graphic driver error actually memory instability?

[jmark1]

Greetings. First time poster here.

First, let me explain what I recently changed. I recently added 4 sticks of Corsair Dominator CM2X1024-8500C5D RAM and upgraded from Windows XP MCE 2005 to Windows Vista Ultimate x64 to utilize all 4 gigs of RAM. Been running great since april. Updated to the most current nVidia WHQL graphics driver 175.19 in July, no problems. Installed August's Microsoft critical updates, no problems. Until, that is, after running some flash content from YouTube...I began getting BSOD with memory dump. I could make this happen every time by playing flash content....even flash applets running on random web pages would make me crash intermittently. After each crash Windows diagnosed the problem as caused by an nVidia graphics driver. Began researching solutions and found several people recommended that uninstalling the driver in safe mode, running a driver cleaner in safe mode, and reinstalling the driver in safe mode would fix the problem. System still crashed with nvlddmkm.sys as the culprit. I began to suspect my new RAM and found some forums discussing adjusting FSB and PCI voltages as solutions to this problem. This made me curious as to the latest version of BIOS out for my mobo. After all, they do suggest only upgrading your BIOS when you are experiencing problems. I flashed my BIOS from 1701 to 2101 and lo and behold it seemed I had solved my problems. The fix lasted approximately 3 days. This led me to investigate the specific BIOS settings as I read that I may need to adjust several of my memory settings manually to make my system stable. I've found this BIOS has several intricate settings to effectively overclock your system. I've never attempted to overclock my system even though I certainly have a system which you could overclock. Although I would be happy just to get my system stable, it appears I am entering a territory which I may as well overclock while I'm there. Which brings me to here. After scanning the Asus forums I discovered that I must first load the default settings prior to changing my memory timings. Only then would it allow me to save my changes so the computer POSTs on reboot. After that, I found a setting "enable SLI memory." CMOS needs to be reset every time I try to enable that. When I try to enable SLI within the nVidia control panel, i get nvlddmkm.sys BSOD crash. I'm thinking I need to make some veeery specific changes akin to overclocking specifics. I'll have you know I am still currently researching these things diligently but any help at all would be most appreciated.

 

[jmark1]

I have derived from Corsair's forums that I must indeed drop my memory divider to 667 MHz and overclock the CPU in my AMD-based system in order to effectively utilize these 4 gigs of RAM. I was able to set my memory timings and actually successfully overclocked the CPU by 280 MHz by using the AI overclock setting in the BIOS. This showed my CPU running at 3080 MHz and only raised my CPU temp from 37 to 38 degrees. Cannot manually overclock however. Any manual settings I try fail to POST upon reboot. Also STILL cannot select "enable SLI memory" in BIOS...even with successful overclock. Fails to POST and must reset CMOS. And last, but not least, I still cannot enable quad SLI within nVidia control panel without nvlddmkm.sys BSOD w/memory dump. Has anyone out there gotten this to work? Please post some settings if so.

 

[jmark1]

Well, seeing as no one is responding, I'll just try to keep anyone who cares updated as to my progress. If I do find a solution, I'll be sure to post it in order to help anyone else who might have this same problem.
OK...been studying up on overclocking. That AI overclock in BIOS is not your atypical overclock. It shows in BIOS as being 3080MHz but when I measure using CPU-z it says actual frequency is 2813MHz. Not much of anything actually. Anyhow, I found resource which details OCing basics for my particular mobo. Apparantly, default voltages need to be locked in on several settings as changes you make may affect these other voltages. Therefore, knowledge of all your system bus voltages is required. You cannot simply leave certain settings on AUTO to accomplish this. In fact, voltage settings need to be thoroughly researched and planned out or else the system may not boot upon saving the settings. Then the CMOS must be cleared by removing the battery on the motherboard (I've already done this several times). If the OC settings you've planned work then the system SHOULD boot. Like I said, I'm still studying this. I won't try any settings again unless someone who knows a lot more about OCing than me responds to this and suggests some settings based upon my system specs or until I educate myself enough and I plan out some good settings worth trying. Whichever comes first.
As far as nvlddmkm error, I've been to Corsair support forum and RAM GUY said,
"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 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."
Well, I know NB stands for North bridge of my CPU, MCH for Memory Controller Hub, and SPP for System Platform Processor. I do believe these are all various terms which refer to the same thing on different systems or BIOS settings. I'm not exactly sure how this applies to my specific BIOS as yet but I will check it out.
I have reformatted and reinstalled Vista since last post. Updated to SP1. Showing all 4 gigs of RAM. This seems to have eliminated random crashes. If I attempt to enable quad-SLI in nVidia Control Panel, I get nvlddmkm.sys BSOD w/memory dump every time.
So I did attempt to apply some of RAM GUY's suggestions and manually set my memory frequency to 667MHz and set my timings 5-5-5-15-2T and set my "DDR voltage" to 2.10v. System booted fine. I decided to be brave and went into nVidia Control Panel and set system to quad-sli. Got nvlddmkm.sys BSOD w/memory dump again. Only this time upon reboot system went into endless BSOD reboot loop. Alleviated this by booting into safe mode, uninstalling graphics driver, running driver cleaner 3, reinstalling latest nVidia WHQL driver 175.19. Was able to boot normally again. Mind you, this happened after recent clean install of Vista Ultimate x64. Absolutely dumbfounded. I've still got to test my memory with MemTest86. I did a pair of the RAM last night for 8 passes with no errors but will try all four sticks tonight. Also am gonna try booting with one graphic card at a time installed to try and isolate any problems with my hardware. Got another lil tool, GPU-z, but am unsure what would be considered normal settings. It sees all 4 GPUs. They running hot but what is TOO hot?
Read something about nVidia graphics drivers could possibly have corrupted my graphics card's BIOS. Is this possible? If so, how to fix?
In case you haven't noticed, I'm kinda obsessed with fixing this. Been at this for few weeks now. No luck so far. Any help greatly appreciated.

 

[jmark1]

Bumped HT voltages to 1.4v and disabled Legacy USB. Ran MemTest with all 4 sticks installed for 6 passes with no errors. I'm thinking the RAM is fine and this is either a BIOS settings system stability issue or some other hardware has gone bad. Graphic card? Mobo? PSU? Still looking...

 

[jmark1]

OK...so I finally taught myself to overclock the system. I do believe I successfully OCed the CPU here:

I began getting random nvlddmkm.sys BSOD crashes however. I decided to try knocking the multiplier down and OCing the HTT instead.

Still got a random nvlddmkm.sys BSOD so I went in BIOS and bumped up memory voltage to 2.150v and CPU voltage to 1.5125v and I haven't crashed yet.

One of crashes put me in BSOD reboot loop so had to boot to safe mode, uninstall graphics driver, reinstalled new 177.98 driver and new nForce chipset driver. Everything seemed fine until another instability crash and now my color scheme automatically changes itself. When I look in nVidia control panel, digital vibrance is set all the way to 100%. Unsure why this is happening. May need to reinstall driver again as random nvlddmkm.sys crashes seem to corrupt driver. Like I said, since OCing HTT and bumping up voltages I haven't had random crash yet. If anyone has any advice, please feel free.

 

[jmark1]

Beyond frustrated here. This is looking more and more like a Vista-driver incompatibility issue. Anyone ever suspect a company of writing poor drivers in order to move more hardware?