Display chaos / Driver failed to restart / BSOD

[Duchovnik]

Hey guys get ready for a tl;dr


I have a HP dv9830 DTR and I've babied it since day one over 4 years ago. Had various problems with Vista, so over a year ago I installed Win 7 and it was running like a dream until now.

I'm at my wits end, I've been trying to solve this for several weeks. I was preparing my comp to run Diablo III, hadn't installed it yet. It all started shortly after updating my GPU with new Nvidia drivers with Diablo III support and a Windows Update. I allowed the Windows updates to install several days after installing the new Nvidia drivers. At the very first boot after installing Windows updates, my computer went to hell. At random intervals, most often when loading Windows, the display would start flickering, sometimes going black and then coming back on, then a BSOD 0x0000016 nvlddmkm.sys. Sometimes the BSOD listed something different, but usually it was the nvlddmkm.sys. Sometimes the driver would be restarted after flickering, but a few minutes later it would happen again.

Everything runs smoothly in Safe Mode and in Normal Mode when using only generic VGA adapter driver.

I suppose I could go on with the generic driver, at least it lets me use the native resolution and unless I use VLC player or Photoshop or play a game you can't notice a difference, but it still sucks for obvious reasons.


Upon googling around I discovered this is some kind of badjuju curse with dozens of different solutions.


This is what I've done so far to try to fix it:

- rolled back the driver to the previous

- uninstalled all Nvidia components from Control Panel > Programs, reinstalled the original factory Nvidia driver

- ran Malwarebytes, UnhackMe and RegRun, all clean

- uninstalled Daemon Tools & sptd [the sptd.sys & spldr.sys kept coming back with code 24 after BSODs] This worked for about a day and a half

- uninstalled Realtek audio

- uninstalled Realtek NIC

- disabled Realtek audio and NIC

- disabled Windows Aero

- ran sfc/scannow twice, no errors

- ran chkdsk /r & chkdsk /b, no errors

- ran memtest86, no errors

- clean install Win 7 x64 and installed drivers in the order Intel lists, which is Win updates > chipset > graphics > matrix storage manager. I used the latest chipset driver but I used the old Nvidia driver that worked with Win 7 the first time I installed it over a year ago. This worked for a handful of days, but Windows was doing new things like hanging while loading Windows or at random intervals, giving no error and then going back to normal. Then the graphics went nuts again no matter what driver I used.

- put all drivers on an external HD, clean install Win 7, disabled updates and wireless during installation. My copy of Windows comes with SP1 but I thought maybe a new hotfix was causing a conflict. I installed the chipset driver that I used when I first installed Windows last year, reboot, installed the older Nvidia driver, and instantly got the same problem upon reboot.

- busted out my ancient recovery discs and did a factory restore [with Vista x64]. On the first boot, I got more display flickering, some black bars and nvlddmkm.sys would be restarted. If I let the computer idle it would go up to 10 mins before flickering, black bars and the driver would be restarted.





I noticed that since the first time I installed the updates, the computer was running warmer than usual when doing ordinary things like web surfing or watching a video, but the GPU never went above 80c. I remembered that when I ran a SystemRequirementsLab test for Diablo III after installing new updates but before I started having the problem, the test reported that the CPU was running @ 2.1 GHz instead of 1.83 GHz. I never overclocked anything before because I don't know anything about it and scared of frying my computer. I looked into overclocking programs to see if I could put it back to 1.83 but I couldn't make heads or tails of any it. All of this was before I reinstalled Windows. I dismantled the computer, cleaned the fan and applied AS5 to the GPU and CPU [left the thermal pad over the Northbridge, it didn't need replacing]. There were no obvious signs of heat damage around the GPU but I don't suppose that means much. Temps lowered.

After the factory restore failed to fix it, I tried to run a stress test on the GPU. FurMark installed but wouldn't even boot up in Vista. At this moment I'm reinstalling Win 7 because it's easier to work and has no bloatware. I will try to run FurMark again after that, but my hopes are low. I fear my GPU was destroyed by one of the updates but I need try everything because I just can't afford to get a new computer right now, and I don't think getting a new motherboard is worth it.






While googling around about this, I saw some other fixes that worked for some people that I want to try before giving up on this computer, but I don't know how to go about it. I read that sometimes the nvlddmkm.sys problem was caused by too much power and that could be fixed, but I don't know how to do that. There is something funky going on with my power all of a sudden because the battery won't charge for more than 15 seconds at a time, I first noticed it a couple of days ago and I don't think it's the battery. 95% of the time I run the notebook with AC and no battery to conserve the battery's lifespan.

I don't have any available options in BIOS other than change boot order, memory and hard drive tests.

There is something else that I suspect may be related - sometimes when using a USB mouse it be slow and skip around for several minutes and then go back to normal, even after a clean install of Win 7 and after factory restore. The touchpad would work fine and I always turn off when using a mouse so there isn't a conflict between the devices. This started happening around the same time the BSODs started. I also used to get the those "this device could perform better" when using external hard drives and other USB things. This computer has 4 USB 2.0 ports and it didn't seem to matter which port is being used.





Specs

Intel Centrino Core 2 Duo @ 1.83 GHz
Mobile Intel 965 Express Chipset Family
4 GB DDR2 memory
GeForce 8600M GS
Intel PRO/wireless 4965AGN
Realtek Gigabit PCI-E 8168/8111 NIC or Realtek 8101 fast ethernet NIC NDIS 6.0 (can't remember and sometimes Windows Update would install different drivers)
Motorola SM56 data/fax modem
Ricoh 5 in 1 card reader
HD is 320 gb ACHI/SATA
Phoenix BIOS ver F.5A, 3.11M


I'll let you know how it goes with FurMark when I get it installed. Please, if anyone has more ideas, tell me.

 

[fantastik250]

Your mobile graphics could be failing.

 

[Duchovnik]

Tried something different and got a different result, I removed the old nvlddmkm.sys that wasn't being replaced when updating drivers or removed when uninstalled and installed 301.24 without PhysX, so now the new nvlddmkm.sys is in the system32/drivers folder. Then I would get a black screen or a freeze but only after trying to open programs like Speccy, Google Chrome, I can mess about in Windows and get into the device manager but not much else, if I let it idle nothing happens. A low res boot will let me open Speedfan and RivaTuner but not FurMark or Speccy.

 

[bignastyid]

The 8600m has solder issues(some manufactures did a recall but those ended long ago) what most likely happened is it got to hot when you ran systemrequirementslab and broke a solder connection. I have had good luck with reflowing. there are several youtube videos that explain the process http://www.youtube.com/watch?v=SEx6g-xt0CId

 

[Duchovnik]

GPU's toast now, I'm getting 0x116 instead of 0x16 and the dump file also lists two DirectX drivers in the BSOD on top of the nvlddmkm.sys. Generic VGA driver it is then, until I can replace the computer.

Thanks so much BigNasty for the tip, I will give it a shot as I don't have anything to lose at this point. I'm so disgusted with HP right now and feel so stupid for not getting the Alienware I had been looking at when deciding to buy my first laptop for gaming. $300 cheaper at the time, but not worth it in the end. At least it lasted almost 5 years under heavy, daily use, and I mean *heavy* use, I used to run FFXI all day and night for days on end [often in windowed mode] and when playing I often had iTunes running and 5-8 Firefox tabs open to FFWiki and FFXIAH. I'm already looking into Dell's payment plan for the M18x, even though I hate having to buy from Dell, I need a notebook with as many different hardware options as possible, I'm never buying a HP notebook again.

Besides the battery fiasco with dv9000s, the 8600M GS had some design issues. I remember being on the phone with HP less than 4 months after buying this notebook, talking to some Indian lady who almost bricked the computer. I sent it in and they claimed to have replaced the motherboard but I doubt it, they also said they replaced the speakers [which I specifically asked them to do and told them to charge me extra for it] and they did not.

The gpu's heatsink in the dv9830 is not connected to a fan, it just has an open vent at the end that is placed next to the fan that is attached to the gpu and northbridge's heatsink. It is just sad to look at.

Anyways I talk too much.

The Northbridge and CPU in the dv9830 aren't that close to the GPU as they are in that video, but I'd still worry about damaging them with the dryer, you think I could try to use some kind of little box around the GPU to protect the other die?

 

[Duchovnik]

I looked around and saw that some people use aluminum foil shaped like a bowl around the gpu, I'll try that.

My last questions before I try this are:

1) Should I remove the thermal paste before applying heat or leave it on while heating?

2) If it only works for a couple of days, could I try it again and use the dryer for 3-4 or 5 mins?

 

[bignastyid]

1) yes remove the thermal paste.
2) yes

 

[Duchovnik]

ZOMG I'm in shock - it worked!!

You're a facking genius for diagnosing the problem!

But instead of using the technique in that video, I did this:

http://petrogaz.com/blog.php?post=39

The hair dryer I borrowed only had low, high and no heat levels so what I did was wrap the motherboard in foil, use a toothpick to cut out a section directly over the GPU and slowly heated up the MB on low heat for 10 mins. Then used high heat over the GPU for 10 mins [I chickened out and didn't do the full 15 min] holding it 5 inches away at first, then slowly bringing it closer to the die every minute and the last five minutes I held it about half an inch away from the die, switching back to low over the whole MB for 5 mins then walking away to let it cool off.

While it was cooling I found a 1965 penny, cleaned it and flattened it with a hammer and a 2x4 over concrete till it was about the thickness of a copper shim. After the MB cooled, I put AS5 on the GPU, penny, tiny *tiny* bit more AS5 on the penny to cover the imperfections then heatsink, hoping the added pressure would help.

Hopefully this fix will last at least a month. The stress test has been running for 20 mins now with no hiccups, GPU's averaging 70C, which is the coolest it's ever been under such usage since the comp was brand new.


I'll post in a couple of days -> week if it's still working.

And THANK YOU again!

 

[Duchovnik]

Best answer selected by Duchovnik.

 

[Duchovnik]

Worked for 7 days. I am disappoint. I redid everything using two dryers this time for 15 mins over the gpu and it's working for now. Next time it fails I'm going to borrow a hot air gun from a friend of mine who works in construction. I won't be putting the MB in the oven though, lol.

It's a good thing I like taking things apart and tinkering with them or I would be very annoyed by all of this.

 

[Jon Bader]

Don't feel to bad about not getting the alienware. I have the m18x and getting GPU caused BSOD on a year old $3000 gaming laptop