Graphics Card random shutdowns and crashes

[ArcRaizen]

I originally posted a topic about this in the BSOD Help Section here: http://www.sevenforums.com/bsod-hel...00116-accompanied-display-driver-crashes.html. Unfortunately, no one has responded in about a week and the issue remains, although it no longer causes BSODs, so I am starting a new topic here.

Basically, for the past few weeks, I have experience an unbelievable amount of graphics card/driver crashes. My graphics card is an nVidia GeForce GTX 580M in an Alienware M17x R3 laptop. Originally, the screen would black out and recover with the OS giving me a message saying that the driver crashed and recovered successfully. The extremely odd thing about the crashes is that they aren't caused by anything in particular and can happen within 2 minutes of each other or take hours or even days. The crashes seemed to happen in repeat bursts once it crashed once, then not again for hours. I've gone through extensive testing and attempts to fix the issue, but nothing yet. It is likely hardware at this point, but I'd still like to be sure.

I've un-installed, re-installed and upgraded my drivers multiple times with no success. Most recently, I installed the original drivers for my card provided by Dell and then let Windows upgrade the drivers to 331.82. No drivers prevented the crashes. I updated my VBIOS to the most recent version as well with no luck. I've cleaned and scanned my computer with multiple anti-virus and spy/adware removal programs. Rebooting the computer with a clean boot still has crashes, so it isn't a specific services causing the problems. Changing the TDR Delay via regedit has no affect on the crashes. I tested my RAM with MemTest86 basically for the hell of it, but its perfectly fine. I also tested my video memory with Video Memory stress Test v1.7, but that also showed no errors. Lastly, I used a Win7 installation DVD to "upgrade" my OS and replace all the Windows files without having to do a complete format and re-install. Still crashes. I removed my battery and ran solely from AC power, which resulted in a crash too. I've attempted to test running only on battery, but my battery is old and barely lasts 20-30 minutes anymore, but I haven't experience a crash only on battery.

As mentioned, the crashes used to result in a BSOD with 0x116 error code, but that is not the case anymore. Now, the display will entirely shut off (i.e. no backlight even on) and will never recover like it used to, forcing me to manually reboot the computer. For the past few crashes I've tried running some GPU performance logging software to see if they could tell me anything notable. I've attached 2 logs from 2 different programs, GPU-Z and EVGA Precision, both running during the most recent crash. One shows basically everything shutting down, one shows a seemingly garbage value for most readings right before or immediately after the crash (not sure which). They're probably useless although both show that they video card isn't overheating, but here they are regardless.

 

[bulletng]

From my experience you can't do much to solve issues like this. I once RMA'd a GTX 280 four times, with a similar problem as yours, until they decided to just give me a different card. Although that may be difficult to do since you have a laptop. Would the Alienware happen to be under warranty? You may be able to contact them for support even if it's not under warranty.

 

[ArcRaizen]

It is not under warranty anymore, otherwise I would have called them long ago. I will likely contact them soon if just to see if I can buy a new video card from them. I know a place from a friend that can and will replace the video card, if need be, since I cannot do it myself, but I still need to get one first.

 

[ArcRaizen]

Ok, I figured out some stuff recently and I'm going to bounce it off anyone here.

First, as mentioned before, I've been running some GPU logs and every single log has says the same thing: Memory Clock at 1500 MHz. It seems 1500MHz is the max speed listed for my video card.

Second, when I run off just battery, I haven't gotten any crashes. Running on AC has resulted in multiple crashes.

Third, I was testing these two things with Furmark running. Twice on battery for 5-10 minutes resulted in no crashes. After the first 5 minute session, I plugged in my AC adapter and it resulted in a crash within 3 minutes. EVGA Precision log again showed 1500 MHz Memory Clock reading right as/before it crashed. Charged the battery again and ran Furmark on battery only for ~10 minutes. No crash. Plugged in AC adapter and within 10 seconds Furmark showed the Memory Clock reading at 1498MHz and then immediately crashed. EVGA Precision log showed it again hit 1498 then 1500MHz right as it crashed.

So it's either one of 2 things. Either the AC adapter isn't supplying enough power to the GPU, or the GPU is throttled when only on battery and isn't able to reach maximum power. If it's the former, I just need a new AC adapter. If it's the latter, I need a new video card. Can anyone chime in and explain if either/both of these ideas is correct or incorrect?

 

[essenbe]

I'm just making a wild guess here, but look at your power plan settings for your GPU. Most laptops have on battery plan and on AC plan and see if there is a difference. There usually is, as on battery is more in a power saving mode.

 

[ArcRaizen]

I have changed that setting to run on max on battery and there are no differences in Furmark.

For the time being, I'm using a spare AC adapter from another laptop although it has lower amp output than necessary for my computer, but it keeps it running at least. No crashes so far with it even after Furmark.

 

[essenbe]

Seems like you found your problem.