Share your GPU horror stories, /g/.
>Mostly use consoles for gaming
>Build first gaming-oriented PC in 2008
>Galaxy GeForce 9500 GT
>Runs okay, but lacking in performance
>Upgrade to Sapphire Radeon HD 5850 the next year
>Upgrade to PowerColor Radeon HD 7850 two years later
>Side-grade to an MSi GeForce GTX 660 for better OpenGL performance
>RMA twice for defective VRAM that BSODs after memory fills to a certain point
>They send it back with the same problem both times
>Sell it out of frustration
>Buy EVGA GeForce GTX 760
>Runs fantastically for a year and a half
>Spontaneously dies with no warning three weeks ago
>No HDMI or DVI output whatsoever
>RMA and buy an EVGA GeForce GTX 970 as an upgrade
>Screen flicker, artifacts, and the display driver consistently hangs my system when playing Elite: Dangerous
>Customer support says to RMA the unit
I swear to fucking Christ, if the replacement 970 has any issues at all, I'm going back to AMD and getting an R9 390. At least my reference HD 5850 still works.
Buy Gainward GTX 980 TI in October 2015, dies on 18/12/2015 (corrupted resolution (800x600), blue video artefacts sprinkled everywhere, BSODs) without any apparent reason just in time for the start of my vacation. Downgrade to old GT 9500 while I wait for my vendor to replace that card.
>gtx 770 in late 2013, dat performance.
>dat master race feel, dat 60fps ultra settings beauty
>get 2nd 770 a year later for under 200... Dat fucking performance
>card dies right before mgs V releases
>can't go back to lesser performance so end up getting another gtx 770 for 150.
>big Boss would've approved.
>2nd card dies during black Friday weekend
I've never been more pissed off considering everyone else I built a pc for never had any issues to this date. It just had to be the psu so I RMA'd last week. 2 pins on the 24 pin plug were melted and the connector on the board had a burned look as well. I thought seasonic was reliable? My motherboard was damaged and 2 GPUs died on me. I'm down to my last 770 and I'm holding onto it till pascal gets released. Ironically, both cards that died were the MSI gaming model.... Gigabyte card is still running. Wish seasonic would fucking compensate me for their product damaging my components. I sent out the PSU last week for RMA and since they have no number to call, I had to email some robot sounding customer service. I'm done with msi, AMD and seasonic.
What do you guys think I should do?
>build PC in 2012 or so, never really had much money
>i5 2500k with 670
>absolutely no problems since, recently bought a second 670 for $100 to hold me over until Pascal
Thank god I've never had any serious issues.
I think xfx uses seasonic maybe? I'm just gonna use this psu for now and sell it, I have a corsair one laying around as well. With the money from this expensive PSU I'll probably get one from evga.
Huh, I've never had a single issue with the 970.
Why did you buy it when the 390 exists now though? Do you not consult other people before making uneducated purchase decisions?
It was a common issue with the game on release as is my understanding from several hundred forums posts claiming there are artifacts with their 680's.
Mostly it was on Underpass, for some reason that map just didn't like being played. Which is ok. It was a shit map anyway.
probably the most known issue since the dawn of modern gpus:
>receive free far cry 3
>want to play mainly cs:go and fifa
>runs great, it plays some AAA titles on high settings
>after two years I notice that sometimes the monitor turns off and only a restart fixes the problem
>gpu finally dies just two weeks after warranty expires
>mfw it's all true, they really do rig modern gpus to only last two or three years
Quite the opposite for me.
In 10 years of heavy gaming, I've never had a GPU, that wasn't bad when I got it, die on me. I eventually gave all of them to family/friends, where they still work if they weren't replaced for performance reasons.
That said, I got more GPUs that were bad on arrival than good ones. One was artifacting bad, one was DoA, and all the others were overheating out of the box. 2011 was a particularly bad year. I got 6 bad GPUs, 1 bad board, 3 bad PSUs (in all 3 the switch broke after flipping it once, must have been same batch).
>Buy Gigabyte 970
>has massive coil whine
>return it, get a replacement
>coil whine again
>return again, get a replacement
>coil whine again
>return again, pay a few extra euros for an ASUS 970
>coil whine again
>return again, pay even more for MSI 970
>coil whine again
>returned and got money back
>using old 660 ti now
honestly if not for the coil whine 970 would've been amazing. Hope Pascal/Polaris(or whatever amd is calling it) won't have major coil whine.
>have a 270x
>BSOD during heavy graphic tasks
>I can open it like a cooked egg
>it was only the plastic shell
>idles at 60 degrees (C)
>will probably fail
>2005 HP shitbox
>old AMD chipset as graphical card
>can barely reach 1280*1024
>still works as lightweight linux machine
I recently broke that plastic thing on the pci-e slot, is it important? I want to update from internal graphics (128 megs) to an external gpu (at least 512 megs).
did i fuck up by breaking this piece of plastic?
Whatever the thing I surrounded is called.
sorry, but my english is lacking
>buy GT 240 in 2009
>it still works fine to this day
>buy GTX 570 in 2011
>it still works fine to this day
>buy GTX 980 in early 2015
>buy a second 980 through EBay for cheap
>both still work fantastically.
I've never had to RMA a part either because of faulty work or DOA, hope I never do. I like to think most such cases of faulty parts are due to user error but I digress. Maybe I've just been lucky for this long.
Exact same thing happened to a friend and I, both got hd 6950s around the same time back when they were first released and they died 2 months apart from each other 3 years later, just after the warranty ended.
I used to have a Dell prebuilt when I was younger. We kept it in a really enclosed area and one day the graphics card burned out and stopped working. It was a GeForce 5200. That experience made me want to learn more about computer hardware.
>Get GTX 660 in summer of 2013
>Dies at beginning of 2015 despite me only playing games from 2009 and older and rendering movies/images with 2D effects
>Error 43 in three versions of Windows
>Proprietary drivers on Ubuntu/Debian just cause computer to halt with black screen
>Hackintosh that works with GTX 650 won't even boot with it; also black screen
>Throw card in a box, put my old Radeon HD 6570 into the computer instead
>No performance drawbacks observed except in Unreal Engine 3 games
>Put the 650 back in my computer months later, wanting more speed for UE3 games and video rendering in 3D
>somewhat faster, but not always up to the task of 1440p
Does a 970 model exist with the last 512MB at a higher speed? Does the 380 really run that hot? Is Nvidia's brand really worth the extra money on the high-end cards? Will I be forced to use that Crimson shit if I use AMD?
>shit psu that can't feed the gpu properly
>somehow gpu dies first
Psu would literally fry the second you turn on the pc. One of my clients learned that the hard way.
>buys the cheapest super gayman 500w psu we had, ignoring my warnings
>it was going to feed an hd2900
>a fucking hd2900
Needless to say, in 2 days he was back with the thing fried, took my advice and bought a 550w xfx. Lucky for him, the supplier bought it when we said it was doa, so we refunded him that one.
Generally, most people say the 390 is better than the 970.
I personally own a 970 though and I've never had an issue. I'm happy with it. The last .5 is slower but I've never been affected by it on 1080p. If you play on 1440p or 4k get a 390.
380/X cards run pretty cool. The Sapphire Nitro doesn't break 70C ever.
Have you ever used Crimson? 90% of the complaints come from people who are to stupid to use the new interface (because they're used to the old one, or it's their first time even looking at a driver control panel. It's good and easy to use.
That is typically how Nvidiots (and by extension, /g/) operate.
>buy R9 380X
>play GTA V for about 15 minutes
>screen goes green, sound loops with distortion, crash/restart
>clean and reinstall drivers
>no change, same crash
>bump up power target
>raise core clock and voltage
>solve issue...for about a week
>back to same crashing
apparently it was incompatible with all 1.0, 1.1 slots and you were at the mercy of the mobo manufacturers to release an update specifically for it. it was really dumb to make a cheap midrange card that can't even be used with the motherboards that came out just a few years earlier.
>Buy Quadro K620
>reading somewhere that max fan inlet temperature is 55°C
>GPU temp is usually 40 in idle
>start freaking out:
>is 55°C the temp of the air or of the GPU it self?
>put case fan to 100%
>computer try to fly away
>I can't hear anithing for the rumor
>Email nVidia costumer care
>waiting for a reply
i know i'm incompetent
>Have a 270 for 1.5 years, runs good, overall great card
>driver update notification
>3 hours later bsod
Still don't get how the fuck a driver can kill a card like that.
>get a dell computer from my mom for graduation
>year passes and I finally have a good bit of money to upgrade
>not a heavy gamer at all but enjoy playing games on at least high with stable frame rate
>buy a 960 because on a budget
>have a 450w psu
>think that'll support it
>actually works the first time I install it
>doesn't work next time I turn computer on
>order a new power supply on amazon
>have prime so expect it Dec 30
>just now get it
>a lot of cords and in confused
>still doesn't work after installation
I either have a faulty card or I really fucked something up and I'm scared.
Corsair PSUs have terrible track record of having faulty noisy fans and/or coil whine issues. I'm having mine month old VS450 RMA'd as I write this, because of this very problem. For proof just yt search "psu noise" and most of the videos you'll find will be about a corsair PSU.
>inb4 "450W ain't enough!!"
I don't have a discrete GPU
Okay. A few questions. What PSU's did you get? What is the new one rated for, in watts? Is it turning on, or is it just not displaying things? Are you sure it isn't the monitor for whatever reason (i.e. plug the laptop into it (if you still have it) and try to get a display from that)? Is there any way you can test the video card by itself?
I got the evga 600 b. Everything works fine. Went through bios of the integrated card. it's working. Not the monitor. Fans are spinning on the 960. I think I just have a faulty card.
My dad asks me for an opinion: what PC to get for use in his office, to do basic picture editing (crop, rotate, etc) and excel sheets.
I recommend a nice AMD build for 580€ including good monitor and decent keyboard/mouse.
Over xmas, my dad announces he's just bought a new pc.
Mfw its a prebuilt gayming rig with twin 970's, 4790K, 32GB ram, and striped 1TB SSD's.
My dad has the most powerful Microsoft Paint machine I know.
>try to install new AMD drivers on linux partition
>look for a fix
>fuck up even more
>cant get to grub screen
>PANIC! SYSTEM MELTDOWN!
>fix it by repairing windows
>return to windows
I miss linux
This just happened to me, 5 minutes ago. Not a GPU story, but horror nonetheless.
>bought 250GB 850 EVO
>spend afternoon removing games to trim down my ~500GB Windows installation to 230GB
>use EaseUS ToDo to clone the drive into the SSD
>boot into it, it works
>install Architect Linux (basically an Arch install script)
>Grub doesn't list Windows
>ok, I'll fix it later
>go into Gnome
>let's install it
>gnome-terminal won't open
>Xterm gives some error opening it
>reboot, let's go play a bit in Windows
>select the SSD in the boot menu since I couldn't choose it through Grub
>"Please insert a boot disk" or whatever
>Either arch or the cloning fucked up the EFI partition, the fact that they were chained in the boot menu could mean that even if the SSD had a fucked EFI partition, it would default to the HDD which would work
>But the HDD now has Arch so Window's EFI partition is gone baby gone
>Spend the next few hours trying to recreate the EFI partition and fucking up because I kept trying to take advantage of an already existent NTFS partition while EFI has to be in FAT32
>At a certain point I manage to get into Windows again
>Unplug the HDD just to be sure
>I realize I recreated Window's EFI partition on the HDD on top of Arch's so now Arch won't boot as well
>I am not a smart man
>I give up and follow a Tek Syndicate tutorial step-by-step
Right now I'm considering wether I should keep Arch at all or try and install Fedora which has functioning Gnome. Either way, I'm scared to lose my Windows EFI again.
yeah, amd drivers on linux are terrible, the open source ones are nice and very stable but still doesn't have very for games.
fglrx on the other side, well, i just don't see any way to defend it. Hope it's the last thing that dies:
Meanwhile i'll be with the green and blue teams respectively.
>buy new 6870 4 years ago
>works fine for 4 years
>give it to a friend, cant let him play dota2 with 30fps
>buy used 660ti
>works without problem
what are you doing with ur gpus, never had anything worse than a psu with a faulty fan
>buy a Fermi back in the day
>it ran hot as fuck
>decide to pure water on it one day to cool it down like a ghetto liquid cooling setup
>shit starts sparking, and turns off
>computer is dead
>call up Nvidia and threaten lawsuit if I don't get my money bad for them not b having warnings about liquid cooling and their gpu.
>they send me two new ones instead
I do and the real issue of the last .5gb is its a XOR situation - bandwidth for the fast memory tanks when the card reads or writes from the last partition.
>got new mobo and cpu
>removes gpu from mobo
>wants to put gpu on the table while holding it's fan
>fan stays in hand gpu falls down
>cant put fan back because screws are broken
>lets fix that
>finds red string
>finds noname shit tier chassis cooler
>it werkz, fan constantly running at 55%, temps ok