X570 Aorus Master: Random Crashes/Freezes

[_adrian]

I’ve reached a dead end trying to troubleshoot random freezes/crashes on my X570 Aorus Master. I’ve even RMA’d the board (Gigabyte claims they “fixed it” but won’t tell me which part they tested/repaired) and the problems persist.

What I Have:

• program crashes / freezes
• whole system crashes / freezes
• i/o (mouse/keyboard) freezes
• system freezes but mouse still moves
• not obviously related to any particular program
• not related to system load (happens at idle, normal usage, heavy usage)
• happens on multiple linux distros (debian, ubuntu, fedora); sometimes within days, sometimes within an hour of boot
• happens on windows (though windows is far less frequent)
• has never happened in bios
• I sometimes see crash reports and entries in journalctl, but I am not convinced these are causal - often don’t seem related, and usually there’s nothing.

What I’ve Tried:

• multiple bios updates, from F11 ip to current (F31)
• multiple kernel updates, from 5.4 up to 5.10.6 (.7 breaks my video)
• memtest86 (no errors)
• stress-ng (no errors/no crashes provoked)
• swapping out literally all other hardware:
  • CPUs: 3600, 3900X
  • RAM: patriot viper 32GB DDR4-3600 (x1,x2), Crucial Sport 16GB DDR4-3000 (x1,x2,x3,x4), Corsair Vengeance 8GB DDR4-3200 (x1,x2), random OEM 8GB DDR-4 2666 (x1,x2)
  • VGA: RTX 2080 Super, GTX 1080, GTX 1060 3GB, RX 480 8GB, FirePro X2100
  • Install media: Samsung 970 Evo, Samsung 960 Evo, AData XPG SX8200, Samsung 860 Pro, various other SSDs
  • PSU: Seasonic Platinum 1200, EVGA 850 G3, Corsair RM1000X
  • Motherboard: Asus Prime X570 Pro. ALL components above work flawlessly on this board.

I’m now nagging Gigabyte for a replacement, but they have ignored me so far. At a loss of what to look at next. Anything is appreciated!

[robbbot]

Only crashes I’ve had with the same board have been related to using their XMP profile, I set the voltages manually, and it’s been stable ever since.

[_adrian]

I’ve done XMP profiles on the Patriot and Ballistix kits. Since I started trying to troubleshoot, I’ve left all kits at stock.

Not sure how to decide best voltages…?

[robbbot]

This would be a good place to start https://download.gigabyte.com/FileList/Memory/mb_memory_x570-aorus-master_matisse_190812.pdf

[t-l-k]

I’d be interested in learning more about what voltages you set for stability.

I have 2 x 16gb Gskill F4-3600C17-16GTZR for 32gb in total on my x570 Auros Master with a 3900x, but my Arch install crashes (powers off) in under a minute if I boot with the XMP profile enabled.

Windows is fine. Live boot environments also seem to work fine.

F31b bios has been most stable for me overall.

[robbbot]

I set the memory multiplier to 36.66, and have 1.35v, although I’m not sure if dmidecode is reporting it correctly.

Memory Device
	Array Handle: 0x000A
	Error Information Handle: 0x0018
	Total Width: 64 bits
	Data Width: 64 bits
	Size: 32 GB
	Form Factor: DIMM
	Set: None
	Locator: DIMM 1
	Bank Locator: P0 CHANNEL B
	Type: DDR4
	Type Detail: Synchronous Unbuffered (Unregistered)
	Speed: 3666 MT/s
	Manufacturer: Unknown
	Serial Number: 00000000
	Asset Tag: Not Specified
	Part Number: F4-3600C18-32GVK
	Rank: 2
	Configured Memory Speed: 3666 MT/s
	Minimum Voltage: 1.2 V
	Maximum Voltage: 1.2 V
	Configured Voltage: 1.2 V
	Memory Technology: DRAM
	Memory Operating Mode Capability: Volatile memory
	Firmware Version: Unknown
	Module Manufacturer ID: Bank 5, Hex 0xCD
	Module Product ID: Unknown
	Memory Subsystem Controller Manufacturer ID: Unknown
	Memory Subsystem Controller Product ID: Unknown
	Non-Volatile Size: None
	Volatile Size: 32 GB
	Cache Size: None
	Logical Size: None

Or if Arch is setting it back to 1.2

[_adrian]

So thanks to @robbbot and a random L1T video I happened to re-watch, I discovered the “Power Supply Idle Control” setting in bios and set it to “typical.” Things looked good for a few days after that, but just had another hang, so turns out it wasn’t the answer in the end.

Also found out which part Gigabyte replaced when I RMA’d the board: https://www.i-components.com/components/Nuvoton-Technology-Corporation-America/NCT3103S-TR.html

So seems I was right to suspect memory (circuitry).

[_adrian]

amusingly, the thing I did ~20 minutes before the crash was brag on IRC about possibly solving the problem

$ uptime > 17:46:26 up 3 days, 22:24

[robbbot]

Can you attach the log if you can after the crash may help narrow it down something like journalctl --since=today.

[_adrian]

Next time it happens, sure. I’m kinda focusing on getting Gigabyte to exchange it ATM.

In the past, there’s been nothing out of the ordinary in journalctl.