Computer restarting

October 8, 2009 4:29 PM

Computer restarts when I use computer games. Just rebuilt it with a Radeon HD 4890 to add insult to injury.

Ok, I just bought and installed a Radeon HD 4890 video card. All is well, except whenever I try to play any graphic heavy computer games (Sims 3) the computer just shuts off and restarts after a while.

I noticed that my device manager says that I've got a display adapter from the 4800 series, not 4890. I looked online for some solutions and found this one, which suggested I update my video drivers. I tried it, but it's not working. I restarted in safe mode and ran the video driver update program from ATI but it wouldn't recognize any of the new drivers. I simply deleted the display adapter and restarted the computer, and it replaced with the bad 4800 again.

My SO thinks it might be because I've got the 32bit XP, so he wants to upgrade it to either that or Windows 7. Is he right? What do I need to do?

posted by stoneegg21 to computers & internet (27 comments total)

Depending on how much more power this new video card uses, you may want to examine your power supply rating. It may be drawing more power than is available, making your system unstable.

32-Bit Windows XP is not a symptom, despite what alternative-OS enthusiasts may suggest. I would examine hardware simply because you aren't getting any error screens.
posted by dinx2582 at 4:32 PM on October 8

The two most common reasons for that kind of behavior are: 1. You've overloaded your power supply. 2. You're overheating.

Neither of those are likely to respond to updating of drivers.
posted by Chocolate Pickle at 4:32 PM on October 8

Of course, it's entirely possible that the new card is bad. Don't ignore this as a possibility, for it has been known to happen.
posted by dinx2582 at 4:33 PM on October 8

Overheating would typically involve glitches, 'random' graphical errors, etc. If everything appears fine the instant after the computer reboots, I would consider this to be less likely of an explanation.
posted by dinx2582 at 4:34 PM on October 8

Also, check that it and the RAM are seated correctly on the motherboard. Madder stuff has been known to happen. Also open up your case and make sure the fan on the card is spinning. If it continues, try running Memtest. If your RAM passes, I'd consider returning your graphics card.

If there's some graphics card testing program I don't know about, you could try running that?
posted by Magnakai at 4:36 PM on October 8

For what it's worth, I just fixed a very similar issue with a roommate's computer. On one hand my thought process is slightly biased due to recent events, but on the other hand, the symptoms sound very, very similar. Thought it was the video card initially, turned out to be a malfunctioning PSU.
posted by dinx2582 at 4:36 PM on October 8

If my memory serves me correct (without doing any research) the Radeon 4XXX series, among other series, requires a connection to your PSU. If anything, this convinces me that irregular power-draw or bad PSU are your best options.
posted by dinx2582 at 4:38 PM on October 8

Overheating would typically involve glitches, 'random' graphical errors, etc. If everything appears fine the instant after the computer reboots, I would consider this to be less likely of an explanation.

Many modern computers have sensors that shut them down or reboot them once the die temperature exceeds a certain threshold. Like mine, for instance, which shuts down every few hours during the summer.

Furthermore, if it's one of these sensor-based systems, the time of low workload during reboot may be plenty of time to cool back down to an acceptable temperature.

Mind you, I'm not arguing for overheating, just noting that it needn't necessarily characterize itself the way you say. I think it's just as likely the power supply.
posted by Netzapper at 4:54 PM on October 8

Power supply. What brand/output? The 4890 pulls almost 200 watts peak. You need atleast a quality 400 watt power supply, or a crappy 600+ watt power supply.
posted by SirStan at 4:59 PM on October 8

Before doing anything else, disable restart on system failure.

1. Right-click My Computer, then click Properties.
2. Click the Advanced tab.
3. Under Startup and Recovery, click Settings to open the Startup and Recovery dialog box.
4. Uncheck Automatically Restart, then dismiss all dialogs by clicking OK.
5. Restart Windows.

This will let you see the difference between driver failure issues - which will cause a Windows blue screen of death - and power supply failures, which will typically just restart your computer without any kind of warning.
posted by flabdablet at 5:03 PM on October 8

Oh, and for what it's worth, I'm also tipping that it will turn out to be your power supply.
posted by flabdablet at 5:04 PM on October 8

PSU is BFG from Best Buy 650 watt.
We had to get an adapter for the motherboard. The PSU didn't have the 4 prongs in a square, it had them in a line. I don't know if that affects anything, but just in case.

The fan is spinning on the video card. the heatsink is warm but not overly hot. Case has 4 fans overall, in case that's likely to pull too much power from the PSU, but I'd think that it was ok with the 650 PSU.
posted by stoneegg21 at 5:09 PM on October 8

The PSU didn't have the 4 prongs in a square, it had them in a line. I don't know if that affects anything

Photo?
posted by flabdablet at 5:18 PM on October 8

Try speedfan, see how the temps and voltages are looking when you start the game.
posted by jenkinsEar at 5:25 PM on October 8

Off the top of my head I'm thinking overheating or not a powerful enough power supply.
posted by InsanePenguin at 5:30 PM on October 8

The fan is spinning on the video card. the heatsink is warm but not overly hot.

You can't tell that. These kinds of components don't have a lot of thermal mass, and between the time of the reboot and the time that you get the case open and put your finger on, it's had plenty of time to cool. Using something like Speedfan is a much better way to see if you're overheating.
posted by Chocolate Pickle at 5:32 PM on October 8

Sounds exactly like not powerful enough power supply. Though it seems strong enough it needs to be able to deliver enough amps on the 12v rails. I'd bet you'd need at least 36. Check the side of the PSU and see if it says it can do it.

You can check Windows event viewer and see if anything went wrong right at that time. This would point to driver/software failure.
posted by dozo at 5:43 PM on October 8

I've got a program that checks speeds and voltages, ASUS PC Probe II. It's not reporting anything majorly different when I run the games.

I did the restart, and I got a BSOD. It said "Stop c000021a. Terminated unexpectedly with a status of 0xc000005. System has been shut down."

Since I got a BSOD, does that mean that it is the driver after all?
posted by stoneegg21 at 5:53 PM on October 8

Many modern computers have sensors that shut them down or reboot them once the die temperature exceeds a certain threshold. Like mine, for instance, which shuts down every few hours during the summer.

Furthermore, if it's one of these sensor-based systems, the time of low workload during reboot may be plenty of time to cool back down to an acceptable temperature.

Mind you, I'm not arguing for overheating, just noting that it needn't necessarily characterize itself the way you say. I think it's just as likely the power supply.

Sorry, I should have clarified that in regards to overheating, I was referring to the GPU, not the CPU.
posted by dinx2582 at 5:58 PM on October 8

Here is some more info on c000021a errors

Try the "Last Known Good Configuration" option, detailed therein.
posted by dozo at 5:58 PM on October 8

Since I got a BSOD, does that mean that it is the driver after all?

Generally if a power supply is overloaded, it will just shut itself down, causing an instant restart. Your computer is alive enough to show you a BSOD, so your power supply didn't do this. So it's now a fair bet that this is indeed a software-related problem.
posted by flabdablet at 6:03 PM on October 8

650w is DEFINITELY more than enough for gaming.

While I can't speak to the nature of your BSOD, it's unlikely that a failing PSU would cause that. However, there's very little evidence to suggest that the BSOD is directly related to the system resets you are experiencing. It very well may be related; I'm just going with my gut, here.
posted by dinx2582 at 6:04 PM on October 8

Something else you can try is booting from an Ubuntu live CD and installing Planet Penguin Racer to exercise your graphics card. If it stays up in Ubuntu but dies in Windows, that puts the finger fairly firmly on the Windows graphics card driver.
posted by flabdablet at 6:07 PM on October 8

there's very little evidence to suggest that the BSOD is directly related to the system resets you are experiencing

Before disabling restart on system failure: system restarts. After: it displays a BSOD instead. That's enough evidence for me.
posted by flabdablet at 6:11 PM on October 8

Ah, somehow I overlooked where it is mentioned that restart on system failure was disable. You make a fine point.
posted by dinx2582 at 7:25 PM on October 8

Install Speedfan to track temperatures. It'll also give you info about voltage spikes or fluctations. Hugely helpful.

What version of the video card drivers?

What motherboard?
posted by Lord_Pall at 9:06 PM on October 8

In 2001 I had this exact problem on a Dell laptop. The issue was that whenever I'd start to play a game that required Direct3D, the laptop would shut off.

I went a few rounds with tech support. Dell replaced a few parts. Eventually I did my own troubleshooting and discovered that it was the Crucial RAM I'd used to do an upgrade. It seems the crappy Dell motherboard couldn't supply enough voltage to the RAM (but only in certain video modes) and the laptop would respond by instantly shutting off.

Eventually I replaced the RAM with a crappy brand that seemed to be able to deal with the Dell motherboard's inability to perform up to spec.

I'm not suggesting that this is exactly what's happening, but if nothing else you may want to consider the RAM as a possible culprit.
posted by krisak at 4:56 AM on October 9