Jump to content
Corsair Community

Desperately seeking solution to nvlddmkm


JMark1

Recommended Posts

Hiya! First time poster here. Unfortunately I wish it was under better circumstances. I'm looking for some help.

I've had this quad SLI system for about 2 years. Never overclocked it. Ran Windows MCE 2005 until this past April when I decided to upgrade to 4 x 1GBs of the Corsair Dominator 1066MHz RAM and Windows Vista Ultimate x64 in order to utilize all 4 gigs. Ran great until just recently. I upgraded to nVidia's latest WHQL certified driver in July, v175.19. Installed August's Microsoft critical updates. Not long after I began getting BSOD with memory dump when watching videos at YouTube.com. Would crash on random web pages if a flash applet was running. It always names nvlddmkm.sys as the cause. When Windows reboots, the self-diagnosis blames the nVidia graphics driver for the crash and that no solution is currently available. It then suggests manually updating the driver among other things.

I attempted to research the problem online and discovered it's a fairly common problem with a multitude of various fixes. None of which is a documented sure thing....

Booted into safe mode, uninstalled driver, ran driver cleaner v3, installed previous version / beta version etc. Crashed no matter what version I tried.

Next, I read that it could be that my new memory may not be configured properly. I went to Asus' website and updated my BIOS from v1701 to 2101 and went in and set my memory timings manually. 800MHz 5 5 5 15 2T 2.1v. The crashing stopped for approximately 3 days before starting again. Next, I read that the RAM wouldn't be stable unless I set them to 667MHz and overclocked the CPU or FSB clock. I've never done this before but am currently trying to research some articles to familiarize myself. Learned quite a bit but haven't attempted any settings yet.

In the meantime, after all the uninstalling and reinstalling of several different nVidia graphics drivers, my Windows Live OneCare began acting buggy and I decided to uninstall and reinstall it. The uninstallation and reinstallation were both unsuccessful. I suspected my Windows Installer of being corrupted. Rather than trying to repair what looked like an increasingly corrupted OS I decided to just take the plunge and reformat and reinstall Vista.

Back to fully updated SP1 clean install, showing 4 gigs RAM, installed latest nVidia beta driver 177.92. No more random crashes but haven't surfed to YouTube either. Aero and Dreamscene run without a hitch. Showing all 4 GPUs functioning. BUT...whenever I go into nVidia control panel and attempt to set quad-SLI, nvlddmkm.sys BSOD w/memory dump. It does this EVERY time. I always was able to set quad SLI when I ran WinXP MCE 2005. This rig ran good with quad SLI enabled in Vista and 4 gigs RAM from April to August. Is this a memory stability issue from untweaked settings, bad graphic driver or could I be experiencing a hardware fault? Bad PSU? Graphic card? RAM? Like I said, this problem is quickly becoming an epidemic online. Any help is desperately appreciated.

Link to comment
Share on other sites

  • Corsair Employees
With 4 modules I would suggest setting the memory frequency at DDR667 or DDR800 and set the memory Voltage to 2.1 Volts and set the NB/MCH/SPP Voltage to +.05 Volts as well and test the system with http://www.memtest.org. In addition, with some MB's (Mostly ASUS) you have to disable legacy USB in the bios when running any memory test.
Link to comment
Share on other sites

I tried what you said and set RAM to 667MHz 5-5-5-15-2T and my "DDR voltage" to 2.10v. I also manually set HT to 1.4v. System booted fine. Tried going into nVidia control panel and setting quad-SLI and got nvlddmkm.sys BSOD w/memory dump again. Only this time system was stuck in endless BSOD reboot loop. Had to boot to safe mode, uninstall graphics driver, ran Driver Cleaner 3, reinstalled with nVidia's latest WHQL driver, 175.19. System booted normally again.

I am fairly unfamiliar with OCing so I am unsure exactly where you want me to bump the voltage. I see a few NB settings....which one? Anyhow, I disabled Legacy USB and ran MemTest with all 4 sticks installed for 6 passes with no errors.

Am I getting a "bottleneck" or could some other piece of hardware be bad in my system. The RAM seem fine. I suspect the settings but am unfamiliar with tweaking BIOS settings. I have been searching for any reference materials related to my hardware and OCing. Working hard to solve this but no lightbulb above my head as of yet.

Link to comment
Share on other sites

Well I finally taught myself enough to attempt some OCing. Bumped up the multiplier here:

 

http://i453.photobucket.com/albums/qq252/JMark1/cpu-3104.jpghttp://i453.photobucket.com/albums/qq252/JMark1/memory-3104.jpg

 

Began getting random nvlddmkm.sys BSOD crashes. I decided to try lowering the multiplier and raising the HTT.

 

http://i453.photobucket.com/albums/qq252/JMark1/cpu-3102.jpghttp://i453.photobucket.com/albums/qq252/JMark1/memory-3101.jpg

 

Still got random nvlddmkm.sys BSOD crash. Another thing is, when it recovered from the crash, my colors changed. Screen looks too bright and chromy. Looked in nVidia control panel and see that it's automatically bumping digital vibrance all the way up to 100%. Does it each time I boot even if I manually tweak colors in control panel and save. Could driver be corrupted from crash? I bumped HT voltages to 1.5v, memory to 2.150v and CPU to 1.5125v and no random crashes so far. If I try to enable quad-SLI, as always, nvlddmkm.sys BSOD crash. I put in a trouble ticket with xFx and am waiting to hear what they say. I've read that the BIOS I upgraded to may be causing this. Apparantly, 2101 is aimed specifically at quad-core Phenoms. Some recommend downgrading to 1201 which was last BIOS exclusively geared towards AM2 FX chips prior to Phenoms. Some say that isn't necessary provided you set your settings correctly.

Also reading that Vista has some stability issues concerning 4 memory modules as opposed to 2. Even with the patch fix. Suffice to say, this memory may be fine but they may not have all the bugs worked out in Vista yet. With that said, I'm wondering if anyone believes I would have been better off with 2 x 2GB modules of this stuff opposed to the 4 x 1GB stuff I currently have? Any feedback is greatly appreciated.

Link to comment
Share on other sites

Hmmm...I wish I would've known about this issue prior to buying. I didn't realize and thought nothing of just buying some good RAM. So it's either limitations of this hardware or the software they write to overcome these bugs. Seems like a lot of little bugs together make for a big pain in the neck. I've seen people go as far as writing Vista installation tutorials in which you strip all your RAM and peripherals from the system until after the OS is completely installed. I have to admit, I'm considering taking 2 modules out and just running 2 until I can buy some 2GB modules. Maybe I'll try that bare-bones OS install and the BIOS downgrade.

Fixed that color issue by uninstalling and reinstalling driver 175.19. Still can't enable SLI but I'll see what xFx has to say. You get problems like these and it makes you think. I'd do just about anything to get this rig to start working flawless again. Considering ditching the 7950s and upgrading to one good card. The latest is GeForce 9800 GX2, right?

Link to comment
Share on other sites

OK...when I run Prime95 blend torture test (which states lots of memory is tested), I fail in the same spot of the 2nd test each time. The documentation that comes with the program states that this is an indicator of a hardware problem.

 

FATAL ERROR: Rounding was 0.5, expected less than 0.4

Hardware failure detected, consult stress.txt file.

 

Could this be a false positive due to overclocking or could my RAM be bad. I successfully completed 6 passes with no errors with Memtest prior to overclocking. I'll run memtest again to be sure but I'm assuming bad memory could definitely keep you from running in SLI. Is there another way I could isolate this problem further.

 

xFx responded with this:

[MICHAELC 9/15/2008 9:38:41 PM] Hello James; thanks for the message, does the problem only happen when you have Quad SLI enabled? Please test each card in the system one at a time just to verify that they are each working properly on their own. What OS were you running previous to Vista 64? Based on the information that Nvidia has given us about the nature of the Nvlddmkm error is that this issue really could be caused by nearly anything, and the system only reports the fault with a graphics card because that is the last device accessed at the time of the failure, and this can be misleading. When did you first see this problem appear? If possible, try testing the card into another computer, perhaps one not running Windows Vista, and see if any of the same symptoms appear. Here’s a site which may offer some help into researching what exactly causes the Nvlddmkm error messages http://www.microsoft.com/whdc/device/display/wddm_timeout.mspx or from Nvidia’s knowledgebase http://nvidia.custhelp.com/cgi-bin/nvidia.cfg/php/enduser/std_adp.php?p_faqid=22 If you have your card or even your system overclocked, try and back off a bit and see what happens, as this may reveal other devices such as a stressed motherboard Northbridge chip or RAM causing this issue even when the video card itself is fine. We highly recommend Driver Cleaner, found at http://downloads.guru3d.com/download.php?det=745 which will come in handy for cleaning out traces of older graphics drivers. (Make sure you have the latest drivers, from http://www.nvidia.com/download/index.aspx downloaded to your desktop.) Download the program, install it, but don’t run it until you’ve rebooted into Safe Mode (tap F8 during bootup) and then choose “NVIDIA” as the cleaning selection. Run that 3 – 4 times, to be thorough, and then you’ll reboot into normal Windows mode in order to reinstall drivers. In case your XFX product will ever need warranty service in the future, please confirm that your customer profile is updated with your full shipping address, phone number, zip code. After logging into your XFX account online, click Account Information to the left hand side of the site, and fill in your information and when done, you must click the Update button directly beneath the address fields. Regards, Michael XFX greatly values customer feedback so if you have a moment, please rate your ticket. This option can be found on the right hand side of the page of the ticket window, under the RATE THIS TICKET heading. Click on the corresponding amount of stars to match your satisfaction.

 

I'll update status as soon as I've had chance to do some more troubleshooting.:(:

Link to comment
Share on other sites

The second set of screenshots above is my current overclock. I lowered my multiplier from 14 to 11 and bumped the frequency from 200 to 282. I set the CPU voltage to 1.5125v. RAM settings are 667MHz 5-5-5-15-2t. Set voltage to 2.150v. My hardware monitor says my CPU is steady at 49 degrees celcius and motherboard is 43 degrees. Running 1176 x 664 resolution and highest color quality. 3D settings in nVidia control panel set to quality. Like i stated previously, these problems began mid-August not long after upgrading graphics driver and Microsoft's August critical updates. Since then I have tried numerous fixes including reformat and Vista reinstall. Problem reamains. I could always enable quad-SLI and now doing so crashes my system every time.

Only thing I changed is upgraded from 2 x 1GB of a competitor's(rhymes with style) RAM to 4 x 1GB Corsair Dominator RAM and WinXP MCE 2005 to Vista Ultimate x64 in April. This computer ran beautiful with these upgrades for a[pproximately 5 months before these problems.

Link to comment
Share on other sites

Ok well I'll attempt that among other things when I resume troubleshooting but I would just like to explain that during my research I've uncovered countless claims in which people have supposedly accomplished overclocks between 3.2MHz to 3.6MHz with same or similar hardware. My overclock is fairly mild in comparison. Stock speed is 2.8MHz with FX-62. 3.1MHz I found was routinely considered to be a minimum overclock with similar hardware. Not claiming to be an expert, mind you.

My system crashed prior to overclocking also. Only reason I overclocked was to see if it could be a solution to my problem. Heck, perhaps I should go back to default settings and run Prime95 then....hmmmm that would be a good troubleshooting indicator. I'll let you know.

Link to comment
Share on other sites

  • Corsair Employees
Test the modules with out over clocking the system or only as needed to test them one at at a time running the modules at their tested settings to be sure one is not failing, and with over clocking it will be hit and miss and running 4 modules will always have an impact on how well the system will over clock.
Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...