Samsung NVMe "Unsafe Shutdowns" increments even WITH ErP Enabled / No power strip cut needed. Is this a Gigabyte/Samsung firmware bug?
Hi everyone,
I’m facing a very specific and frustrating SMART telemetry issue with my NVMe drive on a brand new AM5 build, and I need some technical insight into PCIe power states and motherboard behavior.
System Specs:
CPU: AMD Ryzen 7 9850X3D
GPU: NVIDIA GeForce RTX 5080 Shadow
SSD: Samsung SSD 9100 PRO 2TB (Installed in the top M.2 slot, direct CPU lanes)
Motherboard: Gigabyte X870E Aorus Elite (BIOS version: F9)
OS: Windows 11 Pro
RAM: Corsair DDR5 CL30 32GB with EXPO enabled
PU: Asus ROG Strix 1000w Platinum
The Problem:
My SSD's SMART data is showing an alarming number of Unsafe Shutdowns (157) with only 196 Power-On Hours. The counter goes up by 1 almost every single time I turn off my PC.
Initially, I thought this only happened because I flip the switch on my power strip (surge protector) at night (even if I wait 2 hours after a clean Windows shutdown). However, I just made a major discovery: The counter increments sometimes even if I leave the power strip turned ON, as long as ErP Ready is enabled in the BIOS.
Here is the breakdown of what happens:
ErP OFF + Power Strip ON: If I leave the system completely stock and leave the wall power on, the motherboard keeps the 5VSB standby rail active. The counter usually DOES NOT increment, but about 1 out of 10 times it still does.
ErP OFF + Power Strip OFF (after 2 hours): If I cut the wall power hours after a clean shutdown, the residual capacitor drain from the PSU causes the 5VSB rail to drop slowly. The drive registers an Unsafe Shutdown (+1).
ErP ON + Power Strip ON: If I enable ErP in the BIOS to cut standby power automatically, the motherboard drops the 5VSB rail internally a second after Windows shuts down. The drive STILL registers an Unsafe Shutdown (+1), even without touching the physical wall switch.
(Percentage Used is 0%, Media Errors is 0%) and that Windows is shutting down perfectly (no Kernel-Power 41 crashes in Event Viewer).
What I’ve already configured:
Windows: Fast Startup is Disabled. PCI Express Link State Power Management (ASPM) is set to Off.
BIOS: Memory Context Restore and DRAM Power Down are Enabled to maintain fast boot times.
Note:
I do not want to physically disassemble the PC to move the SSD to a secondary chipset-managed M.2 slot, nor do I want to flash the BIOS to the newer F10 version since my F9 build is rock-solid stable and F10 only mentions generic AGESA updates.
My Question:
Has anyone else encountered this specific behavior with Samsung drives on recent Gigabyte/Aorus X870E platforms? Why does the Samsung controller interpret a standard 5VSB power rail drop (either via ErP or wall cutoff) as an abrupt power loss when the drive has been completely idle and unbothered? Is there any hidden low-level firmware setting that can fix this synchronization gap?
Thanks to anyone who can provide some insight!