On November 12 — the exact day AMD shipped Adrenalin 26.11.1 to stop Radeon drivers from chewing through idle SSDs — NVIDIA shipped GeForce 621.24 carrying an unacknowledged background logging loop that writes up to ten times harder once you start Alt-Tabbing out of a race.

TL;DR

  • The broken branch: NVIDIA GeForce Game Ready Drivers 621.14 (November 4, 2026) and 621.24 (November 12, 2026) can lock NVDisplay.Container.exe into continuously rewriting a binary log file (nvlog.nvlgstg) on C:.
  • Real NAND writes, not phantom counters: Community telemetry matched against CrystalDiskInfo shows ~3 GB/hour at pure desktop idle — escalating to 30 GB/hour (and 70+ GB in marathon sessions) under gaming loads, with 500 MB to 1 GB dumped on every Alt + Tab.
  • Not every PC triggers it — measure first: Before wiping drivers, check NVDisplay.Container.exe in Windows Resource Monitor (resmon.exe) → Disk for two minutes; unaffected rigs log under 2 MB per 5 minutes.
  • The rollback trap on RTX 50-series: Most headlines tell everyone to downgrade to 621.01 WHQL (October 21), forgetting that 621.01 caused browser/video black-screen freezes on RTX 50-series cards — RTX 5070/5080/5090 owners should test Hotfix 621.12 (October 27) or step back to 610.88 WHQL instead.
  • Don’t blindly disable NVIDIA Display Container LS: Killing the service permanently breaks the NVIDIA App and modern Control Panel settings.

What’s actually happening: a staging log that rewrites itself from scratch

Three days ago, in our Forza Horizon 6 AMD Adrenalin 26.10.1 rollback guide, we noted a strange footnote: AMD’s optional November 12 driver (26.11.1) ignored Forza Horizon 6 crashes entirely because AMD had to rush out a fix for atieclxx.exe copying itself every second and writing 0.3 to 5 MB/s to system SSDs.

Over the past 48 hours, forensic posts on NVIDIA’s official driver feedback forums — led by users Andergraude (on 621.14) and AyzekUorren (on 621.24), alongside reports compiled by TroubleGrid, Tom’s Hardware, and TechPowerUp — exposed an eerily similar regression on Team Green’s side.

Starting in GeForce 621.14 WHQL (November 4) and carrying straight into 621.24 WHQL (November 12), the background process NVDisplay.Container.exe (tied to the NVIDIA Display Container LS service) can enter a loop targeting one specific file:

C:\ProgramData\NVIDIA Corporation\nvtopps\nct\nvlog.nvlgstg

Instead of appending small new entries to the end of the log, the process truncates and rewrites the entire binary file on every pass. That detail explains two ugly behaviors Andergraude verified against CrystalDiskInfo‘s hardware Data Units Written counter:

  1. The write rate snowballs as your session runs. On a fresh boot at idle, 621.14 generated 515 MB of process I/O and 270 MB of verified physical C: drive writes every 5 minutes (~3.2 GB/hour), against less than 2 MB per 5 minutes on 621.01. As nvlog.nvlgstg grows during gameplay, rewriting the whole file multiplies the throughput — pushing 621.24 to 30 GB/hour and 70+ GB across longer sessions.
  2. Telemetry toggles and DDU reinstalls don’t cure it. Turning off telemetry inside the NVIDIA App did not stop the loop on affected rigs, and three consecutive Safe Mode Display Driver Uninstaller (DDU) clean installs of 621.14/621.24 reproduced the exact same writes. As of November 18, NVIDIA has not listed the bug under Open Issues.

The honest SSD math — and why Forza Horizon 6 players feel it on Alt + Tab

Whenever an SSD-wear story goes viral, half the internet treats it like your NVMe will catch fire before Friday. Run the actual endurance math first so you know what risk you’re managing.

A standard 1 TB PCIe 4.0 NVMe drive carries a 600 TBW (terabytes written) warranty rating over five years — 120 TB per year.

  • At the idle floor (~3 GB/hr, left running 24/7): Roughly 26 to 28 TB per year, or ~4.5% of the drive’s entire rated lifespan spent doing nothing at the desktop.
  • At the active gaming ceiling (30 GB/hr for 8 hours a day): 240 GB per day, or 87.6 TB per year — burning nearly 15% of a 1 TB SSD’s total rated endurance every year on a single disposable log file, at a moment when NAND flash prices are climbing again.

That won’t brick a healthy drive overnight, and pretending otherwise is clickbait. What sim racers and Forza Horizon 6 players actually feel right now is the I/O hitching.

In AyzekUorren’s 621.24 traces, every single Alt + Tab window switch triggered an immediate 500 MB to 1 GB burst of disk writes, accompanied by gameplay and video hitching as the log swelled. Think about how a normal Forza Horizon 6 session works on PC: tabbing out between events to grab share codes from how Horizon CoLab turned EventLab into a social sandbox, checking SimHub or wheel software, or pulling up weekly playlist maps. Combine 1 GB container write bursts on C: with a game engine whose heavy memory and streaming footprint we already profiled in our RTX 5070 Ti telemetry on FH6’s D3D12 shader upload-heap leak, and you get inexplicable mid-session micro-stutters that look like game regressions. After a month where PC players already had to navigate Series 7’s FPS-dependent Drift Zone scoring bug, the last thing your frame pacing needs is your GPU driver thrashing the OS drive.

(Quick diagnostic split: if FH6 refuses to open at all, you want our Forza Horizon 6 Steam startup crash fix; if it silently closes 90 seconds into shader compilation, see why Forza Horizon 6 dies three minutes in. This NVIDIA bug presents as background SSD churn and Alt + Tab hitching across your whole PC.)

Measure first: the 30-second check in Windows Resource Monitor

Epistemic caveat: not every system on 621.14 or 621.24 exhibits the runaway loop. Several users with overlays and data collection disabled report normal idle I/O. Check your own machine before spending ten minutes in Safe Mode:

  1. Press Win + R, type resmon.exe, and hit Enter to open Windows Resource Monitor.
  2. Click the Disk tab and expand Disk Activity (sort by Write (B/sec) descending).
  3. Leave the PC idle for two minutes, then launch a game, Alt + Tab out once, and watch NVDisplay.Container.exe (nvlog.nvlgstg).
  4. Read the verdict: If NVDisplay.Container.exe stays quiet (under ~2 MB across five minutes), leave your driver alone. If it pins hundreds of kilobytes to several megabytes per second continuously — or spikes hundreds of megabytes when you Alt + Tab — roll back.

The rollback trap on RTX 50-series: don’t blindly install 621.01

The part almost every news recap gets wrong today: telling every GeForce owner to roll straight back to 621.01 WHQL (October 21).

On an RTX 30-series or RTX 40-series card, 621.01 is fine. On an RTX 50-series GPU (RTX 5070, 5070 Ti, 5080, 5090), 621.01 introduced hard video-playback freezes and black screens whenever Chrome, Edge, or VLC ran alongside a game — the exact bug that forced NVIDIA to ship GeForce Hotfix 621.12 on October 27, one week before 621.14 broke nvlog.nvlgstg.

GeForce VersionRelease DateSSD Write Loop (nvlog.nvlgstg)Best Choice By GPU Generation
610.88 (WHQL)July 28, 2026CleanUniversal pre-R620 fallback if all 621.x builds throw nvlddmkm.sys TDR crashes
621.01 (WHQL)Oct 21, 2026Clean (< 2 MB / 5 min)RTX 20 / 30 / 40 series — cleanest WHQL rollback; avoid on RTX 50-series (browser video black screens)
621.12 (Hotfix)Oct 27, 2026Pre-621.14 branch (verify in resmon)RTX 50-series — fixes 621.01 video black screens before Nov 4 621.14 (disables Advanced Shader Delivery)
621.14 (WHQL)Nov 4, 2026Affected (~3 GB/hr idle)Avoid if resmon shows nvlog.nvlgstg churn
621.24 (WHQL)Nov 12, 2026Affected (up to 30+ GB/hr under load)Avoid unless locked in by DOOM: The Dark Ages or WoW: Forever

If another November release forces you to stay on 621.24 until NVIDIA ships a hotfix:

  • Disable the NVIDIA App In-Game Overlay and Instant Replay first, which reduced write volume on part of the test pool.
  • Reset the file size before long sessions rather than permanently killing the service: open services.msc, click Stop on NVIDIA Display Container LS, delete C:\ProgramData\NVIDIA Corporation\nvtopps\nct\nvlog.nvlgstg, and click Start. Leaving that service permanently disabled breaks both the NVIDIA App and Control Panel settings (especially painful since NVIDIA retired standalone legacy Control Panel distribution in the 610.xx branch).

The Bottom Line

November 2026 has officially delivered back-to-back SSD-thrashing GPU drivers from both major vendors inside the same week. NVIDIA’s 621.14 and 621.24 nvlog.nvlgstg loop won’t kill a modern NVMe overnight, so skip the panic — what it will do on affected rigs is burn up to 15% of a 1 TB drive’s rated endurance pace for nothing while dumping half-gigabyte write spikes every time you Alt + Tab for a tune code.

Open resmon.exe right now and check the Disk tab for two minutes. If NVDisplay.Container.exe is quiet, go race; if it’s hammering nvlog.nvlgstg, roll RTX 40-series and older cards back to 621.01 WHQL, put RTX 50-series cards on Hotfix 621.12 (or 610.88 WHQL), and hold there until NVIDIA ships 621.24‘s cleanup build.


Related on ApexSpeedCraft: If FH6 Keeps Crashing on AMD 26.10.1, Roll Back | The FH6 Shader Leak Survived the Patch | Why Forza Horizon 6 Dies Three Minutes In | FH6 Keeps Crashing on Startup: Steam Launch Fix | For Three Days, FH6 Scored Your Drifts by Frame Rate | Horizon CoLab: How FH6 Is Turning EventLab Into a Social Sandbox

Sources: Tom’s Hardware — NVIDIA 621.14 & 621.24 30 GB/Hr SSD Write Report | TechPowerUp — Latest NVIDIA Drivers Excessive SSD Writing | VideoCardz — NVIDIA GeForce Continuous Idle SSD Writes | NVIDIA Forums — GeForce GRD 621.14 Feedback Thread (Andergraude)

Reporting and analysis by ApexSpeedCraft. Full credit to NVIDIA forum contributors Andergraude and AyzekUorren and tracker TroubleGrid for the initial CrystalDiskInfo and resmon traces across GeForce builds 621.01, 621.14, and 621.24 (verified as of November 18, 2026).

Leave a Reply

Trending