PC freezes: bad video card?
August 13, 2006 9:32 AM   Subscribe

I have a PC system that I can install an OS on, but it will always freeze up after boot from anywhere from a few seconds to a few hours. All components are new except the CD/DVD drives and the video card. Could the video card be the culprit? (much longer explanation inside).

Here’s the system:
  • Mach Speed Viper MK8-939A Via Socket 939 ATX with AMD Sempron 3000+ OEM Processor. (new)
  • 1.5Mb RAM (new)
  • Seagate 200Gb drive (new)
  • Plextor CD-ROM burner (new)
  • NEC DVD burner (old)
  • Matrox G400 dual head video card (old)
  • All cables are new.
First I tried to install XP Pro. The version I have on CD is the original release, which requires downloads to upgrade to SP2. I could only format the disk to 137Gb due to restrictions on the original release of XP. The first time I tried to install it got through almost everything and then the screen went black (mouse, keyboard ALT-CTL-DEL did nothing). I tried the installation again (reformatting the disk), this time is worked. However, after the system booted the screen would go black after the Windows startup page. I reinstalled XP, this time it would be boot up and I could login, but anywhere from a few seconds to a few minutes it would freeze (image on the screen would remain but unresponsive to mouse and keyboard).
  • To test the RAM I used Memtest86. No errors detected.
  • I ran Seagate’s disk diagnostic tools. No errors detected.
  • During BIOS startup when it reports the CPU temperature it was never above 19C/66F.
I decided to see if another OS would have the same problem. I installed Ubuntu Linux. It installed fine, but after boot and login ten minutes to an hour later the screen would be black and the system unresponsive. I then decided to rule out the hard drive. I booted Ubuntu off the CDROM (which is done during installation anyway), and after I came back a few hours later I had a black screen and unresponsive system.

I think I’ve ruled out the OS, the hard drive, the CDROM drive, the RAM, viruses (I’ve reformatted the disk for each install). The only thing I can think of is the video card.

A couple of things related to the video:
  • In the previous XP system that used the video card right before Windows started, a screen with rows of patterns of dots would appear for a few seconds (the pattern seemed to always be the same).
  • While starting Ubuntu, during boot I’d see a row of moving dots for a few seconds in the center of the screen.
  • While running XP with two displays, the additional display would occasionally have waves of static running down it (which I’ve seen on two different monitors, so that rules out the monitor).
Obviously, the next step would be to replace the video card and see what happens.

If that doesn’t work, is there anything else I should consider?

Thanks for helping a guy who is on the verge of tossing his computer out the window.
posted by ShooBoo to Computers & Internet (16 answers total)
 
Response by poster: Of course, I meant 1.5Gb of RAM.
posted by ShooBoo at 9:41 AM on August 13, 2006


What's the wattage on the power supply? If it's too low (say 300 or less), that could be your culprit.
posted by PinkStainlessTail at 10:40 AM on August 13, 2006


It really does sound like you've overloaded your power supply.
posted by Steven C. Den Beste at 10:49 AM on August 13, 2006


Could it be a heat issue? That might explain why it only starts cutting out after a bit - and why it doesn't reboot. No doubt wiser mefites can shed more light!
posted by prentiz at 11:16 AM on August 13, 2006


Are all the fans running OK? A stuck video card fan can let a PC run for a while before locking up.
posted by malevolent at 11:37 AM on August 13, 2006


It might be a video card memory issue, as well. Is the heat sink on the video card under its own fan.

To me this sounds like an overheating issue rather than a power supply problem. Have you tried putting a fan blowing across the video card (case off)?

take a look at this forum
posted by ptm at 12:17 PM on August 13, 2006


Response by poster: The power supply is an Antec 380w (it's in a Sonata case). I'll have to check out what my electronics require.

The video card has a heat pipe, no fan (I replaced the fan with the heat pipe a few years ago in the other system).

It will reboot without a problem after it freezes up (of course it will freeze again after the OS starts), and the CPU temp that flashes on the screen during reboot has always been 19C or below.

I'll try the extra fan to see if that keeps things running.

One thing I forgot to say was that after the first installation of XP, I could start it in safe mode and it seemed to keep running. Maybe because XP safe mode uses VGA graphics?
posted by ShooBoo at 12:39 PM on August 13, 2006


In order
  1. It could be heat. If it is, you should be able to correlate lock-ups with heating time. For example: See how long it takes to lock up when starting cold, after sitting unplugged completely for at least an hour and with the case open. See how long it takes to lock up when starting warm.
  2. It could be a bad power supply; however, for practical purposes the watt rating of the power supply isn't relevant.
  3. It could be a bad motherboard, in particular bad caps. For bad caps, and for some/most power supply problems, the lock ups will be really and truly random
  4. It could be bad video, or anything else too..
It is always worth running from a bootable CD like Knoppix or Bart's PE. It is also very helpful to run from both linux and windows.
posted by Chuckles at 12:51 PM on August 13, 2006


And on the power supply issue, appologise if I'm a little inconsiderate, here is a recent question that addresses the debate: Picking a new power supply for my computer.
posted by Chuckles at 12:53 PM on August 13, 2006


I had a similar problem with a different Mach mother board. I struggled for a year, fearing the worst. And it was bad, real bad. I had to -upgrade the BIOS- (insert dramatic music here). Solved everything. The trick of course is finding the correct bios... www.machspeed.com/specs/viper/MK8-939A.html
posted by Gungho at 3:20 PM on August 13, 2006


2 thoughts:

1) Don't believe what the BIOS tells you about the CPU temp. 19c sounds low, but then, when it's 19c here I think it's cold...
2) Do AMD still supply their CPUs with the chip & heatsink separate? If so, did you remember to pull the plastic cover off the heatsink compound? Conversely, did you only pull off the plastic cover & not the whole pad?

I mention those because, despite having a double-sided A1-sized instruction sheet in their retail packs, it used to be remarkably common to either not pull the cover off the heatsink pad, or pull the whole damn thing off - because it was damned near invisible.

And the fact that Linux installs OK & runs a bit longer suggests it may be CPU heat-related.
posted by Pinback at 9:26 PM on August 13, 2006


How long did you leave Memtest86 running? I've seen it catch faults on an overnight run that it missed on the first pass.

19°C sounds too low to me, too. I'd expect a lightly-loaded CPU with a good heatsink and fan to be running somewhere in the mid-30's to lower 40's.
posted by flabdablet at 9:37 PM on August 13, 2006


I'd suspect the motherboard too.

*Looks up board*

Hmm...the Matrox vid card is AGP right? It looks like that MB has something funky called an MGP slot that runs AGP cards on the PCI bus. It has a pretty short list of tested cards, all of which are ATI or nVidia based.
posted by Pryde at 9:59 PM on August 13, 2006


You could try underclocking the CPU, memory, and if possible, video card.
posted by trevyn at 10:13 PM on August 13, 2006


Yeah, that's *way* too cool for a CPU temp.

I like your taste in PS's, though. I *think* 380W is enough for that mobo and video card, though it might conceivable be borderline.

You could be seeing infant mortality though, on the PS or some other component. The PS would be the first thing I would substitute, though.
posted by baylink at 2:15 PM on August 14, 2006


That PSU is totally up to it, wattage wise--I have the same one in one of my desktops, actually. It could of course be a bad unit...

I really do think the MGP slot is a likely possibility. The whole thing is a kludge--the motherboard's chipset is designed for PCI Express, but they shoehorned in an extra slot to appeal to the people still using old AGP cards but it's not true AGP. They evidently did very limited testing using only common ATI and nVidia cards, so it wouldn't be at all surprising if it's not compatible with a Matrox one. There are even a few Newegg reviews complaining about problems with more mainstream cards.

Ideally the OP could test out a PCI-X video card in the system, and also confirm that the Matrox still works in another computer.
posted by Pryde at 5:29 PM on August 14, 2006


« Older time-lock safe legal issues   |   What is the most decorous way for me to pull out... Newer »
This thread is closed to new comments.