GPU Fan Not Spinning at Idle: Is Zero RPM / Fan Stop Normal or a Fault?

2026-10-19 ｜ GPU Repair ｜ 13 min read ｜ HK GPU Fix Editorial

Key Takeaways

A GPU fan that stays still at idle is often normal fan stop, not a fault. The reliable test is behaviour under load: if the fan starts as temperature and power rise, it is normal. If temperature keeps climbing and the fan stays at 0 RPM, stop using the card under load and work through the checks in this guide.

What Zero RPM / Fan Stop Actually Is

Modern graphics cards commonly include a fan-stop feature: at low load and low temperature, the fans stop completely for silence; when temperature or power draw reaches the start condition, the fans start again automatically and follow a fan curve. Different brands use different names for the same idea: MSI calls it Zero Frozr, ASUS calls it 0dB, GIGABYTE calls it Fan Stop, and AMD calls it Zero RPM. The function is essentially the same, but the thresholds vary by model.

The mechanism sits in the card itself. The card's own fan controller — firmware and the fan curve and fan-stop logic in the vBIOS — decides whether the fans spin and how fast, based on GPU core temperature and, on some models, power draw as well. Below the stop condition, fan power is cut and speed falls to zero, and cooling relies entirely on passive cooling: the heatsink fins and the airflow inside the case carry away the small amount of heat the GPU still produces at idle. When the start condition is reached, the fans spin up and follow the fan curve.

The stated design purpose, in the manufacturers' own words, is quiet operation and power saving at low load. MSI describes Zero Frozr as stopping the fans under low load to eliminate fan noise. ASUS describes 0dB as stopping the fans when the GPU core temperature meets the condition, so the card runs silently. GIGABYTE describes Fan Stop as stopping the fans at low temperature to save power and reduce noise.

A point that many informal guides get wrong is that the start and stop conditions are not the same temperature. There is hysteresis — a gap between them. The start threshold is higher than the stop threshold, so the fans do not switch on and off repeatedly when the temperature hovers near a single boundary. ASUS's official table is a clear example: for several models the fans start above 55 °C and stop below 50 °C, leaving a 5 °C gap. MSI Afterburner's Temperature Hysteresis setting exists for the same reason (see the settings section below). When explaining fan stop, always describe two thresholds with a buffer between them, never a single switch temperature.

Another point that informal guides often miss is that temperature alone is not always the whole condition. On the ASUS RTX 30, 40, and 50 series, the official conditions use temperature OR power for starting, and temperature AND power AND 0dB enabled for stopping. That means the fans can start even when temperature is not high, if power draw exceeds the power threshold; and conversely, the fans may not stop even after temperature falls, if power draw is still high or if 0dB mode has been switched off. Any guide that only mentions temperature is incomplete for these models.

Passive cooling while the fans are stopped

While the fans are stopped, the GPU still produces heat — just less of it. The heatsink fins and the air movement inside the case carry that heat away. If case airflow is poor or the room is warm, idle temperature climbs slowly, the fans start, cool the card, then stop again, and the cycle repeats. That repeated start-stop cycling is not necessarily a fault, but it can be annoying. The remedy is usually fan-curve and hysteresis settings plus case airflow, not a replacement fan. Case airflow itself is covered in depth in the companion guide on case airflow and GPU temperatures; this article only needs the principle: while the fans are stopped, the heatsink and case air are doing all the cooling work.

The brief spin at boot is a self-test, not a fault

Many readers worry when the fans spin briefly at power-on and then stop. MSI's official documentation says the fans should spin during the power-on detection of the graphics card. Once the system is running, if the Zero Frozr conditions are met, the fans stop. ASUS's official troubleshooting also separates cards with 0dB from cards without it, rather than treating any stop as a fault. A brief spin at boot followed by a stop is therefore often normal. The absence of even that brief spin at boot is worth noting — MSI's guidance treats it as a reason to contact MSI customer service — but it is not by itself a diagnosis, because boot behaviour can differ between models. Do not write it as "no spin at boot means the card is dead."

Do not confuse graphics card fan stop with motherboard fan stop

Graphics card fan stop is controlled by the card itself — its own controller, firmware, and the card manufacturer's software. It does not go through the motherboard BIOS. Case fans and CPU fans are controlled by the motherboard BIOS (Q-Fan, Smart Fan, and similar) or motherboard software. GIGABYTE motherboard manuals describe a "Fan Stop" that refers to motherboard fans, not the graphics card. Before telling anyone to "turn off fan stop in the BIOS," be clear: graphics card fan stop is generally not a motherboard BIOS setting. Do not send readers into the motherboard BIOS looking for a graphics card fan-stop option.

Official Per-Model Start and Stop Conditions

There is no universal threshold that applies to all cards. The numbers below are verified examples for specific series and models, each with its source. Always check the official page for the exact model in hand. Do not apply one model's numbers to another card.

MSI — Zero Frozr

MSI's official support documentation states that most MSI graphics cards have Zero Frozr, that the fans stop completely when the temperature falls below 60 °C, that the function is controlled in the operating system environment, and that the fans should spin during power-on detection of the card. If the fans do not spin at all even at power-on, MSI directs users to contact the MSI customer service centre rather than disassembling the card.

To check whether a specific card supports the feature, MSI directs users to search for the model on the MSI website, open Specifications, then the datasheet, and look for the Zero Frozr icon in the features section. The switch is in MSI Center, in the Cooling Wizard module, under "Zero Frozr AI Cooling." Older models used Dragon Center, which also had a Zero Frozr switch, enabled by default.

One caution: MSI's "below 60 °C the fans stop" is a general statement in MSI's official FAQ, not a precise start and stop table for every card. Do not present 60 °C as the start temperature for all MSI cards. Start and stop temperatures can differ because of hysteresis.

ASUS — 0dB (the most complete official threshold table)

ASUS's official FAQ on fan start and fan stop conditions for the RTX 30, RTX 40, and RTX 50 series gives the following conditions. Start requires either condition; stop requires all conditions at once.

ASUS adds that actual wattage and temperature may vary with product configuration, usage, operating conditions, and power management settings. That note matters: the table is the official reference, but real readings on a given machine can sit slightly either side of it.

ASUS's official troubleshooting for a fan that stops when booting gives the following steps. For cards with 0dB technology, check GPU temperature in ASUS GPU Tweak II or III; the fans should restart automatically as temperature rises. If the start condition has been met and the fans still do not spin, first update the driver to rule out a driver problem, then shut down, disconnect power, remove the card, and use a fine brush to clean dust that may be jamming the fan. If the fan is severely jammed and will not move, contact ASUS product support. Cards without 0dB technology follow the same driver-update, dust-cleaning, and contact-support steps.

ASUS's official warnings must be stated plainly: using a graphics card with a faulty fan may damage the card, and once a fan is found to be abnormal, the card should stop being used. And: do not disassemble the graphics card yourself, as improper disassembly may void the warranty.

GIGABYTE — Fan Stop (thresholds differ by model)

GIGABYTE's official FAQ on whether the fan stopping at boot is normal exists in two different versions for different models and generations — which itself proves that thresholds vary by model:

• Version A: when GPU temperature is below 57 °C, the fans stop, and the Silent / Stop LED lights up to indicate the fans are not moving; above 58 °C, the fans start cooling, and the LED goes out to indicate the fans are running. The LED is controlled by internal circuitry.
• Version B (a different model): below 44 °C the fans stop, and above 45 °C they start, with the same LED logic.

The right way to use these two versions is as a contrast: two official GIGABYTE answers, more than ten degrees apart, for different cards. Never present only one of them as the GIGABYTE standard, and never apply another card's numbers to the card in hand.

The software switch, as described in third-party guides rather than in the official FAQ text, is in AORUS Engine, under the Fan tab, where Fan Stop can be ticked on or off. Newer software is GIGABYTE Control Center (GCC), which can adjust fan mode, clocks, voltage, and power target, and includes a one-click silent fan setting. Because menu positions vary by version, this guide only states that Fan Stop is switched in the fan settings of GCC or AORUS Engine, and that the exact location depends on the installed version. No precise menu path is invented.

AMD — Zero RPM (Radeon cards)

In AMD Software: Adrenalin Edition, Performance, Tuning, Fan Tuning allows a custom fan curve and switching Zero RPM on or off. With Zero RPM on, the fans stop completely when the GPU is cool; with it off, the fans keep running at a minimum speed. The earlier Radeon WattMan already offered temperature-linked fan curves with Zero RPM fan control.

Driver problems can also turn normal fan stop into abnormal fan stop, and one documented case is worth including with its version numbers. Users reported that Adrenalin 26.5.1 had a bug in which, after the system woke from sleep, Zero RPM stayed stuck: the fans would not start even under load, and temperature rose quickly. Restarting temporarily restored normal behaviour until the next sleep cycle. AMD listed the fix in the Adrenalin 26.6.1 release notes: "Zero RPM mode may re-enable itself after a monitor goes to sleep or is turned off on Radeon RX 9000 series graphics products." The lesson is general: driver and software state belong in the diagnostic checklist, and not every fan that fails to start under load is a hardware fault.

Name reference

How to Monitor: Temperature, Load, Power, and RPM Together

A fan-stop question cannot be answered from a single number. Every observation should record the same four readings, plus a visual check:

1. GPU core temperature (°C)
2. GPU load (%)
3. Power draw (W) — needed because the ASUS stop conditions include power, so a judgement without power is incomplete
4. Fan speed (RPM, and percentage) plus whether the fan is actually seen moving

Software-reported fan percentage is not the same as the fan actually spinning. There are user reports of software showing 20–30% while the fan is visibly not moving, with the fan only starting at a much higher percentage. That mismatch is itself a fault signal. Always cross-check what the eyes see against what the software reports; never trust a single number alone.

GPU-Z (TechPowerUp, free, lightweight)

Open GPU-Z, go to the Sensors tab, and read GPU Temperature, Fan Speed (usually shown as both RPM and percentage), GPU Load, and Power Draw as live readings. Each sensor has a small arrow that can select "Show Highest Reading," which is useful for capturing the peak temperature during a load session. Data can also be logged to a file for later comparison.

One practical trap: GPU temperature falls quickly after a game is closed or the window is switched away from, so readings must be taken while the load is still running, or the highest-reading function must be used. Do not close the load, open GPU-Z afterwards, and conclude that "temperature is low." And do not treat GPU-Z as a fan-control tool: GPU-Z is a monitoring tool. Fan control belongs to Afterburner or the manufacturer's software.

HWiNFO (HWiNFO64, free, with a portable version)

Open HWiNFO, choose "Sensors only," scroll to the graphics card section, and read GPU Temperature (core temperature), GPU Hot Spot Temperature, GPU Memory Temperature (VRAM temperature), GPU Fan Speed (RPM), power draw, and clocks. HWiNFO's advantage over GPU-Z is breadth, especially Hot Spot and VRAM temperature: a normal core temperature does not mean the whole card is cool. Core temperature and Hot Spot temperature are two different readings; this guide notes the distinction without going into Hot Spot mechanisms, which are outside its scope.

HWiNFO can also place a sensor in the system tray for live viewing, and can log sensors to a CSV file, which helps with intermittent problems — for example, a fault that only appears after twenty minutes of gaming and is not visible in an immediate check. HWiNFO is primarily a monitoring tool; fan control should use the manufacturer's software or Afterburner.

Windows Task Manager (built-in, quick sanity check only)

Press Ctrl+Shift+Esc, open Performance, then GPU, and read GPU Temperature. The limitation must be stated plainly: Task Manager shows only core temperature, with no Hot Spot, no VRAM temperature, and no fan speed. It cannot distinguish a fan-stop problem. Do not use Task Manager alone for this diagnosis.

Load testing: creating a condition where the fan should spin

To test whether the fan starts, the GPU must be given real work: a game's built-in benchmark, 3DMark, FurMark, or GPU-Z's built-in render test. The purpose is to create a condition in which the fan should start, then watch whether it does.

The safety warning is mandatory. If the fan is already suspected faulty (not spinning under load), watch temperature continuously during any load test and stop immediately if temperature keeps rising abnormally. Do not "run it to the end." ASUS's official guidance also warns that a card with a faulty fan should stop being used. Never ask a reader to run a long stress test with the fan not spinning. No "safe temperature limit" figure is given for FurMark or 3DMark, because limits differ by card and no universal standard exists.

Diagnosis: Idle Observation to Inspection, in Order

Follow the steps in order. Do not skip ahead to disassembly.

Step 1: Idle observation (leave the machine alone at idle)

Record temperature, load, power draw, RPM, and whether the fan is visibly moving. If temperature is low, load is low, power draw is low, and RPM is 0, the card is probably meeting its fan-stop conditions. Do not declare it faulty yet. Move to Step 2.

Step 2: Load observation (give the GPU real work)

Run a game or benchmark while monitoring. If temperature and power rise to the start conditions and the fans start, with speed rising as temperature rises, that is normal fan stop, and the diagnosis is complete. If temperature keeps climbing and the fans remain at 0 RPM, move to Step 3 — a fault or settings problem.

Step 3: Rule out software and settings (no disassembly)

1. Check whether fan stop is enabled (manufacturer software, Afterburner, or Adrenalin). Try switching fan stop off and see whether the fan immediately runs at a minimum speed.
2. Restore the fan curve to default or automatic. Close all other fan-control tools and leave only one in control.
3. Update the graphics driver to the latest version. Updating the driver is the first step in ASUS's official troubleshooting, and AMD has a documented history of a driver bug that left fan stop stuck.
4. Check for the "software shows speed but the fan is visibly still" contradiction described in the fault-signal list below.

Step 4: Rule out physical obstruction (power off first)

Shut down, disconnect power, and remove the card if needed. Use a fine brush to clean dust that may be jamming the fan, following ASUS's official wording. Check for cables touching the blades, a sagging bracket pressing on the fan, or damaged blades. Flick a blade gently by hand: it should turn smoothly, with no obvious resistance or scraping noise.

Check the fan power cable: the fan cable runs from the graphics card PCB to the fan assembly. If it is loose or broken, the fan stops immediately. Check that the plug is firmly seated, that no pin is bent or broken, and that the cable is not damaged. Reseat it and test again.

Check the PCIe power connectors: a loose or incompletely seated PCIe power plug can leave the card lit but behaving abnormally. RGB lighting being on does not prove that power delivery is complete — this is a common misconception covered in the misconceptions section below.

Step 5: If none of the above works — stop using the card under load and send it for inspection

A card with a confirmed fan fault carries a risk of damage if it keeps being used under load (ASUS official warning). Do not disassemble the graphics card yourself, which may void the warranty. Do not flash the vBIOS. In any suspected short-circuit, burning smell, or smoke situation, do not power the card on again (see the safety section).

When sending the card for inspection, bring the observation records — the four idle readings, the four load readings, and which steps were already tried. That is what lets an inspector judge whether the problem is the fan motor, the power cable, the fan power circuitry or PWM control, or something else.

Fault signals: any one of these means it is not normal fan stop

When only one fan does not spin

First, know that some multi-fan graphics cards control each fan independently: one fan may stay stopped below a certain temperature while another starts earlier. This design exists, but not every card works this way. "One fan not spinning" therefore cannot by itself be called normal or faulty; the load response decides.

Fault signals for a single fan are: the same fan never spins under load, temperature is high, the fan stutters without starting, there is a grinding noise, the fan does not respond to manual control, while the other fans spin normally under the same command.

Troubleshooting follows the same order: check software settings first, then check obstruction and that fan's power cable and plug. If it still does not spin, contact the retailer or manufacturer within the warranty period. Out of warranty, send it for inspection; do not replace the fan yourself, which may void the warranty or damage the card.

Do not write "a three-fan card must always spin all three fans" or "one fan not spinning is always normal" — both are over-generalisations. The correct statement is that it varies by model design and must be judged by the load response.

Differential diagnosis summary (possible directions, not confirmations)

Division of labour with the PWM IC guide

Fan power circuitry and PWM control IC faults are one of the differential diagnoses above. The mechanism, testing, and repair-level detail for PWM IC and fan power circuitry faults live in the companion guide on PWM IC faults. This article lists that diagnosis and points to the companion guide; it does not repeat the full mechanism.

Settings: Switching Fan Stop On and Off, and Setting a Fan Curve

General principles (read before changing anything)

• Use one fan-control tool at a time. Running ASUS GPU Tweak plus MSI Afterburner plus GIGABYTE Control Center plus AMD Adrenalin at once lets them overwrite each other. A curve that "was saved but has no effect" is often caused by exactly this. When testing, close all the other tools and leave one.
• Apply and save the profile. In Afterburner, after changing the curve, click Apply, then save to one of the profile slots. For automatic application at startup, enable "Start with Windows" and apply again.
• Restore defaults first, then test one change at a time. If a custom curve is suspected, restore fan control to default or automatic first, then observe idle and load behaviour. Do not stack more custom settings on top of a pile of custom settings.
• Do not look for graphics card fan stop in the motherboard BIOS. The graphics card fan is controlled by the card itself. Motherboard BIOS Fan Stop, Q-Fan, and Smart Fan settings control case and CPU fans, not the graphics card.
• Do not flash the vBIOS to turn off fan stop. Forum answers are explicit: the only BIOS-level way to disable fan stop is to modify the BIOS and reflash it, which is not recommended for beginners, and a bad flash requires knowing how to recover a bricked card. A software-level override in Afterburner is sufficient; the vBIOS should be left alone. This guide states plainly: do not flash the vBIOS yourself.

MSI Afterburner fan curve (the most common third-party tool)

Path: open Afterburner, click the gear icon on the left for Settings, go to the Fan tab, and tick "Enable user defined software automatic fan control."

The curve graph has GPU temperature on the horizontal axis and fan speed percentage on the vertical axis. Clicking a blank spot on the line adds a new control point, and each point can be dragged. To make the fan start turning at a given temperature, set a non-zero percentage at that temperature point.

Two activation conditions are commonly missed: the gear icon next to the fan slider on the main interface, and the "Auto" label on the right, must both be on before the curve takes effect.

Zero RPM-related options: the Fan tab includes options such as "Override zero fan speed," and "Use firmware control mode." "Override zero fan speed with hardware curve" hands the 0 RPM portion back to the card's own hardware curve instead of forcing the software points. These options behave differently by card; try them one at a time and check the load response. Do not present any of them as "click this and it will definitely work."

Temperature Hysteresis is in the same Fan settings window. It requires temperature to fall a set amount below the start temperature before the fan stops, preventing repeated on-off switching when temperature hovers near the threshold. As an illustrative example from third-party guides — not an official value for any card — with hysteresis set to 10 °C and a start temperature of 50 °C, the fan would wait until temperature fell to about 40 °C before stopping.

Fan minimum speed: some fan motors have a physical minimum speed. A curve point set too low (for example 20%) may fall below the minimum the motor needs to keep turning, producing a fan that stutters, spins for a second then stops, or clicks. Third-party guides suggest raising the lowest curve point slightly (for example from 20% to 30%) and watching the RPM reading. This guide states it as "if stuttering appears, try raising the lowest point slightly," not as a precise percentage standard.

Frequent start-stop cycling: flatten the low-temperature part of the curve (avoid a steep climb right around idle temperature), or use hysteresis to widen the gap between start and stop. Do not teach readers to lock the fan at 100% permanently — that is noisy, accelerates bearing wear, and is the wrong fix.

Manufacturer software switches

MSI Center (Zero Frozr), ASUS GPU Tweak II / III (0dB), GIGABYTE AORUS Engine / GCC (Fan Stop), and AMD Adrenalin (Zero RPM) each have their own switch, as listed in the name reference table. Positions vary by software version; use the installed version's interface as the reference, and do not invent precise menu paths.

If Afterburner and manufacturer software fight each other, try switching fan stop on or off in the manufacturer software first, then adjust the zero-speed handling in Afterburner, simplify the curve to a few clear control points, and test idle and load behaviour again.

Common Misconceptions, Dismantled One by One

1. "Fan not spinning means it is broken." Wrong. Idle fan stop is a design feature. Judge by temperature plus power plus load response.
2. "RGB lighting on means power delivery is fine." Wrong. Third-party guides note that a card's lights being on does not prove it has received complete power delivery. A loose PCIe power plug can leave the card lit but behaving abnormally. Check that the PCIe power plugs are firmly seated.
3. "Software showing a fan speed means the fan is really spinning." Wrong. The software percentage or RPM is a control command or reported value. There are user reports of 20–30% displayed while the fan is visibly still. Cross-check with the eyes.
4. "All cards start at the same temperature." Wrong. Thresholds vary by brand, series, and model. GIGABYTE's two official FAQ versions already give two different thresholds, and ASUS's different models have different start and stop conditions. Check the official page for the exact model.
5. "Turn off graphics card fan stop in the motherboard BIOS." Wrong. Graphics card fan stop is controlled by the card itself, not through the motherboard BIOS. Motherboard BIOS Fan Stop controls case and CPU fans.
6. "Spin at boot then stop means it is broken." Wrong. A brief spin at boot followed by a stop is often normal fan stop engaging. The absence of any spin at boot is the observation worth noting — and even then it must be combined with other observations, not used alone as a diagnosis.
7. "Fan cycling on and off means it is broken." Not necessarily. When temperature hovers near the start and stop thresholds, the fan cycles on and off. That comes from too small a hysteresis gap or from poor case airflow letting temperature drift up and down. The remedy is curve and hysteresis settings plus airflow, not an immediate fan replacement.
8. "Locking the fan at 100% is safest." Not a good idea. Constant full speed is noisy and accelerates bearing wear. The correct approach is a sensible curve that speeds the fan up when needed.
9. "Flashing the vBIOS fixes every fan problem." Wrong and dangerous. Forum answers state that modifying and reflashing the BIOS is not recommended for beginners, and a bad flash requires knowing how to recover a bricked card. A software-level override is sufficient.
10. "Run a stress test and see." Only with a condition attached. If a fan fault is already suspected, a stress test must be watched continuously, and must be stopped the moment temperature rises abnormally. Never run a full stress test with the fan not spinning.

When Not to Keep Trying It Yourself

• Burning smell, smoke, scorch marks, or a visibly burnt component: disconnect power immediately. Do not power on again. Do not run a load test. A suspected short circuit or burnt component can get worse with continued power.
• Confirmed fan fault (fan does not spin under load): do not keep gaming, rendering, or running long high-load sessions. ASUS's official warning is that using a graphics card with a faulty fan may damage the card, and that a card with an abnormal fan should stop being used.
• Do not disassemble the graphics card yourself. ASUS states that improper disassembly may void the warranty. This guide does not teach removing the heatsink, removing the fan, or re-oiling bearings. The deepest steps it gives are: power off and disconnect, clean surface dust with a fine brush, and check whether any cable touches the blades.
• Do not flash the vBIOS yourself. See the settings section.
• Do not poke the blades with a metal tool, and do not block or flick the fan with a hand while it is powered. There is a risk of injuring a hand and of damaging the fan. Any check of blade smoothness must be done gently, after power-off and disconnection.
• Do not apply lubricant yourself. No manufacturer recommends user-applied fan lubrication. This guide does not teach adding oil of any kind. Self-applied lubrication may void the warranty; send the card for inspection instead.

Pricing and Inspection

If the checks above point to a fault rather than normal fan stop, the next step is an inspection and a written quote. The published total price range for the shop's three-tier repair service is HK$400–2,800, with the actual quote given after inspection. No specific repair fee is stated here, because the fee depends on what the inspection finds.

The shop is at Shek Mun, Dali Plaza, Room 1209. Phone: 2186 8010. WhatsApp: 6558 6806. There is no Mong Kok shop and no on-site service; inspection and repair are handled at the shop.

FAQ

Is it normal for a GPU fan not to spin at idle?

Often, yes. Many cards use a fan-stop feature — MSI Zero Frozr, ASUS 0dB, GIGABYTE Fan Stop, AMD Zero RPM — that stops the fans at low temperature and low load. The test is the load response: if the fans start as temperature and power rise, it is normal. If temperature keeps climbing and the fans stay at 0 RPM, that is a fault or settings problem.

Is a brief spin at boot, then a stop, normal?

Often, yes. MSI's official documentation says the fans should spin during power-on detection of the card, and then stop once the Zero Frozr conditions are met in the operating system. The observation worth noting is the opposite — no spin at all even at power-on — and even that must be combined with other observations, not used alone as a diagnosis.

Why does only one fan not spin?

Some multi-fan cards control each fan independently, so one fan may stay stopped below a certain temperature while another starts earlier — but not every card works this way. Judge by the load response: if the same fan never spins under load, stutters without starting, makes a grinding noise, or does not respond to manual control while the other fans behave normally, that points to a fault with that fan or its power branch. Check software settings first, then obstruction and that fan's power cable and plug.

Should I turn fan stop on or off?

There is no single right answer. Leaving fan stop on gives silence at idle, which is what the feature is for. Turning it off makes the fans run at a minimum speed continuously, which some users prefer if frequent start-stop cycling is annoying — one documented user report describes a fan cycling repeatedly between 0 RPM and about 1400 RPM at light load because temperature hovered near the threshold, and the suggested remedy in that report was to switch fan stop off or set a custom curve. Either choice is reasonable; the deciding test is whether the fan starts reliably under load.

My fan keeps cycling on and off. What should I do?

First, check whether temperature is hovering near the start and stop thresholds — that is the usual cause. Flatten the low-temperature part of the fan curve so there is no steep climb right at idle temperature, and use Temperature Hysteresis to widen the gap between start and stop. Also check case airflow, since poor airflow lets idle temperature drift up and down across the thresholds. Do not lock the fan at 100% as the fix.

The fan does not spin under load. What should I do?

Stop using the card under load. Work through the diagnosis in order: check fan-stop settings and restore the default curve with only one control tool active, update the driver, then power off and check for dust, obstruction, and the fan power cable and PCIe power plugs. If the fan still does not spin, send the card for inspection with the idle and load readings recorded. Do not disassemble the card, do not flash the vBIOS, and do not keep running games or renders on a card with a confirmed fan fault.

Can I just use Task Manager to check?

No — not for this. Task Manager shows only core temperature, with no Hot Spot, no VRAM temperature, and no fan speed. Use GPU-Z or HWiNFO for the four readings (temperature, load, power, RPM) plus a visual check of the fan.

Does the motherboard BIOS control graphics card fan stop?

No. Graphics card fan stop is controlled by the card itself. Motherboard BIOS fan settings control case and CPU fans. Do not look for a graphics card fan-stop option in the motherboard BIOS.