NVIDIA has quietly pulled its latest GeForce Game Ready Driver — version
595.59 WHQL — after users reported a
critical fan-control bug that can cause some RTX 3000-, 4000- and 5000-series cards to report and/or spin only a single fan, prompting the company to remove the downloads and recommend an immediate rollback to the prior stable driver.
Background / Overview
When NVIDIA ships a new Game Ready driver it typically bundles game optimizations, fixes, and sometimes new features for recent titles. The 595.59 release was explicitly positioned as the Game Ready package for Resident Evil: Requiem, including support for path tracing, DLSS 4 with Multi-Frame Generation and other modern RTX features. Hours after publication, however, NVIDIA posted an update saying it had “discovered a bug” in both the Game Ready and Studio 595.59 packages and had temporarily removed the downloads while the engineering team investigates. The company’s advisory also advised affected users to roll back to
591.86 WHQL.
Independent reporting and community threads quickly amplified the severity of the problem: multiple users across forums and social platforms reported that hardware monitoring tools and vendor utilities suddenly displayed only one fan sensor, that manual fan profiles stopped being applied correctly, and in some cases one or more physical fans did not spin when expected — an issue that increases the risk of thermal runaway under load. The thread responses and tests indicate the bug affected boards from Founders Edition to many third‑party designs, spanning the RTX 3000 through RTX 5000 families.
What happened — technical symptoms reported
How users saw the issue
- Hardware monitoring apps such as HWiNFO, GPU-Z, and vendor tools suddenly reported a single-fan reading where multiple fans had been shown previously.
- In some systems, manual fan curves (set in MSI Afterburner or vendor utilities) were ignored or only applied to a single fan header, leaving the other fans at idle.
- Reports describe fans not increasing RPM under load, fans stuck at low speed while temperatures rose, or abrupt changes to boost behavior (voltage and frequency capping) after installing the driver. Community troubleshooting noted that rolling back to 591.86 restored expected sensor and fan behavior for many affected users.
Severity and scope
The core concern is that fan-control and sensor-reporting failures may prevent a card from increasing cooling under load, which can produce thermal throttling, system instability, or — in extreme, poorly monitored scenarios — hardware damage. Although not every installation produced catastrophic outcomes, the combination of a driver that interferes with thermal control and the near-infinite variety of PC configurations made the bug dangerous enough for NVIDIA to remove the package from circulation.
NVIDIA’s official response and immediate guidance
NVIDIA’s post announcing the Game Ready driver included an update at 11:00am PT the day the driver was released: the company acknowledged the bug, removed the downloads temporarily, and instructed users experiencing fan-control issues to roll back to
591.86 WHQL. NVIDIA also explained how to reinstall the previous driver via the
NVIDIA App by clicking the three dots in the Drivers tab, and recommended users post detailed reports to the Driver Feedback Forum if they continued to experience problems.
Community-circulated copies of NVIDIA’s advisory and related threads reproduced the message and accelerated the rollback advice across forums and social channels. Independent outlets and hardware communities confirmed the downloads had been pulled from the website and that the package no longer appeared in the in‑app driver listings.
Practical impact for users — what to watch for now
If you installed 595.59 or saw it appear in the NVIDIA App, verify your GPU cooling and sensor reporting immediately. Look for these red flags:
- Hardware-monitoring tools showing fewer fans than your card actually has (e.g., 1 sensor reported for a 3‑fan card).
- Fan RPMs stuck at a low value under heavy GPU load while GPU temperature climbs.
- System instability: black screens, driver crashes (VIDEO_TDR_FAILURE), or abrupt reboots after driver install. Community reports include black screens and TDR events tied to the driver branch.
- Lower-than-expected boost clocks or voltage capping (some users reported the driver clamped voltages to around 0.95V, limiting peak frequency on top-end cards).
If any of the above occur, treat it as urgent. Thermal issues deserve immediate remediation — do not leave a potentially under-cooled system running unattended under load.
Step-by-step mitigation: how to protect your hardware (immediate to follow)
- Verify driver version:
- Open the NVIDIA App or Device Manager → Display adapters → GPU properties → Driver tab. Confirm whether 595.59 is installed. If installed, proceed to rollback.
- Roll back via NVIDIA App (fastest if you have the app):
- Open NVIDIA App → Drivers tab → click the three-dot menu next to the driver → select Reinstall previous driver (or Rollback). Follow prompts and reboot.
- If you don’t use the NVIDIA App — Windows Device Manager rollback:
- Open Device Manager → Display adapters → double‑click GPU → Driver tab → click Roll Back Driver. Reboot if prompted. If the Roll Back Driver button is greyed out, continue to the next step.
- If rollback isn't available or issues persist — use a clean uninstall:
- Boot to Windows normal mode, run DDU (Display Driver Uninstaller) in safe mode to remove all NVIDIA driver traces, then reinstall a known-good driver (591.86 as of NVIDIA’s advisory). DDU is the de-facto community standard for clean driver removal. Multiple community reports specifically advise running DDU when the in‑place rollback does not fully restore behavior.
- After reinstall:
- Confirm multiple fan sensors are visible in HWiNFO/AIDA64/GPU-Z.
- Test with a controlled GPU load (unigine or a stress utility) while watching temps and fan RPMs. If fans still fail to engage, power down and check the card’s hardware connections; if fans are physically not spinning and reinstall/rollback does not fix it, do not continue to use the system for heavy loads — contact the card vendor for RMA guidance.
Deeper technical note: why fan-control bugs can be especially dangerous
Modern discrete GPUs split thermal control between several layers:
- The GPU BIOS (VBIOS) often contains a default fan curve for automatic operation that will run independently of the OS/driver.
- Drivers expose interfaces that let software set manual fan curves or allow vendor utilities (Asus/Gigabyte/MSI) and third‑party apps like MSI Afterburner to control fan speeds through the driver.
- Sensor reporting (so that HWiNFO, GPU-Z, and vendor tools can show RPMs and temperatures) is mediated by the driver and low-level APIs.
A driver bug that either
stops reporting certain fans or
breaks the manual control path can leave a card running on reduced or static cooling behavior — particularly problematic if your system relies on a custom manual curve rather than the card’s VBIOS defaults. Community posts suggest that with 595.59, automatic BIOS-driven fan control often still worked while manual curves (driver-level control) misbehaved — which explains why some users saw only manual-profile‑controlled fans stop responding.
Short-term and systemic risks
- Hardware damage: sustained high GPU temperatures can, in extreme situations, cause component failure or accelerate degradation. While modern cards have protections, relying on those protections is not a substitute for correct fan behavior.
- Data loss / instability: driver faults causing black screens and reboots risk interrupting writes, corrupting files, or losing unsaved work.
- Creator workflows: streamers and content creators reported encoder (NVENC) and OBS/streaming tool regressions tied to this driver family in community reports, meaning the fallout isn't limited to gaming.
- Trust erosion: frequent or high‑profile driver regressions damage user trust and increase the friction users face when choosing to update.
For these reasons NVIDIA appears to have opted for rapid removal rather than a slower fix-in-place rollout.
How we verified the facts and cross‑checked reports
This feature draws from NVIDIA’s own driver release note and update (the company’s post explicitly confirms removal and rollback guidance), contemporaneous reporting from major outlets, and aggregated community feedback on forum and subreddit threads. Key corroborating references include NVIDIA’s release update (which added the removal notice), Tom’s Hardware coverage of the incident and community reports, VideoCardz’s reporting of the “unlaunch” and observed symptom set, and detailed discussion threads on ComputerBase that show user troubleshooting and direct experience with fan sensor changes. Together these sources provide a consistent picture: the driver was distributed, community reports of fan/sensor problems surfaced quickly, and NVIDIA removed the package and advised rollback pending investigation.
Additionally, this article references historical context and past NVIDIA emergency fixes to illustrate that the company has gone through similar rapid hotfix cycles before; internal forum archives and release notes from prior hotfixes were reviewed to confirm that pattern.
Why this keeps happening: release cadence and QA friction
There are structural reasons driver releases sometimes regress behavior on a subset of hardware:
- A single driver build must support a vast matrix of GPUs (multiple architectures and board partners), OS builds, firmware revisions, and third‑party utilities. That multiplies test permutations far beyond what can be exhaustively validated in time-sensitive releases.
- The industry’s release cadence ties driver drops to major game launches; shipping “Game Ready” drivers under time pressure increases the risk of regressions slipping through if automated tests don’t cover every vendor utility and fan-control path.
- Modern drivers interface with firmware and vendor-controlled components; a change intended to improve one stack (e.g., power or telemetry) can have unexpected side effects on fan sensor enumeration or manual-control APIs.
- Finally, community evidence and commentary suggest some users suspect increased automation or AI-assisted processes in driver generation and QA — while plausible, such assertions are speculative and not confirmed publicly by NVIDIA. What is certain is that the volume of changes and velocity of delivery raises the chance of regressions.
Recommendations for NVIDIA and board partners (what should change)
- Staged rollout: reintroduce a phased distribution model by default — small opt‑in groups before full public release — to capture edge cases in real-world environments.
- Better vendor‑utility testing: coordinate with board partners and third‑party utility authors to include fan-control permutations and manual‑curve scenarios in official pre‑release test matrices.
- Transparent post‑mortems: when serious regressions occur, publish a clear post‑mortem that outlines the root cause and the fix schedule. That improves user trust and reduces speculation.
- Improved telemetry opt‑in: allow consenting users to provide anonymized telemetry specifically for fan/control regressions to accelerate root-cause analysis while preserving privacy.
- Faster hotfix channels: maintain a tightly controlled hotfix distribution path that’s easy to access for affected users but won’t be recommended for the general population until validated.
These steps won’t prevent every regression — but they would measurably reduce the likelihood and scope of dangerous issues like a fan-control failure.
Advice for enthusiasts and system builders
- Treat drivers tied to major launches as “optional” until community feedback is available for 24–72 hours. When a driver is explicitly tagged "Game Ready" for a major title, wait to install it on production or critical rigs until the dust settles.
- Maintain a known-good driver image or installer (for example, store 591.86 locally if that’s the stable baseline for your system).
- If you rely on manual fan curves for acoustics or water‑cooling balance, prefer the card’s BIOS/auto curve during high-risk windows, or test driver updates on a non-critical machine first.
- Use DDU to cleanly revert when things go wrong; a normal uninstall sometimes leaves traces that continue to interfere with hardware monitoring.
- Keep backups of work and enable auto-save features in creative apps — driver crashes and sudden reboots can cost more than just a few minutes of time.
What to expect next
NVIDIA’s rapid removal of 595.59 signals the company is treating the issue seriously: fan-control problems present tangible hardware risk and therefore demand a cautious approach. Expect NVIDIA to:
- Investigate the root cause and issue a hotfix or rebuilt WHQL driver after internal validation. NVIDIA has done similarly for earlier conflicts involving Windows updates and driver regressions.
- Repost a corrected driver once the fix is validated, or release a narrowly scoped hotfix branch before the next full WHQL rollup.
- Monitor public forums and the Driver Feedback Forum to triage lingering cases not addressed by a hotfix.
From a user perspective, the safest path for now is to follow NVIDIA’s rollback guidance, validate fan behavior, and avoid installing 595.59 should you still have access to it.
Final analysis: strengths, failures, and the broader lesson
This episode illustrates two competing realities of modern PC hardware:
- Strength: NVIDIA’s quick removal of the problematic drivers and clear rollback instruction shows a willingness to act quickly when a release introduces risk. Pulling the driver and recommending a rollback is the responsible emergency response.
- Failure: The release itself demonstrates gaps in pre-release validation for driver-level fan/sensor interactions across vendor ecosystems. When a driver can break thermal-control paths, the release process has failed in a fundamental safety dimension. Wide hardware and software permutations make exhaustive local QA infeasible, which is why staged rollouts and coordinated vendor testing are essential.
Broader lesson: in today’s tightly integrated hardware-software stacks,
drivers are effectively firmware for the whole platform. Treating them as minor updates increases systemic risk. Vendors must balance the urgency of shipping game optimizations and feature updates against the very real possibility that a single regression can compromise hardware safety.
Conclusion
If you updated to 595.59, roll back immediately and verify that all fan sensors and RPMs are visible and functioning. Monitor community threads and NVIDIA’s Driver Feedback Forum for the company’s follow-up. NVIDIA has acknowledged the problem and removed the downloads — that’s the right first step — but the frequency and nature of these high‑impact regressions demand a longer-term shift to safer distribution practices and more exhaustive real‑world validation before broad releases.
Source: Tom's Hardware
Nvidia rolls back its latest driver update — Game Ready Driver 595.59 reportedly causes fan issues on RTX 3000, 4000, and 5000-series GPUs