My self-built PC freezes up, have been unable to troubleshoot the problem

Kristmas Kthulhu

Hello H/A, I built my first PC this past December, and for two weeks or so was an incredibly proud father. I honestly can't remember when the problem first started, but it may have been there all along, and I never noticed because it just took a long time to install all the software and whatnot I wanted.

What happens is this: whenever I'm playing a game on my computer (and only recently when I occasionally watch videos on Dailymotion), my computer completely freezes up, displays either a black screen or a tan/beige color with vertical black stripes, and loops sound continuously until I perform a hard reboot. I can usually tell when it's about to happen because the screen flickers a few times, I think mirroring what's on the monitor. Sometimes the flickering will happen and I can continue playing normally, and other times I get a freeze with no warning whatsoever.

The strangest part about this problem is that it happens in a predictable manner with some games, unpredictably in others, and never with only three that I know of. What I mean is that in the first category, this freeze only happens on the mission recap screen in Alpha Protocol. I've put 35 hours into that game, and it is the only time I've ever needed a hard reboot.

The second category simply means that the game could freeze up at seemingly any point for any reason. There are times when I can play for hours with no problems, and others when I'm forced to reboot every fifteen or twenty minutes until I give up out of frustration.

The third category consists of Dead Space 1, Batman: Arkham Asylum, and Sequence. There's probably been forty hours played between the three of them, and never once have I needed a reboot. The thing is, there is a not insignificant possibility that the freeze up will happen within sixty seconds of exiting out of these games.

I've contacted my video card manufacturer and my motherboard manufacturer, which had some decent ideas, but neither of which ultimately solved my problem. I've updated every piece of software and hardware I know of, and at this point I've almost resigned myself to not being able to use my computer for some of the things it was built for.

My Specs

Operating System: Windows 7 Home Premium 64-bit
Processor: Intel(R) Core(TM) i3-2100 CPU @ 3.10GHz (4 CPUs), ~3.1GHz
Memory: 8192MB RAM
Card name: AMD Radeon HD 6800 Series
Display Memory: 743 MB

Is there anyone out there that can help me?

Handgimp

You should detail what you've performed in terms of troubleshooting, and provide full system details. Does your motherboard have onboard video? If so, pull the GPU and attempt to reproduce. Run memtest86+ for 24 hours. Run Prime95 for 24 hours, ditto LinX. Basically it'll likely come down to stresstesting, and doing your best to isolate components. It's a lot easier if you have multiple GPUs/CPUs/mobos/sticks of RAM to swap around.

Be advised I am in no way an expert at PC troubleshooting/repair, but this is how I would approach the problem. Well actually I'd just buy a new computer, but I realize that's not realistic for everyone.

Luiane

Hmm, the graphics card would have been my first guess but I suppose that might be ruled out.

Have you tried testing the RAM memory in a program like memtest86? Could be due to a bad section in the RAM causing errors when that section is put under load. This would work best at explaining random errors, but since some of it seems rather systematic it might not be this.

Of course it might also be the processor - could be that there is a part of the contact of the processor that is damaged and causes errors when it is tried to be put under load. Different programs have different programming which could explain why some programs always/often crash at certain points. Other than switching out the processor and trying with another one to check if the crashes persist, I do not know of any good software to test your processor though.

It is also a long shot, but changing the sata wires for the HDD might sometimes help - I had trouble with one HDD that I could not wrap my mind about the cause, and it turned out to be a faulty sata wire.

If I may ask, how large is your PSU? Could be lack of power for some piece of hardware causing it, although I find that unlikely due to problems when shutting off programs - that should not be causing crashes in that case!

Kristmas Kthulhu

When I was on the phone with the AMD guy, we went through the device manager and checked that all of my drivers were up to date, and he said that I had my onboard graphics enabled, which may have been a problem. I went into the BIOS and disabled it, which didn't fix anything. I checked the wattage on my RAM to make sure they were being supplied the required amount; they were. I tried using one stick of RAM at a time, and the problem persisted. I've checked both of the DVI ports on my GPU to see if one of them was causing the problem, and no dice.

What system details have I left out that would be helpful?

I have a 600W Corsair PSU.

As for running memtest86, I downloaded and extracted it, giving me an ISO. I tried mounting it in a virtual drive, but it doesn't autorun, nor can I find anything to execute when I open explore the folder containing the data on the ISO. Please advise on how to run.

Can I run memtest, Prime, and LinX concurrently, or will the whole process take 72 hours?

Luiane

I have not run prime or LinX, but you need to burn memtest onto a cd and then boot from it. You will not be able to run prime or linx at the same time. A thorough memtest run takes time, but if it comes out clear you can be very certain that it ain't the ram at fault.

Ruckus

Bad RAM usually presents as a bluescreen/reboot on Windows machines. His description of the issue sounds like an overheat or bad video memory on his graphics card. I'd try running some games with the onboard video (you'll probably have to tune some settings down) and see if you get the same issue without the PCI-E graphics card installed.

Brolo

Sounds like a PSU issue.

Try load testing your powersupply - run a CPU burn in program like Everest, then a videocard burn-in program like Furmark, then try both at the same time.

If the first two work, but the third fails, it could be you're getting unsteady power on your 12V rails.

Ruckus

That reminds me, you do have the supplemental power connectors attached to the Video card right?

Hahnsoo1

What temperatures are you getting when you play? Do you have a dual monitor setup? Contacting the manufacturer does NOT rule out video card problems. You need to clean out your heatsink/fan with canned air (freeing the dust that's on there) and/or reseat the heatsink/fan with thermal compound before you can rule out the video card on the hardware end. You also need to run a memtest. The fact that it's blowing out during games strongly suggests either the RAM, the power supply, or the video card. You can't completely rule out the video card unless you borrow another card and test the aforementioned games with a different one. There's an outside chance that you are having problems because of DirectX 10 or 11, depending on what games you are playing.

Monitor GPU temperatures at idle and under load, and report back, using any number of temperature monitoring programs out there. I suspect that you'll find a problem there, and that's what I'd check first. Also, make sure all of your fans are in working order. If you have multiple fans, it's easy to miss if one of them goes out.

Kristmas Kthulhu

I tried testing out my onboard graphics, but couldn't get a single game to run on it, so no new information from that. I'll try out Everest and Furmark. What exactly are they?

The only power receivers on my video card are two six pin things, both attached with cables that say PCI-E from my PSU.

I've had a temperature monitor running in the background when I play games, and temperatures don't go over 45C. I'll check again after I post this and give you some more specifics. I have a single monitor setup. How would I reseat the fan or heatsink on my GPU? I have thermal compound.

I unfortunately don't have a spare GPU lying around, although I've had a suspicion that it has been the problem for a while now. The only thing I don't get is why Dead Space and Batman never freeze up while I'm playing.

GnomeTank

I had these problems with my 6870, almost exactly as you describe. It ended up being a two fold problem: First, I had one of the first gen 6800's (sounds like you do as well) which had notoriously bad chips. Get that sucker RMA'ed. Second, my motherboards PCI-e port was bad, and I had to replace the motherboard (replacing just one didn't fix it, I had to replace both).

The 6800 series GPU's are notoriously fickle, and the first production runs were notoriously buggy.

Hahnsoo1

Kristmas Kthulhu wrote: »

I unfortunately don't have a spare GPU lying around, although I've had a suspicion that it has been the problem for a while now. The only thing I don't get is why Dead Space and Batman never freeze up while I'm playing.

All games are not created equal, especially 3D games, when it comes to graphics card utilization. Typically, games that utilize more DirectX 10 and 11 features cause the card to heat up more, for whatever reason (sometimes it's PhysX hardware acceleration on nVidia cards, for example. Not a problem in your case, obviously). Dual monitors can also cause problems because of persistent unfixed driver issues, but it sounds like you don't have this issue. Make sure you are monitoring GPU temps specifically when you test your temps... there are multiple thermometers, all of which can give a different temperature reading.

If I suspect heat problems on a graphics card, I'd first clean out the dust before attempting to reseat the heat sink. The last time I cleaned out my heatsink, I flushed out a dust bunny the size of my thumb! I live with two cats, which means free-floating fur is a problem.

If cleaning the video card/fans in your computer isn't working, then I'd proceed to reseating the heatsink. It will involve removing the heatsink element (typically it's secured with 4 screws around the GPU, but YMMV greatly), wiping down the heatsink and GPU with a tissue to remove the prior compound, and putting a tiny drop (no bigger than an uncooked grain of rice) on the GPU or heatsink and smushing it down. Some folks advocate smearing a thin layer instead and that's fine. Just be aware that too much thermal compound means the heatsink doesn't directly contact with the GPU, and you'll continue to have heat problems. I'm not sure what too little thermal compound would do, because I've literally never seen anyone put too little thermal compound. The last time I did this, I had too much thermal compound on, and I had to remove the heatsink and wipe down the heatsink element and reseat it again before it finally started to cool like it was supposed to.

Your best bet is still finding someone with a video card you can borrow for troubleshooting. If the video card is fucked, and cleaning/reseating doesn't help, you can only figure this out if you have another video card. If all else fails and you can't find a family member or a friend with a spare video card for testing purposes, you might be able to ask someone on the forums for their old card.

Kristmas Kthulhu

After 20 minutes of Dead Space, I Alt-Tabbed and the temperature readings were as follows:

Motherboard 38, 25, 45 C
CPU 55, 55, 55 C (MAX 60)
GPU 40 (MAX 67)

Wow, the GPU got pretty high at some point during that. I'll try cleaning it out with those compressed air things.

Oh, and as soon as I exited to Windows my computer froze.

It's somewhat good to know that the problem likely is with the manufacturing of the card, as I have a 6870. How would I go about proving to AMD that their card sucks and that they should give me a new one for free?

GnomeTank

67C is not hot for a 6800, that's very normal. Mine regularly spikes to 75C under full DX11 load. GPU's run hotter than CPU's, remember that.

Most GPU's can take 100C pretty easily, though you generally don't want them that hot.

Also, you can't go directly through AMD, you'll have to go through the manufacturer. Sapphire, MSI, Asus, whoever made your card. Mine was MSI, and they were relatively helpful in RMA'ing mine, though I was without a video card for two weeks.

Eat it You Nasty Pig.

Yeah, the easiest way to identify the problem is going to be to swap parts and try to recreate the issue. Monitoring your temps and using other testing software might not find the problem if there is a hardware failure somewhere on the card.

If you're still under warranty on the card I'd get it replaced; otherwise maybe pick up a comparatively cheap card and see if the problem persists. It's an expense, but at least then you'll know.

GnomeTank

Really just find any working PCI-e card and try it. Borrow one from a friend. Hell, if you want to pay shipping, I'll send you an old 8800GTS you can try it with it, just to see that for sure it's your video card and not the motherboard.

Kristmas Kthulhu

I might be able to salvage an incredibly old GPU from the defunct desktop at my parents' house, though it'll probably be a week or so before I can go.

I would definitely be willing to pay shipping for your old card if it's reasonable, GnomeTank.

GnomeTank

I'll check what it would cost me to safely ship it. It's already in a static bag, but it would need to be in a box with those air cushion things. I'd imagine it'd probably be ~20 bucks to ship safely, at a reasonable pace (aka won't take three weeks to arrive).

Marty81

I had a similar-sounding problem a long time ago that I finally ended up tracing to a faulty power supply (one outputting voltages outside of the +-5% tolerance on certain rails under load). My power supply actually had gone REALLY bad...when I went to pull it out of the motherboard, its connectors were burned and some of the plastic around them had started melting.

Draygo

If it is not the power supply,

95% chance its the motherboard, ive had similar issues in the past and swapping graphics cards did not resolve the issue.

If the mother board is under 3 years old, RMA it. The downside is you might be without your machine for 2 weeks.

Try swapping your power supply first. Also visually inspect your motherboard capacitors. if any of them are visually buldging then 100% chance the motherboard is the issue and needs to be replaced (or fixed, if you have the know how and equipment).

Sir Carcass

When I built my computer a couple of years ago, I was getting random hard locking. It ended up being my RAM. It was rated for 1800MHz or whatever, which I had it at, but ended up having to down clock it to 1333. Memtest never detected any errors, so the RAM itself wasn't bad. After doing that, it hasn't happened since.