Question NVMe SSD Suddenly Stops Working Under Stress (Games / Stress Test)
Question NVMe SSD Suddenly Stops Working Under Stress (Games / Stress Test)
Hello everyone,
I’m trying to figure out a recurring problem with my PC. The NVMe SSD (a budget RX7 1TB model) seems to vanish unexpectedly during high demand—typically when I start a game or play for a short time after launching demanding titles on the D: partition, or when I run FurMark and CrystalDiskMark together to mimic a tough gaming setup. Once this happens, Windows either crashes with BSODs such as “Kernel Inpage Error” or “Unexpected Store Exception,” or it just freezes. After a while, the drive comes back online.
If I don’t play any games on that partition, the system runs perfectly—just browsing or playing very light games.
System details:
- CPU: Ryzen 5 5600
- GPU: MSI RX 6700 XT (12GB)
- Motherboard: Gigabyte AB350M Gaming 3 (latest BIOS, version F53g, recently updated)
- RAM: Adata XPG Gammix 32GB DDR4 at 3200MHz
- Boot Drive: SATA SSD 128GB (Windows installed)
- Game Drive: RX7 NVMe 1TB (D drive)
- Power Supply: Corsair TX750M Gold
- Case: Gamemax Spark M-ATX
Testing and observations:
- OCCT Combined test (CPU, RAM, GPU, VRAM) also causes the same issue.
- The problem only appears when both GPU and NVMe are under stress.
- It doesn’t happen with FurMark or CrystalDiskMark alone.
- HWiNFO logs indicate voltage drops on the +3.3V rail to 3.06–3.10V just before the drive disappears. During idle or light load, it stays above 3.20V.
- Windows Event Viewer shows repeated events:
- stornvme: Reset to device \Device\RaidPort1
- disk: Event ID 51 – paging failure on D
- ntfs: Event ID 50/140 – delayed write errors
- BIOS NVMe self-test passed; no SMART errors.
- Undervolting the GPU to 1110mV (power limit -6%) reduces power draw but doesn’t prevent crashes under load.
- Running CrystalDiskMark or playing a game (like UE5) with light load doesn’t trigger a crash.
- The unusual fact is that the drive stays at around 40°C regardless of workload, which suggests a sensor or controller issue.
My guesses:
- Voltage instability on the +3.3V rail under heavy load may disrupt the NVMe link.
- It could be a power supply weakness or insufficient voltage delivery from the motherboard to the M.2 slot.
- I’m considering whether this is just an NVMe drive problem or if it’s related to the PSU or motherboard.
Advice:
- Should I focus on confirming that this is solely an NVMe issue, or should I check for a PSU or board fault?
- Is switching to a SATA SSD for gaming a safer long-term solution?
- Any BIOS tweaks (C-states, spread spectrum, etc.) I should try?
Feel free to share the HWiNFO CSV or Event Viewer XML if you’d like.
Best,
[Your Name]
The problem might be related to a thermal issue. The drive seems to reappear after powering off, which could indicate heat affecting performance. Could you provide details about your case and the SSD temperatures during the issue?
Hardware:
Gigabyte AB350M Gaming 3 (latest BIOS)
Please specify the BIOS version for your motherboard.
Power Supply:
Corsair TX750M Gold
How long has the PSU been in use?
Thank you for your quick response. The setup isn't an Adata SSD as discussed; I've included the extra details.
- The case is Gamemax Spark M-ATX
- SSD temperature remains stable without fluctuations and stays at 40°C on HDSentinel, CrystalDisk, HWInfo, etc.
- BIOS version F53g was updated from F51h last night, yet the problem continues
- Heat expectations should influence FurMark testing when no drive activity is present, since the M.2 is positioned directly under the GPU, but the drive was still recognized even at 86°C
- The PSU appears to be an older model, likely from around 2017
Probably the cheap drive's controller can't sustain the heat (from itself or other components like the GPU) and fails when gaming.
Drives are probably the worst thing to cheap out after the PSU. There are so many horrible SSDs out there that cost peanuts but can't run anything without crashing, or fail after a few months. You should back up everything you want to keep from this drive and buy a decent one. A good 1 TB NVMe from a recognized brand (Samsung, WD, Crucial) cost about 60-70$ USD. There's no reason to go with the crap.