GPU Crash (No Signal)

[pim16aap2]

Hey,

 

I've had my current pc for about 3 years now, but about a week ago I ran into an issue while playing BF1.

Half-way through a match, my monitors suddenly went black followed by a "no signal" message, while my GPU fans started spinning at what sounded like 100% for a few seconds. I could still hear the game and verified Windows was still running fine by launching Spotify and starting a track using my keyboard.

 

After trying the usual stuff such as downgrading to the latest driver I knew worked perfectly fine (and one version below that in case I was mistaken) while using DDU in safe mode to get rid of the current drivers, resetting overclocks, clean my pc, adding 10mV and adding 20mV to Core Voltage for the GPU, the crashes still persisted.

 

I have been trying to figure out what happened and why it did ever since.

It takes anywhere from 10 minutes to 3 hours for it to crash (though usually around 2-3 hours). It seems like the chance to crash is a lot higher when playing a few rounds first, stop the game for a bit and then play again, but that might just be a coincidence.

 

These are the specs of my computer:

CPU: I7 4770k @ 4.2GHz

Motherboard: Asus ROG Maximus VII Hero

GPU: MSI GTX 780 Ti Gaming 3G

RAM: 16GB G.Skill TridentX F3-2133C9D

PSU: Corsair AX860

SSD0: Samsung 960 EVO (500GB) (Boot drive)

SSD1: Samsung 840 EVO (250GB)

SSD2: Samsung 840 EVO (250GB)

HDD0: Seagate Barracuda (3TB)

HDD1: Some old 1tb drive

 

First, I tried another game, 7 Days To Die, which resulted in a crash as well, so it doesn't appear to be game-specific.

I then checked Windows Event Viewer, but that was entirely unhelpful, as there were no messages in the ~30 minutes leading up to the crash.

Then I monitored my temperatures while trying to get it to crash again (read: play more BF1). HERE is the result of a couple of seconds before and after the crash.

As you can see, the GPU's reported temperature dropped from 73 (which doesn't seem like a temperature at which it should crash) to 0, maybe Windows couldn't find the GPU anymore? I thought that maybe I didn't put the GPU back properly last time I cleaned it (a while ago, not right before this started), so I tried to take it out and put it back in again (as well as its power cables), but that didn't fix the issue. I also tried to take a screenshot right after the crash, but that didn't really look as useful as I'd hoped either.

 

Next, I ran a few stress tests. First I ran FurMark by itself for ~90 minutes (Results), followed by Prime95 v2.26 64-bit Small FFTs for about an hour (Results). After that I ran both, to see if maybe my PSU couldn't handle the stress anymore for about an hour (Results) immediately followed (no time to cool down) by 90 minutes of FurMark + Prime95 v2.26 64-bit Blend to test RAM a bit as well (Results).

 

During these tests, I encountered not even a single crash, which only confuses me more. However, 5 minutes into BF1 about an hour after stopping the last test, it crashed again. I have yet to run memtest64, as I didn't think that anything would be wrong with my RAM, considering it hasn't been touched in quite a long time, but not long enough to fail due to old age.

 

 

Recent changes I've made to my pc:

- The crashes started right after I had updated to Nvidia's latest driver. As stated before, going back to older drivers didn't help.

- I installed the Windows 10 Creator's Update shortly before it started happening (can't remember if that was a month, a week or a day or how much I gamed during that period).

- I reapplied the thermal paste on my CPU and GPU, after getting annoyed by how loud and slow my GPU had gotten (I noticed it was running (and throttling) at 84c). This happened at least 1.5 months before the crashes started happening and it only seemed to have done the GPU a favor (what was left of MSI's paste had become dust). I also tested the GPU quite a bit after this and didn't find any problems, so I doubt this is causing the crashes now.

 

 

As I cannot figure out what is wrong with my pc and I am starting to run out of ideas, would anyone here be able to offer some insight?

[Technomancer__]

well... this is strange...

i had the same think happened to me but i crashed right after launching a game. i did everything like you did but nothing helped

 

but when i reformated it was fixed. idk why.

try installing windows on another secondary drive and if your pc works fine with that then reformat it

[pim16aap2]

2 minutes ago, Ethocreeper said:

well... this is strange...

i had the same think happened to me but i crashed right after launching a game. i did everything like you did but nothing helped

 

but when i reformated it was fixed. idk why.

try installing windows on another secondary drive and if your pc works fine with that then reformat it

Guess that's worth a shot, yeah.

 

Reformatted not too long ago, so I was trying to avoid it, as it's such a pain

[Technomancer__]

2 hours ago, pim16aap2 said:

Guess that's worth a shot, yeah.

 

Reformatted not too long ago, so I was trying to avoid it, as it's such a pain

thats why is said to try first on another disk and if it works then reformat

[pim16aap2]

All right, I've finally gotten around to trying and and i didn't encounter any crashes on a fresh installation, so I guess I'll have to find some time to reformat my drive

 

Also, I'd forgotten how agonizingly slow HDD's are... I thought my pc was dying or something and kept checking the stats until I realised I was no longer using an SSD. 

[pim16aap2]

All right, after getting a fresh install of Windows 10 everything seemed to be fine at first. Until it started happening again... First I got a few BSOD's playing Getting Over It With Bennett Foddy, and then the usual Video Driver crash which then fails to restart, leaving me with no video signal.

 

Anyone got any ideas I could try? Should I try getting a new GPU? PSU? RAM? I'm not entirely sure where to go from here.

[pim16aap2]

All right, after the last crash, it finally logged something in the Event Viewer.

HERE are the 4 errors related to the crash.

 

Windows Error Reporting also created a DMP file and showed me its location. When I went there, I noticed that there were a bunch of DMP files, so I guess it has been creating those DMP files without adding entries to the Event Viewer.

 

HERE are is the data in the DMP files for the 7 most recent crashes when opened using BlueScreenView.

 

Furthermore, it also sent me to a bunch of tmp.WERInternalMetadata.xml files (HERE is one of them), a Report.wer for the kernel and a Report.wer for ApCrash_bf1.exe 

The AppCrash report was generated 2 minutes later than the Kernel report, but I'm not sure how relevant that is.

 

What stuck out to me, was line 151 of the Kernal report, which is:

Quote

ReportDescription=A problem with your hardware caused Windows to stop working correctly.

 

I don't know how reliable that information is though, would anyone here be more knowledgeable about this subject? Or is this maybe the point where I should contact Microsoft and MSI?

 

Also, it seems to be happening much faster after launching a game now, I can play BF1 for about 10 minutes before it crashes now... At least it makes testing easier...

[iHDCool]

I personally trust the BlueScreenView, accurate info when I had BSODs before. Since you fresh installed the windows, I'm guessing it's not a software prob but driver/hardware. Do you have another rig where you can try out your hardware? Do a Disk scan for potential storage issues. Get CrystalDiskinfo. 

[pim16aap2]

3 hours ago, iHDCool said:

I personally trust the BlueScreenView, accurate info when I had BSODs before. Since you fresh installed the windows, I'm guessing it's not a software prob but driver/hardware. Do you have another rig where you can try out your hardware? Do a Disk scan for potential storage issues. Get CrystalDiskinfo. 

THIS is what I get when using CrystalDiskInfo.

 

Sadly, I do not have access to another rig I could use...

 

Just trying to pin down the issue and then buy new hardware to replace whatever is broken. If replacing the hardware I thought was broken doesn't fix the issue, I'll return it. 

[SupaKomputa]

Your GPU is failing. Can you run stable with iGPU instead?

[pim16aap2]

On 14/12/2017 at 8:09 PM, SupaKomputa said:

Your GPU is failing. Can you run stable with iGPU instead?

I tried gaming on my Intel HD 4600 for about 2 hours without a crash. Couldn't bear to test it out any longer, started feeling sick and getting a headache...

 

I'd not tested this before because I thought that not getting a crash on my iGPU wouldn't be definitive proof that is was my GPU that was failing, as surely the rest of the system would be stressed differently when the GPU is replaced by a potato.

 

I do also think that my GPU is failing though, so I'll probably buy a new one some time next week. I'll post a reply when I have the result.

[SupaKomputa]

Well for me that is a solid proof that the GPU is failing.

There is some reason that the gpu brake:

- (light) heatsink failing,

symptoms : crashing when hot, remedy: take out the heatsink, reapply thermal compound.

- (hard) The BGA solder failed / melted / corroded,

symptoms : sudden / random crash when gpu initialized (starting games), remedy: take it to gpu repair and reapply heat gun / reflow / rework.

- (extreme) Chip Failing,

symptoms : screen artifact, discoloration, random crash, remedy: replace the gpu chip.

 

Those are the most common gpu problems and fix. will temporary fix the gpu until the problem spread to other part of the card. The GPU will eventually died, not worth the effort.

[pim16aap2]

All right, I've side graded my GPU to a GTX 1060 6GB (Gigabyte G1 Gaming) and the crashes seem to be gone now!  

 

Thanks for the help everyone!

[KryptonSL97]

I resolved disabling PCIe 3 in bios settings.