Crash -> CPU debug light

Cassius VII · August 5, 2020

Hi folks, first post here

I was using firefox when my screens turned to black, CPU debug light was lit on the mobo, restart and power button of the case not doing anything. I switched off the power supply, and then rebooted. No blue screen logs, event viewer just lists an unschedule shutdown.

Il built this machine on december 2018, all parts were bought brand new :

CPU : Ryzen 5 2600 @4.025 GHz, 1.3375 V (vcore) - LLC2 - 1.065 V (SoC voltage)
RAM : 2x 8 GB Ballistix Sport 2666 MHz (BLS8G4D26BFSEK.8FD) @ 3000MHz - 1.35 V
Motherboard : MSi B450 GAMING PRO CARBON AC - Bios 7B85v1B - chipset driver 2.07.14.327
GPU : Sapphire Pulse Vega 56 - Core @1647 Mhz / 1080 mV, memory @950 MHz (Driver 20.5.1)
PSU : EVGA 750 GQ, 750W 80+ Gold
Storage : HP EX900 256 GB for the OS
CPU Cooler : Cooler Master ML240L RGB

I'm running windows 10 2004 (19041.388)

Regarding the overclocks :

GPU has been stable for months on various games and folding
RAM passed 6+ hours of memtest (12 GB, 10800%)
CPU passed 2h of prime95 small FFTs, 2h OCCT AVX2, 4h OCCT default. Max temp reached 74.3°C

Strange thing with my cpu OC is that back in august of 2019, It passed 10 hours of prime95 at 4.025 GHz with 1.325 V with a max temp of 72.3°C. I never exceed 65°C while gaming which is my heaviest usage of this computer. So I don't think that I was above any safe limits mentionned here. But I had to increase the voltage to 1.3375 if I want to maintain this frequency now, because of other crashes with the CPU debug light turned on.

I kept logs of HWinfo64 during a prime95 run that ended with a crash if you want to check anything with the log viewer. The only thing I see that i don't understand are the VIN4 and VIN6 voltages : I don't know what are those voltages and i see this behaviour everytime the system is under cpu load.

So I don't really now what could be wrong : motherboard, CPU, PSU, bad CPU overclock ... 99% of the time, this computer serves me more than well.

However, as the end of my 2 year warranty gets closer, if one component is bad, I'd like to RMA it.

My problem is identifying it, if there is any.

So any suggestion would be appreciated !

Thanks

08-05-2020-OC3-prime95.CSV

Applefreak · August 5, 2020

If your chip suddenly needs more voltage to be stable, consider running the system without any oc unless you can afford to have it go up in smoke. Running Prime95 even for 48 hours means nothing long term. If you're lucky your system will run stable without the overclock. Consider yourself lucky, most systems would not even turn on after such an event. Also if you want to run a system long term overclocked, use better cooling solution.

As for VIN, that refers to the input voltage on a specific part. VIN 6 refers to the VRM as far as I remember but don't quote me on that. I have no idea what does mean in terms of stability. I'd run the system at base clocks for a while (maybe keep the ram up if you like as it will less likely kill the motherboard.