[SOLVED] Sudden crashes to black screen while gaming (CPU EZ Debug light on)

[aeqri]

Hi,

Ever since I've installed a few upgrades to an originally pre-built PC, I've started to run into seemingly random crashes. The system only crashes while gaming, not even once while idling, browsing the internet or even stress testing.

Here's what the crash looks like:

• Both monitors lose signal
• No audio
• Keyboard & Mouse LEDs turn off
• All fans continue spinning
• CPU debug LED on the MOBO solid white (CPU not detected or fail)
• Front panel power button stops working, but LED stays on

The only way to restart the PC after this is to power cycle the PSU, which doesn't always work (won't POST, same symptoms as above). Waiting a minute or two before attempting to power on usually results in a successful POST.
I also noticed that launching a game shortly after one of these crashes, it's likely to crash again within 5 minutes, if not instantly.

PC Specifications:
The PC was originally a pre-built, purchased just over 4 years ago. I upgraded the CPU, MOBO and RAM almost 2 months ago, which is when the crashes started occuring.

• OS: Windows 10 Home 2004
• CPU: AMD Ryzen 5 3600 (stock cooler, no OC) [originally Intel i5 6500]
• MOBO: MSI B450M PRO-VDH MAX [originally ASUS H110M-R]
• RAM: Corsair Vengeance LPX (2 x 8 GB DDR4) XMP @ 3200 MHz [originally HyperX Fury (2 x 4 GB DDR4) 2133MHz]
• GPU: Zotac GTX 1070 Founders Edition
• PSU: Corsair VS550

What I've tried so far:

• Installing and upgrading all necessary drivers
• Flashing BIOS
• Running RAM at stock frequency (non XMP)
• Using the old RAM sticks (still crashing, so likely not RAM?)
• Using different Windows power plans (also the ones that come with AMD chipset drivers)
• Windows crash dumps (none get created)
• Fresh Windows install (was crashing in 1904 too)
• Reseating the CPU, along with new thermal paste (no bent pins)
• Disassembling and reassembling the PC
• Temperature monitoring while gaming (CPU maxed out @ 74°C, GPU @ 65°C, doesn't seem like overheat)
• Checking Event Viewer logs (just a vague "Windows did not shut down properly")
• Clearing CMOS

I feel like at this point the only thing to do is to just RMA random parts or buy new ones to try. If there's any other thing I could try to hopefully narrow down the culprit, or which part I could try swapping/RMA'ing first please do let me know.

Thank you for your time.

 

[dchen2105]

It's really hard to say, could be a defective CPU or motherboard easily, maybe some settings on the CPU cause it to crash? Seems like a CPU problem or motherboard problem, but also could be that the PSU is causing issues. Could be that when the CPU tries to boost while gaming, the PSU just can't handle the wattage? I think PSUs start to lose wattage over time, however seems like your build should just be drawing around 300W anyways.

 

[dorsai]

Are you using MSI Dragon Center ? If so uninstall it...that program is garbage and can cause some really weird problems.

You said it's stable during stress testing...but have you stress tested the GPU ? If it's crashing only while gaming the GPU becomes a major culprit in my opinion.

 

[aeqri]

Are you using MSI Dragon Center ? If so uninstall it...that program is garbage and can cause some really weird problems.

You said it's stable during stress testing...but have you stress tested the GPU ? If it's crashing only while gaming the GPU becomes a major culprit in my opinion.

I'm not using MSI Dragon Center or Ryzen Master.

The system is stable while stress testing both CPU and GPU with OCCT. I ran a 1hr test without any crashes, but the moment I launched a game it crashed within a minute.

My first thought was a possible driver conflict. When I initially upgraded the PC, I just kept using the same Windows installation. I ended up doing a fresh install a week later, which sadly did not fix the issue.

 

[dorsai]

I'm not using MSI Dragon Center or Ryzen Master.

The system is stable while stress testing both CPU and GPU with OCCT. I ran a 1hr test without any crashes, but the moment I launched a game it crashed within a minute.

My first thought was a possible driver conflict. When I initially upgraded the PC, I just kept using the same Windows installation. I ended up doing a fresh install a week later, which sadly did not fix the issue.

Have you run any tests on your HHD or SSD ?

Try switching DX versions in game...if you're trying to run 12 see if the game will let you run it on 11 or 10.

 

[aeqri]

Have you run any tests on your HHD or SSD ?

Try switching DX versions in game...if you're trying to run 12 see if the game will let you run it on 11 or 10.

I ran chkdsk on both drives, no errors found on either.

Checked CrystalDiskInfo, SSD remaining life is at 66%. Other than that, both drives look good.

About DirectX:
I don't have many games currently installed. Gunfire Reborn is the game that crashes pretty consistently, only supports DX11. It's the game I use to test if the system is still crashing. CS:GO, which supports up to DX9 also crashes, though not as consistently.

I've also discovered a new way to cause a crash immediately:
In Blender, whenever I use NVIDIA's OptiX AI Denoiser in the render preview, it crashes instantly after it does the first few path traces, every single time.
However, if I instead use OpenImageDenoise, which runs on the CPU, it doesn't ever crash.

I used DDU to clean NVIDIA's drivers and downgraded to version 432.00 (Jul 24, 2019), which doesn't come with OptiX. Blender now uses OpenImageDenoise by default without any issues, but the crashing still persists in games, atleast in Gunfire Reborn.

I'll be googling around a little more about this.

 

[mdd1963]

I'd substitute in a known good PSU before proceeding any further beyond RAM reseats, BIOS updates, and/or quick formats/reinstalls..(any PSU of whatever lineage can work fine last week, and even yesterday, and then start having issues at some point)

(If /when swapping PSUs , be sure to swap all modular cables with it, as they are rarely pinned the same at/for the PSU insertion connection points)

 

[dorsai]

I'd try a different PSU...if you live in the US you can order one or grab one at Bestbuy and return it if the crashing continues. If the PSU checks out then I'd be looking at the GPU as the likely bad hardware.

 

[aeqri]

Well, bought an RM750 and it has stopped crashing. Even the blender test, which caused an instant crash every time. It seems that the PSU was the culprit after all.

Thank you all for your help.