Thread: Computer CSI: Video Card Edition!

So the video card on my new computer appears to have decided to fail, and we're trying to find out the exact reason why, preparatory to sending in an RMA request to Newegg for a replacement. We know it's a hardware failure and not a software failure, and we're not sure why it failed to begin with, though our most likely candidate right now is heat... though it wasn't showing any of the common symptoms of overheating prior to failure. So let's help me find out what failed on the card so we can tell Newegg when I send it back, eh?

System specs are:

Intel Core2Duo 4300 (1.8GHz/1.8Ghz) processor
2GB (2x1GB) G.Skill DDR2 533 RAM
HIS Technologies Radeon HD 4870 512MB GDDR5 video card
550W Antec power supply
22" 1680x1050 16:10 widescreen native resolution Acer LCD monitor (X223W or something like that... I'd need to go back and check)
On-board sound (SoundMAX something or other)
250GB hard drive (unsure of manufacturer)
Stock cooling setup (1x 120mm case fan, giant aluminum heatsink for processor coupled with 120mm fan, heatsink+fan setup on video card)

It should be noted that the system itself is basically a cannibalized Dell Optiplex 320. We kept the motherboard, processor, hard drive, and heatsinks/fans, but replaced the PSU, video card, and RAM, along with a new monitor and keyboard (the keyboard was just a $20 Microsoft Wal-Mart special.)

Dell, like my landlord, apparently believes people are happier when there's absolutely no free space anywhere, so the interior of the case is fairly cramped. We originally bought a Sapphire 4870 but realized it would not fit into the case with the arrangement Dell has in there (the card was both too long and too wide), so that card ended up going to my friend's wife, while we bought the HIS card for my computer.

Additionally, the computer itself was mostly rebuilt at my friend's place and then shipped to me via UPS ground. I added the RAM (replacing the single 1GB stock DIMM) when it arrived, making sure to keep myself grounded at all times by keeping an arm laid across the aluminum case side. I also made sure the processor, video card, fans, and hard drives were still properly seated and secure, since they can sometimes be jostled loose while shipping.

After adding the new RAM and connecting everything, the system boot up normally and appeared to be working just fine... which made sense, since it had been working perfectly well for roughly two weeks at my friend's house while he installed Steam and a few games to run some basic tests.

After getting it running, I first had Steam download and install the games my friend was unable to install, and it's at this time that Steam started to enjoy crashing a lot. Part of it I think may have been because I was having it download about eight games at once, but I'm not sure. At any rate, after L4D finished loading, I paused the other downloads and booted up L4D to test the hardware, and it worked fine. I ran through the first two or three levels of the No Mercy campaign running at 1680x1050 with all settings maxed and maintaining an even, consistent 30-55 fps according to net_graph. At no point, even during crescendo events with lots of action, did the system ever slow down or act wonky like I'd expect if something was overheating.

Firefox had already been downloaded and installed, and I later added Trilian basic for messaging, and my friend had already loaded Avast antivirus before shipping it. Avast and Trillian were running in the background pretty much constantly the entire time.

I did not reboot the computer once after starting it for the first time.

I later installed World of Warcraft and got it setup, running at 1680x1050 with all settings except shadows (which tend to be stupidly expensive at the higher settings) maxed. I maintained a solid 40-60 fps in most areas, and 20-25 in Dalaran during primetime playing hours - about 7:30 PM server time. I spent a couple hours doing some simple raiding and arena, and again, never once experienced any kind of slowdown or artifacting that would indicate heat issues to me.

By the third day I had the system, I was getting ready to give TF2 a whirl, and noticed that I'd have to re-add all the TTP servers to my favorites. I did so, and noticed that Steam would crash after adding a server. I added the DoD server, clicked the button to add another server, and Steam simply disappeared - as if I'd gone into task manager and killed Steam.exe. I repeated the process a couple times to add in the TF2 servers, but by this time I had to leave for work. I figured I'd reboot the system to see if that did anything to help.

When I rebooted the system, my monitor never got a video signal. Uh-oh.

I sat patiently, waiting for the monitor to get a signal, knowing that sometimes it can take a few seconds. I continued waiting, and then heard the little noise that lets you know Windows has logged in and started normally. Oh, shit.

At this point, I've become concerned (duh.) Since the 4870 has two DVI ports, I unplugged it from one (Windows played the little two-note sound to let you know it's aware something has been disconnected) and plugged it into the other. Still no signal. I then hit my power button to cause Windows to shut down and let it go through the process (using the logout sound to know it did it normally.) I then waited about two minutes, and boot the system back up. Again, no signal. I waited patiently and eventually got a signal, revealing a Blue Screen of Death, telling me ati2dxag was apparently not working right and that Windows was shut down in order to prevent damage to my system. It advised to continue on as normal if this was the first time I'd seen the screen - and it was.

By this point, I was out of time to fiddle with it and had to go to work, so I shut the system down, disconnected it from the surge protector/powerstrip, and left for work. I came back about twelve hours later (this morning) and turned the computer on, and immediately got a video signal! Only problem, this signal also had a lot of artifacting, even on the boot screen. Houston, we have a problem.

I let it boot normally (wincing at all the artifacting) and once it finished booting into Windows - about to the point where it asks you to log in an account if the computer is setup to have multiple user accounts, which mine isn't - I promptly lost the signal. Switching to the other DVI port (while powered off this time) simply results in no signal at all.

By this time, I realized this wasn't going to magically fix itself (like similar problems on my old system somehow apparently did.) I decided to try entering Windows on Safe Mode to remove and reinstall the ATI drivers, since the BSOD clearly said they were causing problems. System boots normally, enters safe mode... and the signal goes MIA again. sigh.

I then reboot, open up the BIOS menu, and switch the video adapter preference from Auto (which selects the add-in card automatically if available) to On-board, turn the power off, attach the monitor to the VGA port on the motherboard, and turn the computer on. Success! Perfect signal, with no artifacting of any kind. This is the first part that made me believe that the problem was certainly not with the monitor.

I then boot into Safe Mode with Networking, download the latest Catalyst drivers (9.5), and then go into Device Manager and remove the 4870, and reboot. Still using the onboard video, I boot into Safe Mode again, and fire up the ATI installation program... and it tells me that either my operating system or hardware is incompatible with the drivers. ... What?

Windows automatically recognized and installed the 4870 on booting up, so it's back in Device Manager, and Device Manager says that it's working fine (though not being used at the moment.) Hmm.

I then fired up dxdiag to see what the problem is, and it says it's running DirectX 8.1 (wtf?), and that it cannot detect any kind of video card to test DirectDraw or Direct3D. I then enter Add/Remove Programs and use the ATI Uninstall utility to remove everything related to ATI from the system.

I reboot, set the video card preference back to Auto, turn off power, disconnect the monitor from the onboard video's VGA slot and reconnect it to the functional DVI port on the 4870. Cross fingers, boot up. I get a signal, but it's still got severe artifacting. On the other hand, the system boots into Windows normally, and I still have a video signal (albeit one with bad artifacting)! Elation!

I then install the 9.5 drivers (which this time don't complain one bit), and reboot. Guess what? It's back to the way it was - bad artifacting right up until it boots into Windows, at which point the video signal goes completely MIA. At this point, I'm not left with many ideas left. I pop in the Windows CD that came with the system and boot off the CD. I instruct it to delete the current partition on the hard drive, format the hard drive (the slow way) and create a new partition and install Windows onto the new partition. There is severe artifacting throughout all of this (I was actually having to guess at what key to press to continue on at some points because it was so screwed up.) Windows completes its installation, reboots, and enters the setup/configuration phase. There is artifacting during this as well, but it's not quite as bad (the artifacting is much less severe in Windows, where it's just a couple of wide pinkish bands on the left side of the screen.) Windows boots up and some things become apparent - the system did not install drivers for the onboard sound, ethernet, or some other onboard things. My friend said this is normal, and that I'll have to use another computer to go to Dell's site and get the drivers and install them manually. Thanks, Dell! I'm glad we know enough about computers to do this, but I guess your other customers are just screwed, huh? I guess it keeps their tech support guys in business.

At any rate, I install the more-stable 8.12 Catalyst drivers this time, reboot, and again, I experience severe artifacting until Windows boots, at which point I lose the video signal completely.

Since a complete format of the entire hard drive coupled with repeated reinstallations of video drivers (and the fact that the monitor works perfectly fine with the onboard video as well as when hooked up to the DVI port on my old system's 9600XT) did nothing to fix the signal-loss problems or the artifacting problems, we're lead to believe that something has gone wrong with the video card's hardware itself. Thing is, we don't know what happened to it... or why.

When I rebooted, setting this chain of events in motion, the only things running beyond standard Windows processes (I hadn't gotten around to using Blackviper to prune the useless processes out) were Trillian and Avast antivirus. Steam had just crashed and I didn't see much point in restarting it since I was planning on rebooting anyway.

Overheating causing physical damage to the hardware is the only thing that makes even a little bit of sense to us (since artifacting can result from damage to video cards), but as I mentioned, I never once saw any kind of symptoms that would lead me to believe heat was an issue during my game time (and I played L4D for three hours on end with no breaks in between.) As mentioned, the interior of the case is very cramped which would naturally lead to poor air circulation and the case itself was in the little enclosure on my computer desk (with the enclosure/cabinet door open) which could certainly contribute to heat problems... if heat problems made a lot of sense in this case.

It's also possible the fan may have hit one of the IDE or SATA cables that pass close by it, causing it to malfunction, but I opened the case and thoroughly examined the card for physical damage not long after the problems started, and the fan seemed to be intact and undamaged. I have not actually watched to see if the fan spins when it receives power, but it's something I will do as soon as I get home from work tomorrow morning. Of course, if the fan doesn't spin up during POST under normal circumstances, this won't help anything.

My friend and I are both somewhat at a loss of being able to explain why the video card has failed, especially when it was working perfectly for nearly three weeks (my friend rebooted the computer many times while installing and updating drivers and programs) before abruptly failing.

Any of you have experiences similar to this? At this point, we're almost certainly going to contact Newegg for an RMA to get a replacement, but any advice y'all might have to prevent this kind of thing from happening again would be useful... plus I just want to know what the hell happened in the first place.

I checked it this morning and the fan is indeed working normally. I can't see anything on the outside that would indicate a problem. I'll try removing and then reseating the card later to see if that will do anything, but I really doubt it