Windows 10: Nvidia driver always crashes once on startup and stops after.

About a week ago, i had some strange nvidia driver issues after sending my desktop to sleep which involved a no signal black screen into a system freeze after login.

Since that point I now have a slightly different issue where the first game i play no matter what will cause my nvidia drivers to stop responding and recover within seconds of the game starting, but it will not crash even after spending hours gaming after the initial crash. The event manager just lists a nvidia driver has crashed and recovered and nothing else that seems significant.

I figured that a clean nvidia driver update might fix things as i haven't bothered to update in about 2 months after finally getting what i thought was a stable configuration, but that did not seem to resolve anything at all.

Update: Something different today, after a botched initial startup where the pc started with no graphics signal at all, nvidia driver crashed more than once, which may be problematic.

Tried a DDU uninstall and reinstall of the same driver, which is currently the most recent one, but that did not work. Seems as of right now i can't actually get a game to run without the nvidia driver crashing within a minute now.

:)

 

Windows 10 startup freezing

It's fixed! Thanks for everybody's help!

Just to let you know what I did for anybody reading this.

It did turn out it was my graphics driver causing the issue.

Since I couldn't get into safe mode, I went into the boot options via F2 on startup and on the boot tab I disabled my Nvidia Graphics card.

This stopped the crashing on startup but resulted in a black screen with a cursor.

Once at this stage, I rebooted into the troubleshooting screen where finally the "Reset my PC" button FINALLY worked, rather than crashing everytime I tried!

So relieved, kept all my pictures too!

Thanks!

 

Computer (Windows 10) is BSOD everyday ....

The Event Viewer showed at least five different stop errors such as ATTEMPTED_EXECUTE_OF_NOEXECUTE_MEMORY and SYSTEM_THREAD_EXCEPTION_NOT_HANDLED.

This may indicate a hardware issue.

There was this error in the Event Viewer:

Windows failed fast startup with error status 0xC00000D4

Although this error appeared only once I might suggest to try disabling Fast Startup and see if it helps:

Turn On or Off Fast Startup in Windows 10

The memory.dmp file showed involvement of the nvlddmkm.sys in the crash which may indicate an issue with the graphics.

You could try updating the NVIDIA graphic driver to the most recent version.

 

Hi BobLeSaget

If you have restore points running you could try a System Restore. I would recommend having bootable media on hand should you need it. System Restore is a standard procedure but I like to be safe rather than sorry. To create media you could type Recovery drive in cortana, click on app and follow process. You likely need an 8GB key. (It will delete data on key.)

I would also look at running DDU again and installing an older driver. Newest is not always right.

I would also recommend you look at a product like Macrium Reflect.

Please check out Macrium Reflect as an Imaging tool. It is recommended and used by members on this forum. It is free, reliable and a fast way to restore to a working system.

Macrium Reflect - Backup and Restore

Post back results please.


Ken

 

I'll DDU a older driver first to start, although i am uncertain about many things as the problems had started with a older driver that i had previously DDU installed before due to other related problems. Unfortunately i don't really have any recent restore points for this pc. Nothing in particular changed between the time when it was relatively stable to current instability except that i let it sleep for a couple hours.

Any idea if power supply problems could be related? I read on a few other threads that other people with similar issues found their problems to be a under performing power supply. I have a spare power supply with a greater Wattage 750w vs 500w but i've just been hesitant to take this computer apart as it is a custom case that is somewhat hard to disassemble/reassemble.

 

Bob that is one of the great things about this technology, it could be anything. That said it sounds a little strange as you said you can game for hours. Is it not be responsive enough when the load is demanded, I buy that.

I would use this power calculator to get an estimation of real need. Even if your 500w has lots of head room if you have a power supply I would bite the bullet and install. Your current rig is less than optimal.

To get an approximation on total power required use this calculator. It asks for very specific info on CPU, memory type and size etc.

Cooler Master: Power Supply Calculator


One thing you may also want to try is the Driver Verifier from the tutorials. Maybe it will drive out something.

This tutorial will guide you through the verification process, heed the warnings. Assuming it fails follow the posting instruction in the BSOD section of forum.
Driver Verifier - Enable and Disable in Windows 10

 

I used the calculator and it recommended a 410W power supply. I might just be grasping at straws but what i feel like might be a possibility is this power supply is not outputting enough power until it's warmed up to a certain degree. I'll also look into the driver verifier first.as well.

 

So i decided to test letting my computer idle on desktop for 20m and warm up after starting it before i start up a game/anything graphically intensive, out of just one attempt so far, no driver crashes after reliably getting one for the past week or so.

 

Excellent.. A few more tests to confirm and you're rolling.

 

I have been running an Invidia 660T which started intermittent crashes after installing windows 10 fast ring.
I tried new drivers new power supply and all troubleshooters. Nothing helped. I also run multiple monitors. I'm now on 15031 version and seems to be getting better so I believe it's the OS.

 

Tested it again after about 20m or so of waiting, maybe it's because i have a laptop under load near it, somewhat different outcome. Had a driver crash a minute or so in idle on desktop,the rest of the computer seemed to be working and it took about 10s or so for it to recover. Regardless of the wait i still got a nvidia driver crash about 2-3m into the game test without any particular signs up until it just crashed.

Yet at the same time, it's similar in that after that one ingame crash no further crashing occurred. I wonder if it's possible that it's related to power draw from the socket. I had a wall heater on during one of the moments when the crashing was at it's worst, which was also paired with a strange phenomenon ingame where my gpu was getting very heavily throttled until about 30s or so after i turned off the heat.

 

True story. I had a customer call me because their PC would reboot without notice so they were concerned.

I went to their house and started checking things out and while I was there the lights dimmed, like when something turns on. I asked the owner did they notice that, sure they said, it is when the honey pot turns on. I said what?

Sure enough they have this ten gallon honey pot in the house as they sell honey. To make it run out the bottom tap there is a 1300 watt heater in the tank. Every time the tank cooled, the heater would kick in, the amperage would drop on the circuit and the PC would reboot.

I fixed by installing a good size UPS and attached PC, monitor and modem router to it. The PC doesn't see the dips anymore so no more issues. Now they were supposed to get an electrician to add a "Honey Pot" circuit but of course that never happen.

So your heater story hits home. I bet it is over 1000 watts.

 

I think a reasonable next test would be to see if my desktop starts up properly if i were to not have anything else on the same circuit using power, including lights and etc...

Either that or i can move the computer to a different room and see if it acts any differently.

 

Have a look at Reliability History and see if there are any red flags.If so,then view the Tech Details,which will give you a fairly comprehensive list of data,the most important found at the bottom.A common one is nvlddmkm.sys.

 

There really aren't any things i see of other than the nvlddmkm.sys stopped responding and has recovered message often 2-3 times when my driver has crashed. That being said, thinking about the possibility of power issues i've been trying an alternative to the wait time, which is to use EVGA precision to reduce the power target to about 90-95% of the normal stock clocks. So far it's worked for about two startups. What this means, i am uncertain.

Well about 4/4 on the stock clock reduction preventing driver crash. I've also noticed that my screen no longer flickers for a few seconds after starting up the pc while on the login screen.