Why is City of Heroes overheating my PC?

December 1, 2006 11:36 AM

ComputerHardwareConundrumFilter. I built my PC about two years ago. Works great, mostly. But lately, when playing one particular game, my PC performs for only 16 or so minutes before shutting down completely. As you might expect, there's a lot ...

Thanks in advance for your attention. Here's the conundrum for you:

I've been running an MSI Nvidia GeForce FX5900 XT for a little over two years now. It is no longer the top of the line, if it really ever was. However, it has performed admirably in most of the games that matter to me. Oblivion quickly became a slideshow, but Half-Life 2/Counter-Strike: Source look good and perform just fine with a little in-game tweaking. World Of Warcraft worked well, as did City Of Heroes/Villains ... for awhile.

However, for the last five or six months, I can't get more than 20 minutes of gameplay out of City Of Heroes. Even with the in-game settings dropped to half, my entire system shuts down. Completely. No blue screen, no warning. And the worst part? No event registers in the XP system log.

My frustrated guess is that COH is particularly apt at targeting the RAM chips on my FX5900 XT and overheating them. Of course, my assessment of this overheating is very unscientific. The FX5900 XT has no on-board temperature monitor, so I've only my index finger to depend on.

In an attempt to get a little more life out of this otherwise okay card, I purchased an Arctic Cooling Silencer and installed it last night. I figured that would be an improvement, as MSI's stock heatsink/fan only covered the GPU, not the RAM. (I posted photos of the Arctic Cooling install to Flickr. Perhaps there was a mistake in my installation.)

But even after installing the Silencer and popping it back in the case, I could only stay in COH for 16 minutes. A reach into the case confirmed that the back of the card under the RAM chips was just as hot as before.

I'm sure I need a new card, but I'm quite keen on exhausting as many available options as I can. I'm also one of those poor souls with an almost defunct AGP slot, not a PCI-e, so my replacement options are limited.

So I ask you all ... given my situation, has anyone ever heard of such a thing? Any advice? And if the problem is not my super-hot card, then what else?

Some folks have suggested that the power supply might be dropping the ball on this one, that maybe the power output has diminished over the past two years. This is possible, but wouldn't such an issue manifest itself in more than just one particular application?

For reference, here's my set-up: AMD Athlon 64 3000+, Chaintech VNF3-250 Motherboard, 1 GB RAM Mushkin {3200) DDR400, Western Digital 160GB SATA harddrive, MSI NVidia GeForce FX5900 XT 256MB, all powered by a StarTech 480w Silent Power Supply.

posted by grabbingsand to technology (17 comments total)

I wouldn't expect a complete shutdown from an overheated video card; display might go, but most systems won't throw the panic switch from just the loss of display. I'd expect that sort of behavior from an overheated CPU, and actually had that exact problem with an Athlon a couple of years ago. The next time the machine shuts off, start it up and have a look at your CPU temperature in BIOS; that would be my guess for the source of the problem.

It sounds kind of weird that this would be happening now, but it can be a problem of attrition--when I finally pulled the heatsink off my processor to assess the problem, it turned out that the heatsink had gradually shifted slightly off-center, so the cooling was getting worse with every passing day. Some Arctic Silver and a quick reseating of the heatsink cleared it up.
posted by Mayor West at 11:54 AM on December 1, 2006

Check for BIOS upgrades especially if the MB was built before the graphics card. I had a similar problem and the BIOS update solved it.
posted by Gungho at 11:55 AM on December 1, 2006

Your motherboard probably has an overheat protection switch that gets tripped when the CPU temperature crosses a threshold (90F maybe). When these things get tripped, the computer will shut down hard, no time for Windows to save an event to the event log.

Get a program to check out your CPU temperature, and if it's running hot, take the steps to cool it down (i.e. bigger heatsink, case fans, thermal paste).
posted by knave at 11:55 AM on December 1, 2006

I've been wondering about the CPU temperature. So much so that I've been considering a Freezer 64 Pro for it. Still seems odd that one specific application would throw it, but anything is possible. Thanks for the answers so far.
posted by grabbingsand at 12:09 PM on December 1, 2006

If this is the CPU overheating, maybe it just needs a good cleaning?

Check the heat sink for dust crammed in between the fins. If it's dirty, buy a can of compressed air and get rid of all the crap. Your PSU might benefit from this as well.
posted by utsutsu at 12:18 PM on December 1, 2006

Something similar happened to me when it was determined that it wasn't heat (as I'd originally assumed) causing trouble, but output that was lacking on my mobo.

Normally and with most games it ran fine, but the extra stress of Q3 (dating myself there) killed it within 15 minutes of launching the game.
posted by owenkun at 12:31 PM on December 1, 2006

This sounds like a ventilation issue. I suspect that a couple of case fans to push out all the hot air you're moving inside that box would make a world of good. One fan pushing air in from the front and another pushing air out the rear.


Also, some chipsets are unstable at 8x AGP. Have you tried 4x?

Have you tested your RAM?

Have you dusted everything off?

Have you reinstalled directx, the drivers, and the game?

If that fails its most likely the PS or the card or the system RAM. My best guess is system RAM is not so good.
posted by damn dirty ape at 1:06 PM on December 1, 2006

It certainly sounds like something's overheating, although you may be surprised to find out how hot video cards can get without exceeding the temperature range that they were designed for. The CPU may also be the culprit, or something else.

The last two times I've seen something like this happen, it was a dead fan. Check the fans on your video card, CPU, and power supply to make sure they're all blowing strongly when your PC is running. If they're not, replacing them may solve the problem.
posted by Vorteks at 1:31 PM on December 1, 2006

In my experience, an overheating video card displays an impressive amount of corrupted polygons before crashing. If you're not seeing this, I would say that the video card is fine.

Try running the PC with the case open and a large fan blowing air in. If you still only get 16 minutes, it's not a heat problem.
posted by krisjohn at 2:02 PM on December 1, 2006

This is great information.

Ran the memory test linked above. No errors there. Good little utility though.

The fan on the video card is a-blowin' fine, but the heatsink/fan on the CPU might not be doing the trick anymore. When I installed it, I trusted the already-placed thermal pad from AMD. It's possible that the original stuff has lost its effectiveness.

When I checked the BIOS settings a while ago, the shutoff CPU temperature is set at 65°C. The CPU temp at normal load is 45°C (case temp is 33°C). Next time it shuts down, I'll see what the temp is then.

I've been considering adding one more fan to my case, but that would necessitate cutting a hole in the side panel (I've already added a 120mm to the back near the CPU).
posted by grabbingsand at 3:07 PM on December 1, 2006

You can get apps that display your CPU temperature. Find one for windows, then get prime95, which will run your CPU at 100% and sit and watch your CPU temp. If your machine dies when it hits 65 then there's your problem.
posted by markr at 7:22 PM on December 1, 2006

As far as Windows monitoring apps go, SpeedFan is awesome. It has a logging mode, which I've used to diagnose heat problems in the past.
posted by addyct at 8:24 PM on December 1, 2006

Last night, I upgraded my BIOS. Accidently, I might add.

I was only intending to check out the built-in AwdFlash on my motherboard on a reboot, to see what it needed. Unfortunately, I'd copied the new BIOS onto a floppy already, one that was still in my machine. Instantly, it started the flashing. Then it rebooted ... to a black screen. On top of that, the motherboard started making long beeping sounds.

But after finding the CMOS jumper and resetting it, plus pulling out the CPU heatsink and treating it to a vacuum hosing, the machine rebooted just fine. I had to temporarilly use a PCI video card, as the default BIOS settings did not seek AGP first, but after making a couple of changes, all was well.

And since the BIOS upgrade, my CPU's idle temperature is around 33°C (12°C lower than before). At peak (during COH), it barely hits 45°C!

So the solution was a combination of cleaning out the heatsink and fan, plus upgrading the BIOS. Thanks AskMe!
posted by grabbingsand at 9:00 AM on December 2, 2006

I would rather piggyback off of this question than add another one, since my problem is so similar. I have a Dell Optiplex GX270 minitower, which is known for having bad caps. Mine does, and Dell is sending a replacement motherboard for me to install.

But I have noticed something that I can't explain with the powering down of my computer. When I run it with 756MB of RAM, it never shuts down. As soon as I replace a 256MB stick with 500MB, however, I get the same problems mentioned above. Why does that extra RAM make such a difference?

I should also mention that the GZ270 has a 180V power supply and I am running an Nvidia 6200 card, which supposedly requires much more than 180V. The RAM isn't defective, and I installed SpeedFan, and the computer temperature has never gone above 48C.
posted by billtron at 4:41 PM on December 2, 2006

FWIW speedfan lied to me about the temp of my harddrives. HHD Health seems to be better.
posted by Medieval Maven at 9:10 PM on December 2, 2006

For your video card -- and likely for your extra RAM as well -- you're going to need a bit more juice. A little research has found a replacement power supply for your Dell that provides 250 watts. Might be enough for you.

If you want to run some numbers, this PSU Calculator is pretty handy to check your actual needs.

(Oddly enough, my success from last night took a turn this evening, so I'm back to looking at my own PSU as a shutdown culprit.)
posted by grabbingsand at 9:12 PM on December 2, 2006

grabbingsand, your link unfortunately didn't survive your post. Do you mind sending it along again? I can only find a 210W ps for the Small Form-Factor GX270, so I want to know what you found.
posted by billtron at 2:23 PM on December 3, 2006