A problem with my home PC - Crash to green screen

[Duranis]

Didn't know if I should post this here or in the hardware section as it is my home PC.

 

Basically I upgraded my home rig a few months ago with a new gfx card, cpu and an ssd and started with a clean win10 install. The problem is that occasionally it crashes, the screen goes a solid green/lime green colour and the computer completely locks.

 

Sometimes it can go hours without doing it, other times it will do it twice within 10 minutes of first turning it on. Also it has never crashed when running stress tests/bench marks (tried a bunch of different ones), only while gaming. There is nothing in the windows log other than an improper shut down. Because it completely locks the PC it is hard to get a log of what is going on as it crashes but manually watching temps and voltages seems to show no big spikes in temp or voltage dips just before it happens (can add temp logs later if needed). I haven't manually tested voltage with a multi-meter yet and on board voltage sensors are never that trustworthy so it may still be a power issue.

 

Spec:

AMD FX-8320

R9 380 Windforce

8GB (2x4GB) DDR3

Gigabyte GA-970A-DS3 Socket AM3+

Corsair 750w PSU

120GB SSD

X2 1TB WD HDD

 

What I have done so far:

• Updated all drivers for everything. Updated BIOS with latest version. Tried the last 3 versions of the GFX card drivers.
  
• Clean reinstalled Windows 10, again.
  
• Unplugged all but the system drive (SSD) and the harddrive with the game on.
  
• Moved the game to the SSD and had this as the only drive plugged in.
  
• Completely took all the hardware apart, cleaned contacts and rebuilt.
  
• Replaced the GFX card with my old one. This actually seemed to solve the problem for awhile but then it started happening again.
  
• Sent the R9 back under RMA, they tested it and said it was working fine so probably not the GFX card.
  
• Put a 12cm fan on the Northbridge heatsink as it was super hot. Apparently that is pretty normal though for the AMD 970 chip.
  
• Unplugged every USB device except the mouse and keyboard.
  
• Tried various combinations of 3D settings.
  
• Added extra support to the GFX just in case it was too heavy (cable tie to case).
  
• Run with 1 stick of ram, when crashed run with just the other.
  
• Calculated voltage/ampage usage for the pc and compared to what the PSU should be giving (assuming it is working fine). In theory the PSU is more than capable.
  
• Googled for answers and found a few people that mention their PC crashing to a solid green screen but no real solutions.
  
• Run stress tests overnight (8+ hours) and not had issues. Run memory checks and hardrive checks, again everything appears to be fine.
  

 

I thought I had solved this until last night when I bought a new game and it started happening again within an hour of playing it. After taking everything apart again and trying different in game settings I noticed 2 things.

 

When I switched it to full screen it then run for about 3 hours and hasn't had a problem.

The Northbridge heatsink gets hot enough to burn you (which is apparently normal).

 

So it seems like running processor/GFX intensive games in a borderless window might be the cause? Load usage, temps and voltages for CPU/GPU don't seem to be any different between borderless and fullscreen though so not sure why it would make a difference.

 

Just really looking for ideas of where to go with this next. I don't have spare components to swap out for testing and I cant afford to buy a new parts when that might not even be the problem.

Edited May 12, 2016 by Duranis

[ZeroHour]

Have you tried disabling the onboard sound?

Also you wouldnt happen to have a ups handy to test the power feed with?

 

Very interesting problem, although no doubt frustrating.

[BassTech]

I had a similar issues a few years ago - I needed to disable the temperature sensors in the BIOS and that did the trick, oddly.

[Duranis]

Thanks, the onboard sound was already disabled. I have tried turning off the temp sensors and any bios power management. Haven't got a UPS here unfortunately but might be something to try (could maybe take to someone else house to test). Is really frustrating. Even worse when I can't force it to happen by stress testing as I have to make a change then run a game for sometimes a couple of hours to test it.

[Homer3334]

This is a good one. Does it happen always when gaming, and are there specific games it happens with too?

[Duranis]

I only have a few games installed at the moment but it has happened with all of them, path of exile, Stellaris and euro truck simulator 2. It has crashed again tonight so borderless window/fullscreen probably not a factor. That was before I disabled any bios power management and sensors though.

 

Used MSI afterburner to log useage while playing and everything seems to be ok. CPU temp maxed at 35C but was most around 25-30C, GPU temp maxed at 57C but was mostly around 50C, Ram usage was around 4GB, pagefile around 5GB and this was over a period of about 2 hours. I haven't managed to get a log that shows anything odd when it crashes so if something is spiking it is doing it before afterburner can write it to the log.

 

Hoping disabling the bits in the bios will help but if it doesnt the only thing left I can think to do is remove the SSD and put my old CPU back in to see if the problem is still there. If it isnt then put both bits back one at a time and see if it triggers it. That of course wouldn't mean that the SSD or CPU was 100% the problem but at least it might give me some clues. Just so odd that benchmarking programs don't seem to cause a problem. I suppose it could be that I have just got "lucky" when running them and whatever is causing it didnt kick in then.

[BassTech]

So it hasn't crashed (yet) since you disabled bios power management/sensors? Mine never used to crash using benchmark software either, ran it for ~24hrs once with no issues whatsoever. Thought it was fixed and within 15 minutes of playing a game it started happening again.

 

The games you're playing, are they downloads/steam or do you have physical discs?

[Duranis]

Yeah I played Stellaris for a few* hours last night after disabling temp sensors and all the parts that change CPU speed, etc and it seems to be working (so far).

 

I don't think I have bought a physical copy of a PC game for at least 10 years

 

*Few = Between 10pm last night and 3am this morning. Stupid gaming making me forget I need sleep to be a functional human being.

[BassTech]

Yeah I played Stellaris for a few* hours last night after disabling temp sensors and all the parts that change CPU speed, etc and it seems to be working (so far).

 

I don't think I have bought a physical copy of a PC game for at least 10 years

 

*Few = Between 10pm last night and 3am this morning. Stupid gaming making me forget I need sleep to be a functional human being.

 

Yeah, I was wondering if it was possibly an optical drive issue Did you manage to fix it in the end?

[Duranis]

No. Even after changing the bios settings it has still crashed a couple of times. However it is not as frequently as it was (maybe once every few days rather than at least once or twice every time it is used). Hate trying to fix things that are just completely random and have no consistent way of triggering them.

 

At some point I put the old CPU back in and remove the SSD to see if that helps but at this stage I'm fed up of messing with it and would rather chuck it in the sea than take it apart again

[BassTech]

haha sounds like a nightmare! It is a really odd issue...