Could this be a video card error?

whisper

whisper

Limp Gawd
Joined
Mar 16, 2003
Messages
466
Hi everyone, hope you can help me here. Please see my sig for details on all my components. Yes, it's an older rig but has always been very reliable for me. Recently I've been having a lot of problems with system freezes and they are always immediately preceded by a couple quick flashes on the screen (usually displaying varying patterns/colors of horizontal and vertical lines). The system will then completely lock up and I am forced to hit the reset button or completely shut off/turn on again.

I noticed that this can happen even when I am simply in my bios - I can get into my bios on startup by hitting the delete key. I let it sit there for a minute or two and at random it will just flash a couple times then lock up. So the problem does not require windows to load up and therefore seems unlikely to be related to any driver issues.

For the time being I've clocked down a bit to 245x9 on the DDR333 divider, but my OC has been stable for almost 4 years so I'm having a hard time believing that's the problem. I was recently doing some routine testing and ran memtest for ~30 minutes with no errors. I also ran prime95 for ~30 minutes with no errors.

A long time ago I built a rig and the video output was completely messed up - it displayed patterns similar to what I'm seeing now. Turns out it was a defective vid card which once replaced, solved the problem. Unfortunately I don't have another PCIE vid card handy to swap in right now, but what do y'all thing about this? Does it sounds like a dying vid card? Is it normal for a vid card to just suddenly give out?

Thanks - it's frustrating having a very unstable rig after having a lot of success in the past. I hope I can solve this!
 
Obviously it's impossible to say for sure over the internet and without actually swapping a new card in, but to me it sounds like a bad graphics card yes.

One thing you could check is whether the fan on the video card is spinning. Often the fan is the first part of a video card to fail, and could possibly overheat the memory giving you corruption. I doubt it though, since it's happening even in the BIOS.

I've seen bad monitors produce similar problems. I had one which would crash my computer all the time. I ran all sorts of tests (Prime95, Intel Burn Test, memtest, etc) and didn't find any errors, but when I replaced the monitor all the problems went away. By the way I find Intel Burn Test will find instability in your overclock much faster than Prime95 (and it also gets it to higher temps). YMMV though.
 
Hey man, sorry to hear you're having problems. Yeah, videocards do frequently up and die like that quite often. However I'm inclined to think its something else if it occurs even in the BIOS.. doesn't seem to me like a graphics card issue, even if its hardware would cause random crashes in BIOS but I could be wrong.

My first thought was actually memory but you said you ran Memtest... did you let it finish a complete pass or did you just stop it after 30 minutes? Some of the later tests can catch memory errors that won't show up in the earlier ones.

Could be a motherboard issue, might wanna look it over for bad caps (bulging, leaking). Its not as common as, say, RAM problems, but motherboards do go bad too, my P7N SLI Plat. went screwy on me awhile back and I had to RMA.

I haven't been doing this long enough to experience it myself, but from what I've read on these forums it also isn't unheard of for an overclock that had been stable for years to suddenly become more stable, either requiring more voltage or simply a lower clock. That is an old system, so I'd try it at all stock just to confirm or eliminate that.
 
Thanks for all the feedback. A couple things I've noticed and worked on over the past day or so:

First, I opened up my case and most things looked in order. However, I noticed a large amount of dust and material covering the CPU heatsink (I have cats so over time this can be a problem in my house). Using Core Temp I had also noticed I was idling around 49-50C which is very high for my CPU. After thoroughly cleaning this up I am now idling around 37C, and hitting about 49-50C under full load, which seems more normal. Maybe this will help some.

I also installed a utility called SpeedFan but I'm still trying to figure out if the temp readings are accurate and what they are referring to. One of the temps is labeled "GPU" and it's reading 68C... if this is referring to my vid card it seems to me like a very high temp. Could anyone comment on this? Does that seem normal or could this indicate a problem?

SpeedFan also monitors 3 fans in my system but I'm not sure which is which. I imagine one would be the fan on my CPU cooler? When I was running Windows XP my fans would automatically adjust speed (RPM) under load conditions, which I could audibly detect as the noise went up. I installed Windows 7 about 2 weeks ago, before any of these problems started, and I'm noticing that the fan speeds never seem to change anymore. I ran Prime95 for a while this morning with FanSpeed running and RPM of all 3 fans (whatever they may be) never changed a bit. Anyone know how I can get my fans to start responding appropriately again? I'm wondering if this has been causing a heat issue.
 
^

That's a pretty normal temp for a video card believe it or not. Some of the Radeon 4800 series cards even run at 90C+. They are designed to handle these temps. You could try manually adjusting the GPU fan to 90-100% for awhile to reduce temps and see if that makes any difference, but I very much doubt 68C is overheating.

I'm not familiar with SpeedFan, but you could just disable all fan throttling in the BIOS (running the fan at 100% speed), and see if that helps. If you are stable in IBT/Prime/memtest it seems unlikely to be a heat problem with CPU/mobo/memory though.
 
Today I also bumped up my vcore by +0.02V. In the bios I set it to 1.37V but both Core Temp and SpeedFan are reporting 1.38-1.39V. At any rate, with this new setting and perhaps combined with cleaning the heck out of my CPU cooling fan, I have not experienced a single lock up today (fingers crossed). Among other things, I played WoW for about an hour and later encoded a 1 hour episode of Dexter using Handbrake with no hitches.

At 245x9 (2200 MHz) I'm still running slower than previously, but I'll see how it does over the next few days. It's no secret my system is pretty long on the tooth and I'll probably bite the bullet over the next few months and upgrade everything. But, I'd like to get just a little more out of it even if I have to clock down slightly.

On the topic of upgrades, I'm thinking of replacing the vid card first as I could hopefully see some immediate gains in certain applications (games) and I could carry it over to a new rig later on. It might also answer my original question about whether my problems this past week are related to my current card. I shouldn't have any problems with a PCI-E 2.0 card would I? My older mobo does not officially support this but I believe it's backwards compatible unless I misunderstood things. I'm looking at a GeForce GTX 460 by EVGA which to me seems to be in the sweet spot for price/performance. Any comments or advice on that choice?

Thanks again, as always the feedback here is very helpful:)
 
Good to hear you are lockup-free so far :)

whisper said:
On the topic of upgrades, I'm thinking of replacing the vid card first as I could hopefully see some immediate gains in certain applications (games) and I could carry it over to a new rig later on. It might also answer my original question about whether my problems this past week are related to my current card. I shouldn't have any problems with a PCI-E 2.0 card would I? My older mobo does not officially support this but I believe it's backwards compatible unless I misunderstood things. I'm looking at a GeForce GTX 460 by EVGA which to me seems to be in the sweet spot for price/performance. Any comments or advice on that choice?
Click to expand...

Yeah a PCI-E 2.0 card will work with 1.0, and it really doesn't degrade performance either unless you set Crysis on all extreme settings or something. I just ordered a EVGA GTX 460 so I can't argue there. Depending on what resolution you game at you may be perfectly fine with the cheaper 768MB version too. The 1GB seems to only really pay off at 1920x1080/1200 or higher.

Just keep in mind your CPU will definitely be a bottleneck. I had an AMD X2-4200+ OCed to about 2.7ghz (similar to your Opteron) and I was having trouble running some new games even then (over a year ago). Like you said though you can use a new video card as you upgrade other stuff. That's what I tend to do. Last year I upgraded my mobo/cpu/memory, this year I'll do video card, next year hopefully SSD. :)
 
You must log in or register to reply here.
Back
Top