Ryzen 1600X (B350) build has consistent system shutdown after loading any games (CPU debug LED on)

Hello everyone, I am at wit's end. Sorry for the huge post, but I don't want to leave out any possible leads.


TL;DR: My rig completely restarts whenever I load any game. It is extremely unstable, and I suspect a defective 1600X CPU.


I have a weird problem happening with my Ryzen 1600X build, and I’m running out of ideas for diagnosing the issue. I think I may have to RMA most likely my CPU, though possibly my motherboard and/or RAM.

My system specs: (all running at cleared-CMOS default frequencies and voltages)
-Ryzen 1600X (Hyper 212 Evo Cooler) on Ryzen Balanced Power Plan chipset
-16GB (2x8GB) Corsair Vengeance LPX DDR4 (rated for 3200MHz) running @ 2133MHz default
-MSI Tomahawk B350 (updated to bios v19)
-MSI Duke GTX 1080 8GB (WHQL 388 drivers)
-EVGA Supernova NEX750G
-Windows 10 (fall creators update)

I have the AMD balanced power plan chipset installed, I have a fresh install of windows, and windows is updated to the newest fall creators update along with all appropriate updated drivers.

My issue:
My system completely shuts down when booting up games like Arma 3, PUBG, and RBS Siege. It shuts down the same way as if your CPU was too overclocked and your system would shut down to prevent damage. The system completely goes black, my keyboard LEDs cycles off, my case fans and LEDs stay lit and spinning, but my system fully reboots.

My system is currently capable of posting, getting into the bios (to play around with settings), and even boots into windows without issue. In fact, I could use my system seemingly reliably for Microsoft Office, web browsing, etc as long as I don't load games.

My system has been flawless after putting it together around June this year, especially ever since I reinstalled windows. My GTX 1080 had no issues destroying games at 1440p, and my 1600X could demolish even Arma 3 maxed out at 60 fps. I used to be able to play games for hours on end with complete reliability. I never bothered overclocking.

However, within the past few months I tried the following overclock:
- CPU freq. @ 3.9 GHz
- Loadline Calibration set to mode 1 (10%)
- BIOS CPU voltage set to 1.375v (the actual CPU voltage reported by the mobo would go as high as 1.42v under load but would drop down and not go over that voltage)
- RAM XMP profile with frequency @ 3200MHz, 1.6v, and 14-16-16-16-18 timings
- AMD Cool'n'Quiet enabled
- CPU Overvoltage protection enabled
Note: when messing with overclock settings, my system would post instantly and always. I never had to wait for my system to do any "memory post retries".

My temperatures would never go above 50*C while gaming, so they were never a problem.

At first I was able to game for a couple hours before my system would crash. Soon after, I started noticing how it would take less and less time during gaming for my system to crash and the CPU or DRAM debug light would show up on my mobo. I have since removed all overclock setting by clearing CMOS and games like Arma 3 barely make it to the menu before shutting down. Even RBS Siege crashes at the menu for me.

It got so bad today that I tried to swap my 2 RAM DIMMs from the B slots to the A slots (thinking it may help), and that caused BSOD in windows. It got so bad that my PC completely stopped posting. I took apart my entire rig, re-seated my CPU, put new thermal paste on, put my RAM back into the original B slots, and I am lucky enough to boot into Windows now. Games are at a 100% shutdown rate now, where I can't even make it the main menu.

I tried to overclock in bios again just to try to diagnose problems, and my rig has a hard time posting with super conservative overclocks like 3.7GHz @ 1.36v. It would take my PC 5 attempts to self-post until it finally gets into bios again. Whenever it posts and crashes, it almost always has the CPU debug light on, indicating an issue with the CPU.


I use my PC every day for college classes, so I don't have time to mess around with all the potential RMAs for my parts. If I have to RMA an item, I need to be sure that that is the defective part. Perhaps someone here can point me in the right direction.

Thanks in advance.




***UPDATE***
I have been able to run Prime95 Blend Test, CPU-Z stress test, and Cinebench R15 for over an hour without any crashes or artifacts. However, as soon as I load up a Heaven benchmark on top of these stress tests, it only takes about 1-2 minutes in until my system shuts down and reboots.

I took a recording of my motherboard's debug lights as the crash happens, and it shows that the CPU and DRAM LEDs come on first, hang for a second, turn off, and then the VGA LED turns on for a few seconds (indicating an issue with the GPU).

I went ahead and recorded the debug LEDs as they crashed in RBS Siege, and it happened exactly in the same sequence and order as it did for the stress tests.