress of heating/cooling and also power spikes (poor leccy or failing PSU)

[OneIT]

Personally I think I've been relatively lucky with my hardware, but I tend to buy high end stuff so chances are more cares going into quality control.

Since I started building PC's back in 2005 or so I have had:

1 Failed GPU - I let my CPU fan take a chink out of my finger and jolted the whole PC when it was on, couldn't render 3d after that until I replaced the GPU.
3 Failed hard disks - 2 of them old first gen SATA Maxtor drives and 1 Hitachi 1TB
2 Failed GPUs - I was using phase change cooling at an iSeries LAN and with the ambient temperature all the ice around the CPU socket melted and ran down the motherboard and killed both of my GPUs.
2 Failed OCZ RevoDrive SSDs
1 Failed CPU

I work in computer repair / support and in our workshop we tend to see a LOT of failed hard disks. Even in systems only a couple of years old. Other than that most things are pretty reliable.