Diagnosing random power-offs

[missster.anderson]

(Full specs listed at bottom)

I've run into an inconsistent yet reoccurring issue where my computer will seemingly randomly completely shut off. This started happening shortly after I upgraded a significant portion of my machine in fall of 2019 when I swapped CPU, MOBO, RAM, and added an NVME SSD.

After doing these swaps, I noticed that about once a week or so my BIOS would reset and I'd have to go in to re-enable certain features like XMP, custom fan profiles and such. Eventually, the power drop off issue started happening so naturally I figured, "well darn, one thing I was hoping to get away with was not needing to swap the PSU, guess I have to now" and swapped out the PSU. Exact same issue still persists even after putting in a new PSU. Next I figured it was likely either a RAM issue or an overheating issue. I've ran memtest86 twice and no errors were found either time. Then I pulled out the CPU and GPU and re-applied thermal paste to make sure I had good contact underneath to diagnose the potential overheating (machine runs warm but never anything close to HOT but wanted to make sure).

Even with memtest86 returning no errors, I have found that the issue only ever happens when running in single channel mode. If I am using only one stick (tested both) of my 2x16gb kit (with or without XMP enabled) I never run into the issue of BIOS resetting or power offs. I have tried every possible arrangement of RAM possible with the 2 stick kit I have. dual channel in the 'correct' slots, dual channel in the 'incorrect' slots, single channel with both sticks (slots 1&2 or 3&4) as well as only one stick in every slot (tested both sticks in this manner). This leads me to believe there is something wrong with my MOBO or CPU since those are all that is left after the part upgrades I did. I've checked for bent/loose pins on the CPU and everything seems in order there as well.

 

If anybody knows of any way to narrow down whether it's a MOBO or CPU issue without the need of a secondary machine for testing, you'd be doing me a massive service.

 

System specs:
CPU: Ryzen 5 3600x
MOBO: GIGABYTE x570 AORUS Elite
RAM: Corsair Vengeance LPX 32GB 3200MHz C16 (2x16GB)
GPU: PNY XLR8 1070 8GB
PSU: MSI MPG 750W
CPU Cooler: Deepcool Gammax GT

 

STORAGE;
NVMe SSD: Intel 660p 1TB
SSD: Samsung 850 EVO
HDD: Seagate 2TB

 

[ninjitsu]

Are you running the RAM in single channel?

Have you monitored the actual temperatures while gaming? If you leave HWMonitor running while gaming, it will record the max temperatures you hit. Have you tried updating motherboard bios?

[missster.anderson]

31 minutes ago, ninjitsu said:

Are you running the RAM in single channel?

Have you monitored the actual temperatures while gaming? If you leave HWMonitor running while gaming, it will record the max temperatures you hit. Have you tried updating motherboard bios?

currently back to running in dual channel. because I run into the issue either way. I've run HWmonitor a few times and have never seen anything approaching *panic* or even hot for that matter. I'm all air cooling but I run fans pretty aggressively since i don't care about noise. I've also tried multiple different BIOS updates. basically every time a new one comes out just to make sure the issue persists. With regards to the temperatures though, as I said, this issue happens seemingly randomly. It's not just while gaming; sometimes it's while the system is just sitting idle, but it is usually under load when it happens. I'd say the crash is correlated with being under load about 80% of the time if i had to put a number on it.

[ninjitsu]

Have you looked at logs in event viewer after a crash? Did you do a fresh install of Windows after your upgrade? Are the AMD chipset drivers up to date? Also, what timing is XMP setting your RAM at? With my 3600, it's a little picky with timing. Have tried running at 3000 instead of 3200?

[missster.anderson]

2 hours ago, ninjitsu said:

Have you looked at logs in event viewer after a crash? Did you do a fresh install of Windows after your upgrade? Are the AMD chipset drivers up to date? Also, what timing is XMP setting your RAM at? With my 3600, it's a little picky with timing. Have tried running at 3000 instead of 3200?

nothing shown on logs; didn't do a fresh install, probably should; chipset drivers are up to date; running 3200mhz on the XMP but again, I seem to only run into this issue when running multiple sticks of RAM no matter the XMP profile. Using either stick at 2133mhz or XMP seems to be fine, but as soon as i add in the second stick, whether at stock or XMP, BIOS will occasionally be reset, and the computer will occasionally fall completely on it's face and then start right back up like nothing happened. 

For reference, I've just run furmark and CPU burner at the same time for 30 minutes and the CPU temps topped out at 78° C and GPU at 75° C.

[ninjitsu]

Are you able to test with a different RAM kit? Maybe something with smaller dim size, like two 8 gig sticks?

[missster.anderson]

3 hours ago, ninjitsu said:

Are you able to test with a different RAM kit? Maybe something with smaller dim size, like two 8 gig sticks?

unfortunately no. I've exhausted the extent of my "try this" with regards to hardware availability without opening up the parts cannon AKA my wallet

 

NOW, I did find while digging through the depths of the GIGABYTE bios on this MOBO an option for "Power Down Mode" in the RAM settings. I've read this can be a serious cause for instability if not explicitly disabled instead of 'auto'. I've gone ahead and set this as well as calling out "high frequency support" to Level 1 for 3200MHz instead of 'auto'. The scientific method would point me to only try one at a time, but I'm sick of this happening and am willing to toggle all the possible switches at once if it means it never happens again haha

[VladTheImpaler]

14 hours ago, missster.anderson said:

unfortunately no. I've exhausted the extent of my "try this" with regards to hardware availability without opening up the parts cannon AKA my wallet

 

NOW, I did find while digging through the depths of the GIGABYTE bios on this MOBO an option for "Power Down Mode" in the RAM settings. I've read this can be a serious cause for instability if not explicitly disabled instead of 'auto'. I've gone ahead and set this as well as calling out "high frequency support" to Level 1 for 3200MHz instead of 'auto'. The scientific method would point me to only try one at a time, but I'm sick of this happening and am willing to toggle all the possible switches at once if it means it never happens again haha

The bios resetting is a red flag.. No software or external hardware can reset the bios like that.. I'd try different bios versions

It's highly unlikely since the system should have external power at all times, but maybe replacing the CMOS battery?

Honestly the bios resetting suggests a deep problem with the motherboard..

 

 

[missster.anderson]

2 hours ago, VladTheImpaler said:

The bios resetting is a red flag.. No software or external hardware can reset the bios like that.. I'd try different bios versions

It's highly unlikely since the system should have external power at all times, but maybe replacing the CMOS battery?

Honestly the bios resetting suggests a deep problem with the motherboard..

 

 

This was certainly my thought. I haven't had enough time to really put my previous bios settings through the wringer. I've read a decent amount about RAM instabilities with Ryzen and how the 'auto' settings in BIOS's with RAM settings in particular seem to be a sticking point. I've tried 6 different BIOS versions since swapping to this new MOBO back in 2019 and have checked the CMOS multiple times with a voltmeter just to make sure. At this point I'm fairly certain it's either 1) a faulty MOBO or 2) some wack RAM settings causing instability. I've currently set basically everything I can think of to remedy instability and am hoping this will prevent me from needing to open the wallet

[ninjitsu]

I've had my bios reset itself after messing with RAM timings. If it fails to boot a couple times, it fails back to default. Something similar might be happening to you. Maybe it's time for a 12th gen intel system? :D