Drivers 3083 Published by

NVIDIA released GeForce Hotfix Display Driver 621.35 on November 20 to stop a bug where the background process NVDisplay.Container.exe keeps rewriting a log file, pushing up to 30 GB of unnecessary writes to your SSD every hour even while idle. The regression first appeared in Game Ready 621.14 on November 4 and worsened with 621.24 on November 12, yet NVIDIA's own 621.24 release notes listed zero open issues. The fix ships as an optional, non-WHQL beta available only through NVIDIA's Customer Care site, and the company itself says the safest option is to wait for the next certified driver. NVIDIA's unusually fast sub-day response came days after AMD shipped a nearly identical SSD-write bug, which the community widely credited for speeding up the fix.





NVIDIA fixes GeForce driver bug that hammered SSDs with up to 30 GB of writes an hour

NVIDIA pushed GeForce Hotfix Display Driver 621.35 on November 20 to stop a background process from eating your SSD's endurance. It's optional, beta, "as-is," and available only through Customer Care.

NVIDIA released the driver on November 20. It patches a bug that had a background process writing up to 30 GB of data to your drive every hour. That's with your PC sitting idle. No rendering. No downloads. Just your desktop sitting there, doing nothing.

Nvidia

Keep in mind this isn't a standard driver update. It's a "hotfix," NVIDIA's word for a targeted patch that skips the usual game-aligned release schedule. It rides on top of the latest Game Ready Driver, currently 621.24, and it does one thing. It stops excessive writes to a log file called nvlog.nvlgstg.

That file lives at C:\ProgramData\NVIDIA Corporation\nvtopps\nct\nvlog.nvlgstg. It's maintained by NVDisplay.Container.exe, which runs under the NVIDIA Display Container LS service. NVIDIA frames the whole thing in a single line on its Customer Care page, logged under Bug ID 7116722.

"GeForce Hotfix Display Driver version 621.35 is based on our latest GeForce Game Ready Driver. This hotfix addresses the following: Excessive writes to file nvlog.nvlgstg may occur in background."

A bug that compounds

The mechanics are where this goes from annoying to genuinely strange. The process is supposed to append to the log. Instead, it truncates the whole file and rewrites it each time. That compounds. The bigger the file grows, the more data a single rewrite pushes. Hence why the file tends to hover near 20 MB. It keeps swelling, then gets chopped back down.

The regression traces to Game Ready 621.14 on November 4. Version 621.24 shipped November 12 and made it worse. And here's the kicker: NVIDIA's own 621.24 release notes list no open issues at all. Across roughly 1,431 lines of text, nowhere does it mention disk writes, nvlog, nvtopps, or the Display Container service.

The headline number is 30 GB an hour. But that's the tail end. The baseline matters. NVIDIA's 621.01 from October 21 was clean, writing under 2 MB per five minutes. Then 621.14 landed on November 4 and pushed that to about 515 MB per five minutes, roughly 3 GB an hour. By the time 621.24 shipped, some users were seeing up to 30 GB an hour. One reporter logged 70+ GB across a session and tracked about 500 MB to 1 GB of extra writing with every Alt+Tab.

Those are process-level figures, and they can overstate what actually reaches the NAND. Writes into the file cache count too. Still, one careful observer cross-checked the SSD's own SMART "Data Units Written" counter and found it matched Windows' numbers. Worth trusting that over a driver's per-process tally.

Now the wear math. A typical 1 TB NVMe drive is rated for a few hundred terabytes of total writes. At 3 GB an hour sustained, that's close to 26 TB a year, for zero benefit. Not catastrophic on its own, but noticeable over time. The scale is what turns this from a background nuisance into something worth acting on.

The AMD factor

NVIDIA wasn't the only maker tripping over a write-hungry process lately. Days earlier, AMD shipped a nearly identical bug. Its atieclxx.exe, part of the AMD External Events Utility, was repeatedly copying its own executable, with reports of 400+ GB a day. AMD dropped an optional, non-WHQL Adrenalin 26.11.1 on November 12.

The two bugs look identical in Task Manager: a background process hammering the drive at idle. But they have different causes and different fixes.

What actually works

If you're living with the bug right now and can't wait for a certified driver, there are workarounds. The fastest is to stop the NVIDIA Display Container LS service, delete the log file, then start the service again. One measured test saw five-minute writes drop from 515 MB to 48 MB, a 91% cut. Temporary, though. The file grows back within hours.

Rolling back to 621.01 is the move that actually holds. NVIDIA recommends doing it through Add/Remove Programs using an older setup.exe, not Windows Device Manager rollback, which won't reliably restore everything. Microsoft says that limitation is intentional.

Set the service to Manual and the writes stop for good. The catch is it breaks the NVIDIA Control Panel and NVIDIA App too. G-SYNC, color profiles, resolution overrides, all of it becomes inaccessible. Ignore anyone telling you the container is just telemetry you can live without. That advice is wrong, and it costs you functionality.

Don't bother turning off telemetry in the App or running a DDU clean reinstall of the same version. Multiple users tried both and saw no change, because the problem lives in that specific version, not in leftover files.

Where to go from here

The 621.35 hotfix sits behind NVIDIA's Customer Care site only. You won't find it in the NVIDIA App or the usual download portal, and it isn't WHQL certified. NVIDIA's own boilerplate is blunt about this.

"These Hotfix drivers are beta, optional and provided as-is. They are run through a much abbreviated QA process. The safest option is to wait for the next WHQL certified driver."

So there's the tension. The urgent fix is also the least-tested one. NVIDIA says the correction will eventually land in the next official driver, at which point the hotfix gets pulled.

If you're seeing disk usage pegged at 100% or random stuttering after installing 621.14 or 621.24, a clean roll back to 621.01 is probably your safest bet until a certified patch lands. Otherwise, head here to NVIDIA's Customer Care page to grab the fix, or just wait.