Free tools Windows power users keep installed
One-click scans. No signup required.
iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
GPU throttling is a control response, not automatically a hardware failure. A graphics processor lowers its clock when it reaches a thermal threshold, an enforced power limit, or another protection condition. The fastest reliable diagnosis is to record temperature, board power, clock speed, utilization, fan speed when available, and the active throttle reason during the same sustained workload. Thermal evidence leads to cooling and airflow checks; power-limit evidence leads to limit, tuning-profile, workload, and power-delivery checks.
What GPU throttling means
Modern GPUs dynamically adjust clocks and voltage to stay within electrical, thermal, and reliability boundaries. Under a demanding game, renderer, compute job, or AI workload, a lower clock can reduce frame rate or throughput. That reduction is expected when a configured limit is reached; it is not proof that the card is defective.
NVIDIA distinguishes two common causes: power throttling occurs when average GPU power reaches the configured power limit, while thermal throttling occurs when the GPU reaches a predefined temperature threshold and reduces clocks. The thresholds and available sensors depend on the exact model and platform, so neither a single temperature reading nor a clock value can identify the cause by itself. See NVIDIA’s TensorRT performance guidance.
The Tool Desk
Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Other limiters can look similar. A workload may stop using the GPU fully, a configured clock setting may constrain frequency, or a hardware-slowdown condition may be asserted because of high temperature, high power draw, or an external power-brake signal. Treat the word “throttling” as a description of reduced performance until telemetry identifies the limiter.
#1 Best Overall
- High performance cooling fan, 120x120x25 mm, 12V, 4-pin PWM, max. 1700 RPM, max. 25.1 dB(A), >150,000 h MTTF
- Renowned NF-P12 high-end 120x25mm 12V fan, more than 100 awards and recommendations from international computer hardware websites and magazines, hundreds of thousands of satisfied users
- Pressure-optimised blade design with outstanding quietness of operation: high static pressure and strong CFM for air-based CPU coolers, water cooling radiators or low-noise chassis ventilation
- 1700rpm 4-pin PWM version with excellent balance of performance and quietness, supports automatic motherboard speed control (powerful airflow when required, virtually silent at idle)
- Streamlined redux edition: proven Noctua quality at an attractive price point, wide range of optional accessories (anti-vibration mounts, S-ATA adaptors, y-splitters, extension cables, etc.)
Log the slowdown before changing anything
Reproduce the performance drop with the workload that causes it. Capture a short idle baseline, then collect telemetry continuously while the workload is sustained. Every value must come from the same period; a peak temperature from one run and a later power reading from another cannot establish causation.
Signals to record
- GPU temperature, including any separate hotspot or memory temperature that the vendor exposes.
- Board or GPU power draw and the configured power limit.
- Graphics or compute clock speed.
- GPU utilization and, for games, a frame-rate or frametime trace.
- Fan speed and fan response under load, when reported.
- The active throttle, limiting, or violation reason.
Record driver and firmware versions, the exact card model, the application, and the test duration with the measurements. A brief burst can produce a different result from a long render or gaming session because temperature, fan speed, and average power change over time. Save the log rather than relying on a screenshot of one peak value.
NVIDIA’s nvidia-smi reference documents temperature, power and clock telemetry on supported devices: nvidia-smi documentation. A typical sampling example is:
Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Fix the driver behind crashes, sound loss and screen glitches3Clear out junk files and repair common Windows errorsnvidia-smi --query-gpu=temperature.gpu,power.draw,clocks.gr,utilization.gpu --format=csv -l 1This example is a starting point, not a universal command. A field can be unavailable on a consumer card, driver, operating system, or older GPU. Add a supported fan or throttle field only after checking the command reference for that device.
NVIDIA DCGM documents separate software power-cap and software thermal throttle indicators, as well as a hardware-slowdown indicator that can represent high temperature, an external power-brake assertion or high power draw: NVIDIA DCGM field constants. AMD’s SMI documentation provides power and thermal violation metrics for supported Radeon and Instinct generations: AMD SMI GPU violations.
How to read the signals together
| Observed pattern during the slowdown | Most useful next check |
|---|---|
| Temperature rises to the model’s documented threshold, clocks fall, and a thermal reason is active | Fan response, heatsink and vent obstruction, dust, ambient temperature, and chassis intake/exhaust |
| Board power approaches the configured cap, clocks settle lower, and a power-cap reason is active | Power-limit setting, tuning profile, workload, PSU adequacy, and board power delivery |
| Utilization falls without a thermal or power reason | Workload behavior, application settings, clock settings, driver state, and other hardware-slowdown indicators |
| A hardware-slowdown flag appears with both high temperature and high power | Use the temperature and power traces to determine which condition is dominant; do not label it thermal automatically |
The table is a diagnostic map, not a set of universal thresholds. A temperature that looks high on one card may be normal on another, and some products expose a hotspot or memory reading while others do not.
Use the evidence to choose the right branch
Thermal limit or thermal violation is active
When the logged reason is thermal, inspect the cooling path before replacing hardware. Observe whether the GPU fans accelerate as load rises. Check the card’s intake area, heatsink fins, filters and case vents for dust or obstruction. Confirm that the chassis has a clear path for cool intake air and warm exhaust air, and note unusually high room temperature or a side panel that is restricting airflow.
Quick wins for a faster PC:
Repair Windows errors before they cause bigger problemsFix Now →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Let the system reach the same sustained workload after each check. If opening the case changes the temperature, that is evidence of a chassis-airflow problem, but it does not identify which fan, filter, or obstruction is responsible. If the fan remains at an unexpectedly low speed while temperature rises, check the vendor’s support advisories for the exact card and firmware before assuming the cooler has failed.
Rank #2
- 【High Performance Cooling Fan】 Automatic speed control of the motherboard through the 4PIN PWM fan cable interface, which can determine the speed according to the temperature of the motherboard, with a maximum speed of 1550RPM. Configured with up to 55cm of cable for PWM series control of fans, ideal for cases and CPU coolers.
- 【Quality Bearings】The carefully developed quality S-FDB bearings solve the problem of pc cooling fan blade shaking in lifting mode, keeping fan noise to a minimum while providing maximum cooling performance when needed and extending the life of the fan.
- 【Vibration reduction and low noise】 The case fan is equipped with four soft material silicone corner pads on all four sides, which can reduce the vibration and friction caused by the rotation of the fan, perfectly reducing noise and allowing low noise operation, so that cooling can be carried out in low noise.
- 【Silent Fan Size】 Model: TL-C12C X3, Size: 120*120*25mm, Speed: 1550RPM, Noise 25.6dBA Connector: 4pin pwm, Current: 0.20A, Air Pressure: 1.53mm H2O, Air Flow: 66.17CFM, Higher air flow for improved cooling performance.
- 【Perfect Match】The PC fan can be used not only as a case fan, but is also suitable for use with a cpu cooler to create a cooling effect together, which can take away the dry heat from the case and the high temperature generated by the CPU in operation, allowing for maximum cooling; Ideal for cases, radiators and CPU coolers.
Intel’s Arc guidance specifically recommends air exchange with both intake and exhaust fans; that is an Arc-focused example, not a universal case layout: Intel Arc overheating guidance. A case fan is relevant only when the measurements point to inadequate chassis airflow. Verify the case’s supported fan size, mounting points and intended airflow direction. A fan will not remove an enforced power cap, repair a failed GPU cooler or correct model-specific firmware.
Also distinguish a blocked chassis path from a blocked card cooler. A clean case with a card pressed against another expansion board can still have restricted intake. Conversely, a well-ventilated case cannot compensate for a fan that does not respond or a heatsink whose fins are obstructed. Do not disassemble the card or replace thermal paste until the documented fan behavior, airflow, and telemetry point to a cooler-level problem and you understand the effect on service coverage.
Power cap or power violation is active
Compare live board power with the configured limit. A GPU that reaches its designed cap can be operating normally; NVIDIA describes this condition as power throttling by design. Check whether a vendor tuning profile, an application-specific setting or a workload that saturates the board explains the limit. On AMD hardware, the SMI guidance recommends checking power-cap thresholds, live power and PSU adequacy when power violations are high.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Check the configured limit before interpreting a percentage or wattage as a fault. A profile may intentionally reduce the cap for noise, temperature, or system-power reasons. Restore the vendor default or another known-good profile, repeat the same workload, and compare both the cap and the resulting clock. If the card reaches the same cap at the same point in every run, that is different evidence from a cap that changes unexpectedly or from a power reading that falls while a hardware power-brake indicator is active.
Do not raise the power limit as a diagnostic shortcut. Increasing it can add heat and may exceed the cooling or power-delivery capacity that the system was designed to handle. First restore the default vendor profile, verify all required PCIe power connectors are firmly seated, and confirm that the system’s power supply is appropriate for the exact card and platform. The available controls and readings differ among AMD products; AMD describes those differences in its WattMan support article.
Adequate PSU capacity is only one part of the check. Look for a configured software cap, a system tuning mode, and whether the workload is intentionally power-limited. Do not turn a normal cap into a hardware diagnosis simply because the card does not maintain its highest advertised boost clock; dynamic clocking is influenced by power, temperature, and utilization.
No matching reason, or signals conflict
Reset overclocks, undervolts, custom fan curves and third-party tuning profiles to a known-good vendor profile. Repeat the identical logged workload. If the reason remains unclear, check the exact card’s support page for a firmware or driver advisory, then update or reinstall the vendor-supported driver when appropriate. NVIDIA documents additional clock limiters, including clock settings and hardware-slowdown conditions, so a slowdown without a thermal or software power-cap flag needs broader investigation.
Recommended Free Tools
Conflicting readings can result from unsupported fields, different sampling intervals, or comparing two runs with different application scenes. Confirm that the tool is reading the GPU under test, that the timestamps overlap, and that the reported clock is the relevant graphics or compute clock. If a metric is unavailable, state that limitation instead of substituting an unrelated temperature or power value.
Rank #3
- Streamlined Fan Connections: Daisy-chain multiple fans together and control them all through just a single 4-pin PWM connector.
- Precise PWM Speeds: Set your fan speeds up to 2,100 RPM while providing up to 72.8 CFM airflow to your system.
- CORSAIR AirGuide Technology: Anti-vortex vanes direct airflow at your hottest components for concentrated cooling, pushing air in the direction you need when mounted to a radiator or heatsink.
- High Static Pressure: RS fans work well as radiator fans with a static pressure of 4.15mm-H2O to push through obstructions.
- Quiet Operation with Magnetic Dome Bearing: CORSAIR’s Magnetic Dome bearing minimises noise and reduces friction for greater longevity.
Temperature readings need model context
There is no cross-vendor temperature number that proves a GPU is throttling. NVIDIA’s TensorRT material uses “around 85 degrees Celsius for most GPUs” as an explanation, not as a universal consumer-card cutoff. NVIDIA’s command reference notes that target-temperature support is unavailable on some devices, and AMD’s controls and sensors vary by GPU. Check the exact board’s current specification and monitoring support rather than judging a card from a number seen in a forum or from another model.
A reported edge temperature, hotspot temperature, and memory temperature are not interchangeable. Use the sensor that the vendor associates with the relevant limit, and note which sensor your tool actually exposes. A single maximum value also does not show how long the GPU stayed there or whether clocks changed at that moment.
Intel’s Arc support article discusses gaming cases in which GPU and VRAM temperatures were observed above 90 °C. That describes an Arc troubleshooting scenario, not a safe or unsafe limit for every GPU: Intel’s Arc overheating article.
Platform-specific checks
NVIDIA GeForce and data-center GPUs
Use nvidia-smi or the supported management stack to correlate temperature, power, clocks and utilization. The tool reports target-temperature support only where the device exposes it and describes dynamic clocking based on power, thermal and utilization conditions: nvidia-smi reference. DCGM’s reason fields help separate software power-cap and software thermal limiting, but its detailed throttle documentation is aimed largely at NVIDIA-managed environments; consumer-card field availability can differ.
Treat a hardware-slowdown flag as a prompt to inspect all correlated signals. It can represent high temperature, an external power-brake assertion or high power draw, so labeling it “thermal” without the temperature and power traces is unsafe. For platform-specific power and thermal controls, consult NVIDIA’s TensorRT hardware and software environment guidance.
Do not assume that a missing target-temperature field means the GPU has no thermal protection. It may mean that the device does not expose that control or sensor through the selected interface. Use the exact model’s documentation and the available throttle-reason fields instead.
AMD Radeon and Instinct GPUs
Radeon Software’s performance-tuning view can expose a histogram for GPU activity, temperature, power, fan speed and clocks, with fan and power-limit controls on products that support them. AMD documents that feature availability depends on the GPU in its WattMan guide. AMD SMI violation metrics can show the percentage of time a supported GPU was throttled for power or thermal reasons; use those percentages together with live readings and configured limits rather than as standalone fault codes: AMD SMI GPU-violation documentation.
Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchWindows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallA violation percentage is useful for sustained jobs because it describes how much of the monitored interval was affected, but it does not by itself explain whether the limit was expected. Pair it with the configured cap, live power, temperatures, clocks, and workload timing. Command names and metric support differ among Radeon and Instinct generations.
Rank #4
- High Performance Cooling Fan: The design of nine fan blades, the maximum speed reaches 1200 RPM, and it is connected to the motherboard through the 3 PIN interface, providing a good cooling effect for the case
- Low Noise: Every fans is equipped with four soft silicone cushions that can absorb vibration at high speeds. The maximum noise is only 32.1 dBA. Keep the case in a relatively quiet environment when working
- Hydraulic Bearing Design: High-quality bearings can make the fan rotate more stably, reduce noise, and prolong its service life. Each fan can work an average of 30,000 hours
- Simple Installation: This computer fan's size is 120 mm and is compatible with all types of cases, making it easy to install. You can do it even if you have no installation experience
- Good Insulation and Heat Resistance: Case fan uses PBT environmental protection material, with good insulation and heat resistance, tough and durable quality
If thermal violations dominate, inspect fan operation, cooler and heatsink obstruction, ambient temperature and chassis dust. If power violations dominate, compare the cap with live power and verify PSU capacity before changing tuning.
Intel Arc
For Arc cards, check case intake and exhaust first when temperatures rise under sustained gaming or compute load. Intel’s overheating article was last reviewed on December 3, 2024, and describes Arc gaming cases with GPU and VRAM temperatures over 90 °C; use its airflow recommendation as Arc-specific guidance rather than a universal limit.
Intel also documents a specific fan-control firmware issue. Cards with OPROM 20.1053.1 could keep the fan near 1,200 RPM while temperature rose rapidly. The advisory, last reviewed December 19, 2024, says driver 31.0.101.4644 included a firmware update for that issue. Verify the current support article, exact card and firmware before applying version-specific advice: Intel Arc fan advisory.
Step-by-step troubleshooting procedure
- Identify the exact platform. Record the GPU model, board vendor, driver, firmware or OPROM version, operating system, and the monitoring tool. Sensor and throttle-field support varies by model and software.
- Return to a known baseline. Disable overclocks, undervolts and custom tuning, or select the vendor’s default profile. Do not change BIOS thermal-trip settings.
- Capture idle telemetry. Note temperature, power, clocks, utilization and fan speed with no demanding application running.
- Reproduce the slowdown. Run the same game scene, benchmark, render or compute job long enough for the behavior to appear. Save telemetry from that interval, including the limiter reason.
- Check correlation. Align temperature, power, clock, utilization and fan samples by time. A later reading cannot explain an earlier clock drop.
- Identify the active limiter. Use the vendor’s documented throttle or violation field where available. Do not infer the reason from temperature alone.
- Follow one branch. For thermal evidence, inspect fans, dust, obstruction, ambient conditions and intake/exhaust. For power evidence, inspect the configured cap, tuning profile, workload and PSU or board power delivery.
- Apply one conservative correction. Clear an obstruction, restore a default profile, correct a case-airflow issue, or apply a verified model-specific driver or firmware fix. Avoid changing several variables at once.
- Retest under identical conditions. Compare clock, throughput, temperature, power, fan behavior and limiter reason with the baseline. A successful fix should change the relevant signal, not merely produce a different one-off result.
- Escalate when evidence remains abnormal. Check the exact model’s support advisories, then pursue a driver reinstall, firmware-specific fix or hardware support if the cooler, fan or power delivery does not behave as documented.
Common mistakes and unsafe “fixes”
- Calling every high temperature a fault: limits are model-specific, and a temperature reading without a throttle reason does not establish throttling.
- Assuming every clock drop is thermal: power caps, utilization changes, clock settings and hardware-slowdown conditions can all reduce clocks.
- Comparing unrelated runs: different scenes, sampling intervals, drivers, ambient conditions, or power profiles can invalidate the comparison.
- Raising the power limit first: this can increase heat and hide the original constraint instead of diagnosing it.
- Adding a case fan without airflow evidence: a fan helps only when chassis intake or exhaust is inadequate, and its size and mounting must fit the case.
- Changing BIOS thermal-trip points: NVIDIA warns against changing the documented Grace ACPI trip points in its platform-specific guidance: Grace power and thermals.
- Replacing thermal paste or disassembling the card immediately: confirm fan, airflow and telemetry evidence first; disassembly can damage the card and affect service coverage.
- Overclocking while troubleshooting: AMD notes that its warranty does not cover damage caused by overclocking in the WattMan guidance: WattMan support.
- Treating a vendor-specific advisory as universal: Intel’s Arc fan issue and its firmware remedy apply to the documented card and version, not automatically to other GPUs.
When the result is normal operation
A sustained workload that reaches the configured power cap, followed by a stable clock and no abnormal temperatures, may simply be operating within the GPU’s design. Likewise, a thermal limiter that activates at the model’s documented threshold is performing its protective function. The actionable question is whether the limiter appears earlier than expected, whether cooling or power delivery fails to respond as designed, and whether performance changed from a previously comparable baseline.
Maximum advertised boost clocks are not guaranteed to remain constant. Dynamic clocking responds to power, temperature, and utilization, so a lower but stable clock under a heavy workload can be normal. Investigate a change from a known-good baseline, an unexpectedly low cap, a fan that fails to respond, or a limiter that appears without the corresponding environmental or power condition.
Keep the complete before-and-after log: workload, driver and firmware versions, temperature, power, clocks, utilization, fan behavior and limiter reason. That record gives vendor support enough context to distinguish a normal cap from a fan, airflow, firmware, driver or board-level problem.
Frequently Asked Questions
Is 85 °C the temperature at which every GPU throttles?
No. NVIDIA uses “around 85 degrees Celsius for most GPUs” as an explanatory example in its TensorRT guidance, not as a universal cutoff. Temperature thresholds, target-temperature controls, and exposed sensors vary by model. Check the exact GPU’s documentation and correlate the temperature with a thermal throttle or violation reason.
If my GPU reaches its power limit, is it defective?
Not necessarily. NVIDIA describes power throttling as occurring when average GPU power reaches the configured limit, which can be normal by design. Check the configured cap, workload, tuning profile, clock behavior, and temperatures. A cap that is unexpectedly low or changes without an intentional profile deserves further investigation.
Best Value
- HIGH STATIC PRESSURE: Efficient even with resistance – the generated airflow easily penetrates dense radiators, narrow perforated panels and mesh structures and ensures reliable cooling
- PWM CONTROL WITH WIDE SPEED RANGE: The speed can be progressively adjusted up to 3000 rpm via the 4-pin PWM connection – the fan stops completely at less than 5% PWM
- PRECISE MANUFACTURING FOR MAXIMUM SMOOTH RUNNING: Minimal gaps, automatic balancing and high-precision measurement noticeably reduce vibrations – for quiet, efficient and long-lasting performance
- SMOOTH-RUNNING FLUID DYNAMIC BEARING (FDB): The self-lubricating bearing minimizes noise during operation – ideal for quiet, efficient cooling and a long, reliable service life
- NEW FAN BLADE DESIGN FOR MORE PERFORMANCE: The redesigned rotor blades offer an optimal balance of performance and low noise – especially efficient at low speeds
What should I log while the slowdown is happening?
Record temperature, board or GPU power, configured power limit, clock speed, utilization, fan speed when available, and the active throttle or violation reason during the same sustained workload. Also record the card model, driver, firmware, application, and test duration.
Why can a hardware-slowdown flag not be labeled thermal immediately?
NVIDIA’s DCGM documentation says the hardware-slowdown indicator can reflect high temperature, an external power-brake assertion, or high power draw. Use the temperature and power traces and the other reason fields to determine which condition is present.
Will adding a case fan fix GPU power throttling?
Only if the evidence shows inadequate chassis airflow contributing to a thermal limit. Intel’s Arc guidance recommends both intake and exhaust air exchange, but a fan does not remove an enforced GPU power cap, repair a failed card cooler, or correct a model-specific firmware issue. Verify the case’s fan size, mounting, and airflow direction.
Should I raise the power limit to stop throttling?
No, not as a first diagnostic step. Raising it can add heat and exceed the system’s cooling or power-delivery capacity. First restore a known-good profile, compare live power with the configured cap, check the workload and PSU adequacy, and retest.
What if my monitoring tool does not show a throttle reason?
Tool and field availability depend on the GPU, driver, operating system, and platform. Confirm that the command supports the exact card, align the available temperature, power, clock, and utilization samples, reset custom tuning, and check the model’s current driver or firmware advisories. Do not substitute a temperature number for a missing reason field.
When should I contact the GPU or system vendor?
Escalate when a fan fails to respond as documented, a cooler or power-delivery condition remains abnormal after conservative checks, a verified model-specific advisory applies, or performance remains below a comparable baseline with no explainable limiter. Provide the complete workload and before-and-after telemetry log.
The Bottom Line
Diagnose the limiter, not just the symptom. Log temperature, power, clocks, utilization, fan behavior, and the active reason during one repeatable workload. Thermal evidence points to fans, dust, ambient conditions, and chassis airflow; power evidence points to the configured cap, tuning profile, workload, PSU, and board power delivery. Restore known-good settings, change one variable at a time, and use the exact GPU’s documentation before applying model-specific fixes.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
