Why is the GPU almost always the first component to fail in a PC?

>tfw I'm still using my HX1000W I bought in march 2008.

Attached: Bild020.jpg (1536x2048, 657K)

heat and dust are the #1 PC killers.

most people barely clear out the dust clogging the gpu heatsink and combined with the gpu running hot during games or mining no wonder it fails.

I've never had a GPU fail, but I have had a few hard drives fail. Not catastrophically though.

*May

Read the date as MM/DD/YYYY (actual date was 03/05/2008) for some stupid(american) reason.

Oh and i;ve had a few PSUs fail also, RAM, and a mobo. But never a PSU or CPU

I've never had a single component fail on me and I feel like I just jinxed my luck after opening Jow Forums today and reading this thread.

I've had 1 CPU fail, 1 GPU fail, maybe 2 DIMMs, and like 4 or 5 HDDs.

Anecdotally HDDs fail most often.

That's not true. The processor is always "firing electrons" (cycling) when turned on. Operating systems with proper power-saving modes under-clock the CPU so that it doesn't use as much power and stays cool, which increases its lifespan. Install a Linux distribution without power-saving features pre-installed and you will see that your CPU is simply cycling at the clock rate set in the UEFI/BIOS

my GTX670 died on me after only three years, one day it just completely stopped

nope, a clock signal is literally all the time hitting your shit at the clock rate

My PSU failed a year in. Nothing else has failed yet and I'm 4 years strong. My HDD is on it's way out next though it seems.

same
7 year warranty PSU fails after a year

Those high end Corsair were good.
My 860AXi was still running, although not as efficiency as originally and a little loud, so I wound up replacing it.

I had a Geforce card, I think it was a 8800GT, die within a month of its warranty expiring.
Got a 9600GTX which was a lemon and I had to replace within like a year. Didn't bother RMAing it since I didn't want an Nvidia card any more, and the 5770 I replaced it with was such a major upgrade.

And I know Nvidia cards don't necessarily die more often (though AMD does overbuild their reference PCB much better), and I just had bad luck. But that was still enough to get me to switch and after my 5770 and then 7970 I never had to buy another GPU for so long to even think about considering Nvidia again.

I've never had an AMD card die. 9700 Pro, 5770, 7970 are all running to this day.
I know a lot of that is luck, though I suspect a lot of people who have AMD cards die are running on AIB models which have a WORSE PCB than reference. Gigabyte does that fucking shit a lot.

Usually components fail within the first year, or they'll last 7+ years fine.

Actually I had 2 7970. Fan failed on one, but that was it. Otherwise works perfectly fine still up until I sold the 2nd one.

I don't know about OP but for me it was always PSU or HDD read heads that failed on me

Both great. Better than anus and niggerbyte

lol the classic make a statement that is erraneus in order to get correct answer instead of just asking the question normally. boomer.

9400 gt still working here

Stop buying shitty cheap GPUs.

>Those high end Corsair were good.
The HX1000W is basically two 650-750W dumbed down to 500W per side as multiple sources has overloaded it beyond 1200W and it still works well.

hardwaresecrets.com/corsair-hx1000w-power-supply-review/8/
hardocp.com/article/2008/04/21/corsair_hx1000w_psu_review/4
hardwarecanucks.com/forum/hardware-canucks-reviews/6951-corsair-hx1000w-power-supply-review-11.html

>My 860AXi was still running, although not as efficiency as originally and a little loud, so I wound up replacing it.
My fan has become slightly noisy since it's a brush motor in it and it's been going almost nonstop since 2008, so I'll have to replace my Yate Loon into something else, maybe a Noctua 140MM or something.

literally never had a PSU fail on me and am currently using one from 2007.

I think my local computer shop isn't shit and even their 60$ psu's are decent.

Hard drive and psu nigga

Ya like I said, just bad luck.
Nvidia just makes the dies. Graphics card failures are almost always the PCB and not the GPU. But it still drove me to get AMD at the time.
Well and 5000 series benchmarks were just amazing.

actually my 9600GTX being a lemon was likely the die. It was unstable as fuck all the time doing anything. idk. Could have been the board, still.

I've dealt with two failed hard drives over the past month.

Because it's the part which heats up and cools down very often, is fragile AND is soldered instead of being socket-ed like the CPU - which means lots of stress for those solder joints

I’ve had one gpu fail but it was due to bad drivers.

this is only true if you cryptomine or some shit

How underage are people ITT to make claims like "never had x fail".
I've had 3-4 PSUs fail, a graphics card, like 5 mobos, many RAM sticks, half a dozen HDDs, an SSD, a wifi card, monitors, sound cards, dozens of fans. Not sure if i've ever had a CPU go bad, i think it's usually been the mobo, but never tested thoroughly to discern which since it's always old/obsolete at that point.

I've only had 1 GPU fail on me and it long in the coming (system would randomly freeze, etc for months and months) until one day it finally died.

Probably was faulty from the very start though because it was only like 1 yr old.

do you live in the bermuda triangle or something

The only time a GPU failed on me was the fan wore out from dust

Nope, just been using PCs for 25 years, electronics fail sometimes. Not everything does, eg my pentium 75 never failed, i threw it away when it was 10 years old and was so slow i had no more uses for it.

because you buy intel

Don't buy single fan coolers if it's a hot GPU. Don't buy single slot cards period. The card's cooler relies on a good supply of air from the case, so make sure case airflow is good. Get the best power supply you can. If your PSU is feeding bullshit through the 12V rails, then your GPU is going to suffer.

Heat. Heat kills components. Not to mention the range of temps can sometimes be larger on a GPU vs a CPU for example. GPUs pump out a lot of heat given the die size. A graphics card also has its own power stages and vram, all which have to be cooled, often on the same cooler as the main die.

2nd:
GPU's are binned much differently than CPU's. A lot of AIB partners that do factory overclocking have lower percentage stability across the board. Balancing heat output, power delivery and clocks means more room for trouble in the field. This is reflected their returns policy which is often more open and proactive. Also, because AIB's are competing, they are fighting for the best clocks which can mean pushing slightly lesser binned GPUs to their 'natural limit' for market share but maintain customer loyalty through efficient RMA. This is less the case these days as lots of AIB's have nailed a good formula that works for them. But its something to consider and as architectures change- new challenges are provided.

Reference cards are approached in a Conservative manner for reasons of reliability. Reference GPUs are clocked well under the 'natural limit' to maintain low returns and long life. This is why you can achieve high clocks on an average bin reference GPU (more so with NVIDIA, AMD tend to push their base clocks a little higher). They are not really expected to be overclocked and as such, they can use average binned dies and still be within the target QA.

Attached: 20180610_153911.jpg (1080x608, 449K)

My stats for comparison:
1 gpu failed

My R9 270x failed last year. Was playing FFXIV when suddenly the screen got a bunch of artifacts. After rebooting, the GPU wouldn't display properly unless I was in safe mode.

The only other PC component that failed on me before was a Rosewil PSU which exploded during a Dota 2 match because I was pushing overclocks too much.

have had one GPU die on me (form a pre built)
and 2 hdd and a ssd

Seems to be the motherboard for me

For me it's PSUs, luckly always PSUs, but unluckly they die quite a bit often at a point i can do a metal gear out of my dead PSUs.

I have been using computers for over 18 years and the only thing that failed besides Input/Output devices were 2 GPUs.

>bought corsair psu
>it won't sleep or hibernate, just cuts power if you try
>for years I'm convinced it's a problem with my motherboard
>buy a then siny new LGA1150 board
>still hard crashed if you try to sleep or hibernate
Fuck corsair. Never again.

>saying it is makes it so
I am laffing Jow Forums

I have never had a GPU fail. Saw plenty of motherboards fail.

I've had an AX760i for over five years now, the shit fan is the only issue...well that and the awful software Corsair provides for controlling/monitoring it. I won't ever buy a piece of Corsair hardware again if it relies on software they provide--both the Corsair Utility Engine (Headsets/etc) and CorsairLink (PSU/Closed Loop Coolers/etc) are god awful.

Only if you buy novideo

I got two GTX 970s and decided to go with the nVidia reference design since I figured the same as you, and found one had a really marginal ASIC and is limiting my overall overclock because it can't scale as high as the other 970. Watching afterburner while playing a game you can see its voltage always tracks higher. Only managed to get another 225mhz on the core because of it.

It's a shame because I was right to assume the stock nVidia reference blower design would be overpowered for the 970--I'm nowhere near the card's maximum temperature and could get much more out of them if that second card could keep up.

it's the high power consumption and temps, that makes the VRMs and caps fail. Often it desolders or develops cracks in the bumps and stuff connecting the GPU to PCB, too.
Dedicated GPUs are crap stuff. Unwelcome in desktop, absolutely harmful in laptops.

I’ve been running 2 1080 TIs and I’m getting audio stutters and hang freezes that I can hear through my headphones. Turns out my driver is crashing. I’ve always ran both GPUs under 75c and now they are failing on me. What are signs of psu failure just to rule out that it isn’t that. I have been mining with them for sometime ether but like I said nothing rough on temps. I even made sure to get the best 1080Ti (FTW3). Now they can’t hold peak 2000 MHz

Attached: 13C31FE2-6162-4CD3-B1BF-39DBD03DADA1.jpg (600x411, 70K)

Mine failed last year. I've had a ton GPUs fail over the years for me both nvidia and AMD other stuff has been fine other than the occassional PSU and HDD. My current evga power supply has been doing good for me for about 4 years so far. Few years ago, friend of mine Rosewill break less than 2 years and the replacement of it broke within a week.

I have HD 5770 bought in 2010, running the whole time on 96°C under load and it just won't die. Planing to build a new PC this summer, hopefully they will release gtx 1180 (1170) in next two months before BfA is out.

I'm still using a Shaw trifan 860w PSU I got in 2013 and it's still powering my PC perfectly fine. In fact I've had nothing fail on me yet.
It's powering my i5 3470 and HD7950 just fine.
In fact, my PC still performs fine.

Have desktops really stagnated so badly or is it just that the software is struggling to catch up to what the hardware can accomplish?

How comes laptops consistently run easily over 50c (more realistically near 65c) all time and last long if properly taken care of?

I dont think its the die themselves that go bad.

nice name user

>anecdotal
Never had problems with my fucking cx430m
You probably received a faulty unit

topkek my 8 years old Lenovo laptop's i7 runs at 60°C idle and 102 under load and is still going strong

I've got a 6670 in one system that I keep expecting to fail. Thing just keeps trucking along. My newer R9 280x had a fan die, but the 6670 has been perfect since 2011. Go figure.

Lol my 1080 has started artifacting reeeee

It hasn't been so bad for me.
>GT 640 died after 1 year only bought it for multi display on family PC
>R9 390 RMA'd and fixed
>Seagate drive 1TB died after 2 years had nothing important on it besides steam games.

Meanwhile my 500GB WD Blue from 2010 is still running to this day lol.

I think I've never had a gpu along with a cpu or ssd die on me. Plenty of psus, hard drives, motherboards and buckets of doa ram.

>Heat
really makes me question why GPU cooling is so fucking bad for all manufacturers. You can literally buy some 40€ ayylmao design and shave 20°C off

It's usually not. The reference cooler is bad compared to aftermarket but it's designed with different intentions.

4 and a half years later and still going strong.

Attached: ss (2018-06-13 at 06.25.35).png (718x562, 40K)

I think we should already switch to superior ss:DD/hh:YY/mm:MM

The only components I've had fail on me since the days of Core 2 Duo til now was a couple of power supplies and a motherboard.

this is my experience
feelsbadman

Never had a GPU fail on me ever. Have had HDDs fail, but expected that since they are mechanical in nature and would work well for five years. Also had a SSD fail within a year.

Probably the worst tech failure I ever had was the AC stopped working while I was out on vacation, and during that time a stick of memory just stopped working altogether in the humidity.

I bought mine from ebay new just under a year ago nd it's already dying
Where the fuck are the new gpus?

Have a theory that lots of GPUs die from card sag which makes bga contact points to fail/break because of torsional twist on pcb.

Attached: gpu_sag_comp.jpg (1024x535, 215K)

I reduced the sag using string on my gtx 760, attached through vent holes at top of case and cooling fan frame on card. Just be sure it is not in path of fan blade before power on :^)

>ca.pcpartpicker.com/list/GCrt3b
On the topic of PSU, could anyone tell me if this one is good to last me with this build? I don't know much about psu.

This is what I currently have, and I'd prefer to keep it if I don't need to buy another. But I don't want to bottleneck anything.

MSI master race reporting in

Using computers since 1997., never had a gpu fail.
My list:
1. hard disk
2. hard disk
3. hard disk

It's a combo of people putting high temp cards in shitty cases with terrible cooling and people letting ancient power supplies crap on their system when they start to get too old.

YYYYMMDDTHH:MMZ(+/-)hh:mm
where the lowercase hh:mm specifies the timezone offset, T and Z are delimiters.

what's that cockring on the right of the case for?

They run the hottest.

im getting artifacting (blue horizontal lines) and cant even install drivers properly anymore (error 43)
it's dead, isn't it?

It think it's the case if your PSU is too weak.

My seasonic failed after 4 1/2 years, thanks god their warranty lasts 5 years and got a brand new one.

Mine died on me at the start of the year. Brought it in 2010.

>Usually components fail within the first year, or they'll last 7+ years fine.

That's so true.

Maybe you need something like this.

Attached: VGA holder.jpg (4608x3456, 3.31M)

Do you live near an Ocean or a dump with acidic wastes ?

This. Also the VRMs and capacitors fail first. The GPUs almost never die. Same on mobos, thats why its so hard to get good oldschool boards, while you get old highend CPUs dirt cheap.

>he doesn't use Corsair™ power supplies

Attached: 1528479528068.jpg (320x454, 21K)

Buddy please. These fuckers just love to die. And don’t give me shit about how your SSD will work eternally. Just wait until one dies on you for the first time. You get zero warning, it’s pretty tragic.
I have actually never had any component die other than storage media.

Attached: DC77480F-4732-450A-888B-83CEF12B9A60.jpg (1250x700, 74K)

Motherboards get fucked all the time, I had numerous Athlon 64 motherboards shit themselves along with a Z77 board.

pretty sure the hole "PSU fail alot" thing are just retards in asia and middle America who buy PSU as a afterthought for 50$ and the shop they buy them from gets literally the worst shit from china.

or maybe its people just living in shitty areas with bad power etc.

only GPU that have died for me ive killed with arctic silver or have died because they faulty 8xxx 9xxxx nvida etc. (was huge class action law suits against them for this)

I did lend a GPU to a friend that started artifacting after maybe 2 years of SLI but wouldn't be suprized if he overclocked them both and just one failed. was 6970 amd.

you realise you can just screw one screw into the back of your case where the gpu connectors are to make them strait. jay two cents has a video about it called GPU sag or flop or some thing.