GPU failing, purple screen of death.

kudohlol

Distinguished
Jan 5, 2011
74
0
18,640
Hi im relatively new to computers and computer parts etc, i am aware of updating drivers and what the main components of a computer are, im not aware of the latest and greatest what works best with what and what type of products are the best bang for buck because it changes all to often..

Heres my problem!

Recently my harddrive crashed on me and i had 2 replace it, but while i had my previous harddrive i'd often get my screen to cut off of power for like 2-5 seconds randomly while the pc kept running fine and i was playing games or something, then the screen would come back to normal, i also would randomly sometimes get Purple screens of death which look like the same ones a lot of people seem to be having problems with, with my graphics card.. geforce gtx 280. Since ive replaced my harddrive, ive been having the purple screen of death on practically anything regarding video. videos on the net set it off, games set it off.

I wanted 2 know what i can do to fix this, is my Graphics card dead? can it be my Mobo? my graphics card is probly only 2 years old, mobo is 4ish

Yesterday, i had my pc running fine for a good 6 hours straight games video everything, went to bed.. now purple screen of death practically anytime a video is needed 2 be played.

my pc specs are.

Machine name: SHANE-PC
Operating System: Windows Vista™ Home Premium (6.0, Build 6000) (6000.vista_gdr.100218-0019)
Language: English (Regional Setting: English)
System Manufacturer: HP-Pavilion
System Model: GM441AA-ABA m8125x
BIOS: Phoenix - AwardBIOS v6.00PG
Processor: Intel(R) Core(TM)2 Quad CPU Q6600 @ 2.40GHz (4 CPUs), ~2.4GHz
Memory: 3070MB RAM
Page File: 1189MB used, 5129MB available
Windows Dir: C:\Windows
DirectX Version: DirectX 10
DX Setup Parameters: Not found
DxDiag Version: 6.00.6000.16386 32bit Unicode

Card name: NVIDIA GeForce GTX 280
Manufacturer: NVIDIA
Chip type: GeForce GTX 280
DAC type: Integrated RAMDAC
Device Key: Enum\PCI\VEN_10DE&DEV_05E1&SUBSYS_099819F1&REV_A1
Display Memory: 2286 MB
Dedicated Memory: 1007 MB
Shared Memory: 1279 MB
Current Mode: 1680 x 1050 (32 bit) (59Hz)
Monitor: HP w2007 Wide LCD Monitor
Driver Name: nvd3dum.dll,nvwgf2um.dll,nvwgf2um.dll
Driver Version: 8.17.0012.6099 (English)
DDI Version: 9Ex
Driver Attributes: Final Retail
Driver Date/Size: 10/16/2010 10:55:00, 10023528 bytes

Sound device is realtek sound onboard.
 
Solution
Yeah, those coloured screens are usually caused by the GPU.

I'd say that would be a good idea to redo the heat sink. Take it off, clean it off, and apply some new stuff. Should help, I'm sure.
Could be a heat issue. If you run MSI Afterburner it'll show nice graphcs with temps, or else a smaller program to use would be GPUZ. Could be a heat thing.

To fix it you would have to remove the card, remove the heat sink cover, and try to clean all the dust out of it.

Could be something with drivers... maybe. Try running Driver Sweeper and delete all the drivers, then reinstall the latest from Realtek and Nvidia.

Might also be fixed if you do a BIOS update. You would have to check with HP for mobo BIOS or whatever brand your 280 is (GPUs have a BIOS as well).

Right now tho I'm leaning on heat just because HP cases aren't known to have very good ventilation and after 2 years there's a very good chance of dust cloging the heat sink(s).
 

kudohlol

Distinguished
Jan 5, 2011
74
0
18,640
Well, the tower is pretty hot, it only has 2 basic fans, and the graphics card has its own fan, so it could be heat, ill try installing one of those programs to see what the results are and post them here.

Theres no dust, because when i was first having problems like this, i opend up my tower got my compressor out and blew compressed air in all the fans and got all the dust out, everything is beloved patriot n span clean in there.

But when i first originally bought this pc, it was running with a very low end graphics card and a low watt power supply maybe max 450Wats.

Now its got 1000 watt power supply, and a monsterous graphics card that isnt even being fully used because the mobo isnt SLI capable, and only has 1 pci e 16 slot...

As for drivers i just recently reinstalled everything, this is a clean flash drive 2 terabytes bearly has 40 gigs worth of stuff on it, and all the drivers are the latest n greatest i can get on the internet.

I don't know how to update my bio's how ever, this is the stuff ive been "googling" and it seems its not something that comes with an easy answer its more technical then searching something and following 4 steps and your done, so i have to contact HP to find out what my mother board is specifically and then find the drivers for that online?


thanks for the help / advice so far.

Edit1: im running MSI Kombustor now, its doing its thing on test, ill update info as it comes... I don't get it tho, its pushing the graphics card hard right now (average frames 540-550) and its temp is going up slowly, why hasnt it crashed yet? does this perhaps explain its a mobo problem?

Edit2: haha shortly after saving this message, i crashed, i think your right its strictly (or maybe 1 of many problems) heat, it went to 101 degree's i think Celsius its calculating in, and it purple screen of deathed me, what can i do to cool it down do u have any recommendations? Part of the program you gave me has options to adjust fan speed on my graphics card, its auto'd at 44%, should i raise that to 100% or that is a bad thing?
 
Well first you can try running it with the side off your case and check the GPU temps. Then you'll know it's bad case ventilation, and probably can add some intake/exhaust fans to remedy it.

I do advise a custom fan profile. It's a great feature, although for me it didn't work on my GT240 but it worked fine on my 5850s so I don't know for sure if it will work for you. But yeah, definitely give it a try. What I did was figured out that 80% fan speed was about as loud as I ever wanted it, so I set it to hit 80% speed at around 80C.
 

kudohlol

Distinguished
Jan 5, 2011
74
0
18,640
My case is completely open at the moment with the harddrive actually laying on the old one on the floor, so it has been open for the time of which the problems have been showing.

My only questioning about it being a heat problem (because i am sure that is a problem regardless of being the only one or not); the only question i have is, why are they affecting it so much now? and not before, ive been running my rig like this for god at least 2 years now, and only now its acting up... can it be anything else?

I pumped fan speed up to 90% and it crashed in a game instantly when i opened it.


Argh this is very frusterating, i would love to buy a new GPU and Mobo and see which one is failing, but i don't know what mobo to buy for a q6600..
 
Hmm well you could try underclocking the GPU just to see. I've had grey screen issues from having my VRAM clocked too fast. Of course if that solves it then you'll know the card is getting to be faulty. Why? I don't really know. Could be wear and tear or could be from heat. Might try giving the manufacturer an email.
As for mobos... well... tough to figure it out on your own besides ruling out everything else.
 

kudohlol

Distinguished
Jan 5, 2011
74
0
18,640
Well, ive been doing more research and dabbing into all kinds of information thrown at me from more experienced friends.

Here is some info ive gathered.

1. My card stock doing nothing in a decent ventilated tower should be sitting anywhere from 30-50 Celsius max, the fact that when i started the test at 90 means the card is really in bad shape already.
2. a solution that is cheap and 100% attempt worthy is buying some thermal gel or pads and taking the old gunk off and replacing it with new.
3. Im going to buy an external fan 2moro and leave it pointing directly at the GPU 2moro and see how that does.

4. My friend says the card is probably faulty and its over heating way to fast and the fact that sometimes when i run the test it starts at 60 and other times 80 + means the card just isnt allocating its heat properly anymore.


Im going to also go buy a new GPU from a local store, bring it home test it out if it works perfectly for a few hours im gonna return it and order something more specific online.


Sadly my GPU is made by BFG which i just recently found out haha (today! :) ) that they no longer are servicing their "life time waranty" because the company stopped producing / manufacting gpu's as of last year. Lucky me, ill probly be paying full price for a new card!

Edit: The gpu worked for 2 years, so heating wasnt an issue when i first got it, it must be something with the card now, but im still gonna try some of the methods ppl are recommending just to cool the card down for temporary use, or to use it till its truely fucked, pardon my french/
On the bright side, its probably not my mobo or anything else, so at least i know its most likely the GPU now.