Is my graphics card/motherboard/PSU dying?

[Meloy]

So a couple of times in the last month or so my computer has crashed whilst gaming (the display went all weird - random colours, pixels, green lines etc and the system froze up), each time I restarted and it was all OK again.

Yesterday it crashed but when restarted it wasn't OK; the display messed up a few minutes into operation each time. Many times it's messed up from the moment it's switched on. On the occasions it starts up OK and then crashes, the whole system freezes. On the occasions it's messed up from power on, I can still use my machine but it's difficult to use because of the display.

Uninstalling/updating/rolling back drivers on the graphics card had no effect, and the fan on the card is working fine. The temperature on the card is also fine. There appear to be no lose connections inside the machine. I get periods of time of varying length when I can use my computer just fine with no issue, and then the problem usually occurs like this:

1 - In the middle of operation (usually when I'm doing something graphically intense, but not always) the graphics go weird and the system freezes.
2- I switch the power off and on again. It starts up with the graphical problem still present, but the computer is usable. I start in Safe Mode. It often displys a "Windows has recovered from an error" message when I sign back in.
3- I disable and uninstall the Graphics Card, and restart the computer.
4- Using the onboard graphics at this point, the graphical problem still often persists. I run Windows in normal mode and reinstall the graphics card.
5- I restart the computer and run it in normal mode, using my Graphics Card. It generally starts up fine, but then crashes (and freezes) about a minute after sign in.
6- I repeat steps 1-5. Usually, if the graphical problem is no longer persisting at stage 4, then it will no longer crash at stage 5 and I have a few hours of computer use because the problem occurs again.

Anyone have any suggestions/advice? Some people have advised me that it's likely to be the graphics card, others are sure it's the PSU. Being not particularly rich, I don't want to fork out to replace the wrong part!

I've been using this system for nearly 2 years now, my card is a MSI GeForce 8800GT 512MB, my motherboard is an ASUS S775 Intel P35. I'm running 64-bit Vista. My PSU is a rubbishy 480W one that came with the case.

Many many thanks to anyone kind enough to help.

[Karnac]

I would download Speedfan and check temps for overheating, reseat all cards and ram, then swap in a known good power supply and see if the problem persists.

http://www.almico.com/sfdownload.php

[Meloy]

Thanks for the response =) I already have SpeedFan (aswell as SensorsView and GPU-z) and all temperatures are reporting fine. The only thing that's perhaps odd is when detecting the voltages, I can't get a 5v reading, and the 12v reading is coming back as 11.4/11.5v (apparently that's exceptionally low?).

Thanks again.

[Karnac]

The 12v numbers ok, but the lack of the 5v raises questions.....borrow a power supply and swap it in overnite so you can eliminate it as a cause. There's a lot on instruction available on you tube if you are uncertain.

[Meloy]

Thanks for the advice =) Unfortunately I don't have access to a spare PSU. Everything would seem to suggest it is indeed the problem though so I expect I'll just buy a new one.

[Meloy]

Incidentally, do you think it's safe running my PC at the moment, or should I leave it off or run it with my onboard graphics? In other words, is the PSU likely to blow completely?

[Karnac]

That's what it looks like....make sure you get a decent PSU...the cheap ones don't last.....let us know how you make out.
How much ram is in the PC?......
I think you're safe to run it......power supplies are fickle, sometimes the problems intensify over time and sometimes you'll go to turn it on and nothing.....it can take months before they go completely.

[Meloy]

There's 4GB of RAM on the PC. I've run a memory test and it came back fine. I'm fairly sure that the problem isn't the graphics card, because it persisted when I uninstalled it. However, I can't physically remove the Graphics Card to make 100% sure since the onboard graphics has nowhere for me to plug my monitor in =/

[Karnac]

Ok ... memtest was my next suggestion, but I'll stand by the PSU going bad.

[Meloy]

I'll get hold of another one ASAP and let you know how it works out. I just hope it's not the GRaphics Card or Motherboard >.<

[Karnac]

Well, it's always good practice to install a new PSU when installing a new MB.....if that's any consolation...Good luck.

[Computer_Commando]

...I can't get a 5v reading, and the 12v reading is coming back as 11.4/11.5v (apparently that's exceptionally low?)...

SpeedFan Config---Voltage tab:  check all boxes
I'm not sure what uses the +5.0V rail anymore.  Maybe the +3.3V rail has taken over.  Otherwise with no +5.0, the computer wouldn't even turn on.
Your +12V rail is a bit low, should be over 11.5
Even if your computer doesn't use the +5.0, the fact that's it's 0, says the PSU is bad.

[Meloy]

Update: it lists voltages in my BIOS too, and gave me a 5v value that was very close (I can't remember exactly what it was). It also gave me a 12v value of just over 12v.

I happened to have SpeedFan running when the last crash occurred, and it listed "AUX" as running at 127c....what is AUX? That seems far too high.

[Karnac]

You can disregard the 127c ... it's the default for when no sensor is found I believe.

[Meloy]

That's a relief!

I ordered a new PSU today so should come tomorrow or Wednesday. I'll let you know how it works out Is replacing the PSU a simple case of taking the old one out and putting the new one in and plugging it into the various components? Or do any kind of BIOS modification etc need to be made?

Thanks again!