GTX 560 Ti 2GB - Possible driver problem and crashing on Win7 64bit

Good morning guys

Yesterday I did a fresh install of a brand new computer and it was working fine. All of a sudden (after ~5 hours of use), my system froze and the video went all crazy (actual photo from my monitor):



In this first freeze, the computer did not respond to anything for like 2 minutes, and then I rebooted it. The POST and the EFI display fine, then in the Welcome Screen the system crashes again, but now after 1 min it restarts.

To isolate the problem source, I ran a series of tests:

• Ran the computer on Safe Mode. It ran nicely.

• Then I turned iGPU on and plugged my computer to the onboard port. Ran nicely again (although the max res I could get is 1280x1024, which looks horrible on my 1600x1050 monitor).

•Then I uninstalled all the Nvidia drivers and let Windows install its default drivers for my GPU. It ran nicely once again.

This makes me think the problem is exclusively driver-related (i had installed Nvidia drivers 266.58, 275.33 and 285.62 and none worked). Funny is, even with the drivers installed EVGA Precision says it has detected no hardware and Virtu says it can't turn on.

My system configuration is this:

EVGA GTX560Ti 2Gb
Windows 7 Ultimate 64bit
Asus P8Z68-V Pro
Samsung SyncMaster 2233
Intel i7 2600k @4.0Ghz
16Gb Corsair Vengeance 1600Mhz
WD Caviar Green 2Tb
Corsair Force 3 90GB
Corsair HX650
Thermaltake Dokker Lite
3x Aerocool Shark Blue
2x Cooler Master Excalibur
Zalman CNPS 9900MAX

Although I had overclocked my CPU, I never touched my video clocks. System and drivers (including the Nvidia ones) are installed on my SSD and everything else on the HDD.

Any clue on what's going on and what can I do to solve this thing?

Have you tried resetting the CMOS and running it without the OC on the CPU?

I indeed reverted my CPU overclock to default settings to no gain, and clearing the CMOS made my VGA POST LED turn on. Now my card is rendered totally useless for it won't display even the EFI.

Thanks for the help. Anyone else has a guess on what's going on?

One more update to the problem:

After uninstalling all the drivers, I tried using the computer thru Virtu. CPU-Z now recognizes the 560Ti clock and memory as being from the Graphics 3000, and vice versa.

is there any way to be sure everything is working fine now?

Any help, guys?

One picture is worth 1000 threads:



The computer is running fine now, with the monitor plugged into my iGPU port and still running Virtu. I haven't taxed the graphics out yet to see if I am really using my GTX or not.

Updates on the problem once again.

After formatting, clean installing everything, reverting drivers, downclocking the card and everything, nothing worked on definitive.

Sometimes it worked right away, sometimes after a single reboot, but often i had to change back to the iGPU. Uninstalling Realtek Audio drivers made it work for a while, then disabling the Driver Signing worked for a while, then changing the TDR parameters on the registry also worked for a couple of boots but in the end it was all bad again.

Running Cinebench with Virtu enabled gave me a 13 fps mark which is pretty bad, and checking the Virtu Control Panel one can see it does not support Cinebench and not any application I use (CAD). Did not run Tropics though.

So, I decided to move the board from the PCIE 2.0 x16 (blue) to the x8 (grey) one.



To my surprise it's working a charm! I rebooted like 20 times and did not have a single hang yet. Here's prints from GPU-Z, taken right now as I'm writing this post - do not mind the 32°C idle temperature, I live in Brazil and the room temperature is now ~26°C:



As its running @x8, I decided to run Cinebench and 3dMark (both 11 and Vantage) to see the card performance under load.

At Cinebench I got a stable (10 times) 63.77 FPS in that car chase scene and the temps never went higher than 42°C (I've changed the fan settings on Precision to make it respond almost linearly to the workload).

At 3dMark 11, which ran on the Benchmark only/Performance Settings/Centered (my monitor is a 1050p one), I got a P4718 score. You might see on the linked report that my CPU has been OCed to 4.4Ghz. Temps topped at 57°C at 75% fan.

With 3dMark Vantage, Performance Preset, i got a P19796 score and temperatures topped at 61°C @75% fan.

3d Vision also works fine, though I haven't run any bench on that.

As everything appeared to be normal with those three tests, I tried plugging the card back to the main PCIE blue slot. Once again everything crashed. After switching back to the x8 slot, everything is working properly and the benchmark results were pretty much the same.

Can I say everything is working as it should right now? And what the hell might be causing the crash on the x16 slot? Maybe I should RMA the board, which is brand new and with the newest stuff installed.

Thank you all for the responses!

Thanks for the responses. But would I see any performance gain from plugging the board to the x16 slot? Because as it seems, everything is running good right now.

Fine then malmental. I'll contact the vendor and see if I can have a replacement soon - when the new board arrives I post news on the problem here.

I did, to no help.

Once again I thank you for your prompt help. About your checklist, lets see:

1) The board was brand out of the box when it started giving trouble, but will try that.

2) Will try that. Will lend a computer from a friend and see if it works, both with the video card on another MB and another video in my MB.

5) I reformatted the computer a lot of times, installing the drivers and other utilities in a myriad of combinations. If I install the Nvidia drivers first it will work... until I decide to install anything else. Never tried installing the video drivers and rebooting the computer 20 times for stability though.

About the BIOS options, here is a page with pretty much everything the UEFI has to offer us: [url=http://www.legitreviews.com/article/1615/4/][/url]

interesting, what nVidia drivers are you installing.?

well then let's rule out drivers and deal with hardware..

More confusion thrown on the issue.

I've plugged in a Quadro FX380 into the PCIE x16 slot. And it worked damn well - without even needing to uninstall the same drivers that make my computer crash.

My Windows now crashes with a black screen, then switches to the pink screen mentioned in the first post, then shows a black screen with a very dim Start Menu button and the clock, then a blue screen with white artifacts, then a black screen with white artifacts. I'll post a video of it if asked. If I reboot the computer a considerable number of times, though, I can get it to work - but if I try to run 3dMark for example it crashes again to a black screen.

I've been praying for a BSOD (wait, who the hell prays for a BSOD?) to show me something to begin my investigation with. After a number of trials, i finally got this:



I've tried a number of TDR registry modifications, from increasing the timeout to changing the level. Nothing works.

Please, any light on that?

OK.
can we try something once more or a little drastic before you start sending back hardware.?

remove all drives but leave the SSD.
then reformat and install clean win 7 64-bit.
make sure you wipe SSD clean before install.
install 560 Ti in the first x16 slot and then proceed with install of OS.
make sure you have an active internet connection..

dude, my bad..
you did say that, I mis-read..

First thing I ever tried is Driver Sweeper malmental, too bad it does not work either.

The Quadro FX380 I plugged in worked really well, not a single issue in the whole day. Also using the 560Ti in the Quadro's motherboard (which is a P5KPL/1600) worked fine too.

My mind is warping trying to figure out what can cause the problem:

• Drivers: tried installing every version, letting Windows update do the work (it installs the 275.33 version) and everything but no driver work. Standard VGA Adapter works though.

• BIOS and OCing: resetting the CMOS does not help. Running everything on factory clocks does not either.

• Corrputed system/registry: nope. Everything is brand new and out of the box, including Windows. Tried hacking the TDR parameters on the registry to no help too.

• Bad memory: I don't think so. Ran memtest for a couple hours and did not have a single error.

• Heating: nope. Both idle and load temps are within normal thresholds.

• Bad Power: Corsair HX650. Nuff said.

• Bad PCI slot: don't think so. Another card worked well there.

The only thing that solved the problem so far is changing the PCI slot to the x8 one. But as malmental pointed out, it's paying for nothing if I leave it like that.

I'm feeling kinda like this (think my computer is that f*cking door):



Yeah, will do. Thanks for the help, I hope your hair is intact after that thread.

this making me think it is a power problem

either with the PCIE x 16 slot supplied 75 watt
or your PCIE 6 pins

your 560Ti should be using two 6 pin power connectors

when you plugged in the Quadro how many 6 pin connectors did it use?

because if it only used one which I think most quadros do

but if 560TI worked with both 6 pins in the x 8 then

maybe that is not it
(sorry brainstorming here)

maybe your PCIE x 16 slot is not supplying power properly

I like Mals idea

clean edge connector of 560TI with strong alcohol (90 percent Isopropyl only)
also clean out the PCIe 16 slot

I think that the power pins on the edge connector of 560Ti is not making proper contact with the PCIe 16 slot

really clean the 560Ti edge connector first with pencil eraser then with iso alcohol

just a thought before returning everything for RMA

Did you breadboard you system before setting it up? I know drivers could be an issue, but hardware is pretty sensitive anything can go wrong.

1. Boot up with just your CPU, heatsink, MoBo, and PSU. If it beeps, you're good. If not, it's probably a bad MoBo (or is it PSU?).

2. Add the RAM. If it beeps, you're good. If not, bad RAM.

3. Add the GPU. If it beeps, you're good. If not, bad GPU.

4. Eventually add the other pieces of hardware and boot up. If any issues occur at this point that do not allow for a successful boot, it's either a bad PSU or MoBo.

If the issue cannot be solved, I would start all over and RMA everything. I know it sucks delaying your build for a week or so, but it's better than the frustration and headaches you are going through.

It's basically a blueprint of your hardware before you place it in your case. It allows you to determine any hardware issues or to just test if your system will work.

google your error from your BSOD screenshot and this is what I came up with

quote

I have found and fixed the problem today. This is what appears to happen.
during the installation of the most current drivers 100.65 Vista, an OLD file
nvlddmkm.sys is copied into windows/system32/drivers and not the current one
in the install. As a result the new drivers are attempting to access a file
dated 11/2006 instead of 2/2007 ver 7.15.11.0065 which is in the newest WHQL
driver ver 100.65 vista 32.

Fix: Go to windows/system32/drivers and rename nvlddmkm.sys to
nvlddmkm.sys.old. Go to the nvidia directory and find the file nvlddmkm.sy_
and copy it to windows/system32. Using the cmd window (DOS box) type
EXPAND.EXE nvlddmkm.sy_ nvlddmkm.sys. When the expansion is complete, copy
the new nvlddmkm.sys to windows/system32/drivers and restart the computer.

Your computer should now work properly.

You will notice that any uninstall and reinstall of nvidia drivers will not
remove the old nvlddmkm.sys file and will not overwrite it with the newer
version. You have to do it manually. I do not know why this happens but who
cares as long it is fixed


that is from Vista
but should be similar procedure for Win7

your error was nVidia BSOD nvlddmkm.sys error
which this refers too

soure - http://forums.techarena.in/vista-h [...] 689444.htm


google search page link - http://www.google.com/#sclient=psy [...] 25&bih=847

did you try the fix?

also this which I have read a few threads stating changing ram helped
even though the ram passed Memtest
it was the RAM causing the error

quote


Well, the RMA guy tested the GPU on their test machine and it worked fine, through various tests - no issues whatsoever. They ran memtest on my RAM and again no errors. They tried another vid card on my machine and same nvlddmkm errors. They pulled out my RAM and tested it on their test machine - no issues but as soon as they put that same RAM back into my machine - the nvlddmkm errors came back. Eventually they just tried another set of RAM sticks, same manuf but slightly better model and no more nvlddmkm errors!! They tested extensively for around 3 days with various tests and no issues whatsoever.

So I've just had my system returned back yesterday and it's been running fine, not a single error (so far) so I'm smiling again - fingers crossed.

Sum up:
I thought my vid card was dying due to the artifacts and inability to run anything remotely 3D but I was wrong! It seems this nvlddmkm error is a very general error and causes symptoms that point the finger at drivers, OS and other hardware! There seems to be compatibility issues between RAM-mobo-GPU that is not well understood (at least by me).

source - http://forums.techguy.org/windows- [...] bsods.html

found a few threads where changing memory sticks solved the problem

I would try the other power on fix first
then try changing RAM if possible

Boy, when I saw this my eyes started shining again. I tested it and... did not work!

Also the memory thing did not work also, I popped in a stick of Corsair Dominator (only 'same manuf but slightly better model' i could think of) and same thing.

Now looking at my BIOS monitor I came across to this:



Can those voltages be any trouble? Also I don't know how to change them, the values shown are read-only.

Thank you for the help.

Yes, a single stick. The RAM timings are correct in the BIOS, so I'm almost giving up on the issue. Will contact both ASUS and EVGA though.