For an RTX 3090 running AI workloads, the reliable way to cut heat or power is to measure a representative task, change one setting at a time, and compare power, temperatures, throughput, and stability on your own card. A supported power cap is reversible, but it can lower clock speeds and extend task time; better case airflow may help when cooling is the problem.
Know which temperature and power figures apply to your card
NVIDIA lists 350 W graphics card power and a 93°C maximum GPU temperature for the RTX 3090 Founders Edition. These are manufacturer specifications for that design, not universal limits or targets for every add-in-board RTX 3090. Check the specifications and guidance for your exact card before changing settings. NVIDIA RTX 3090 specifications
As an Amazon Associate I earn from qualifying purchases.
The RTX 3090 has 24 GB of GDDR6X memory, but the GPU core temperature does not tell you the memory temperature. NVIDIA’s nvidia-smi documentation distinguishes the readings, and memory-temperature reporting is available only on supported products. If your system does not expose a memory reading, do not infer it from the core sensor. NVIDIA System Management Interface documentation
Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Scan for outdated or missing drivers - takes under a minute3Clear out junk files and repair common Windows errorsEstablish a repeatable baseline
Record the board model, case and airflow setup, ambient conditions, driver and software versions, and AI task settings. Use the same model, inputs, batch or inference settings, and run procedure for every comparison. Otherwise, a change in workload or environment can look like a tuning effect.
#1 Best Overall
- Model:PLD09220S12H
- Voltage: DC 12V
- Current: 0.55A
- Fan Dimension (L x W x H): approx. 87x87x22mm
- Power Connector: 4-Pin Header
- Before the task, run
nvidia-smi -qto inspect GPU details and the power and temperature fields exposed by your system. - During inference, run
nvidia-smi dmon -s pcuto monitor power, clocks, utilization, and temperature. Check the installed command’s help if an option or field is unavailable. - Record sustained power and temperatures, task duration or throughput, stability, and fan noise. Note whether the workload is actually power-limited or shows signs of thermal throttling.
NVIDIA documents these monitoring commands, but available sensors and fields vary by device and system. NVIDIA System Management Interface documentation
Try a power cap when reducing board draw is the priority
Where supported, NVIDIA documents nvidia-smi -pl <power_cap> for setting a GPU power limit. The accepted limit and ability to change it depend on the device, driver, system, and permissions; check your board documentation and the help for the installed command. NVIDIA’s TensorRT for RTX guidance notes that reaching a power cap can lower GPU clocks. A cap can reduce power use, but it is not a free performance gain. NVIDIA TensorRT for RTX performance guidance
Rank #2
- For ASUS ROG Strix GeForce RTX 3080 V2 OC Edition 10GB GDDR6X
- For ASUS ROG Strix GeForce RTX 3080 V2 10GB GDDR6X
- For ASUS ROG Strix GeForce RTX 3060 Ti V2 OC Edition 8GB GDDR6
- For ASUS ROG Strix GeForce RTX 3060 Ti V2 8GB GDDR6
- For ASUS ROG Strix GeForce RTX 3070 V2 OC Edition 8GB GDDR6
- Save the existing power-limit setting and baseline measurements.
- Lower the supported limit cautiously, then rerun the same AI task under comparable conditions.
- Compare average power, GPU temperature and memory temperature if reported, task time or throughput, stability, and noise.
- Keep the change only if the reduction in draw or heat is worth the measured performance cost. Restore the original setting if stability worsens or throughput falls too far.
Do not assume a particular cap is optimal for every RTX 3090. The right trade-off depends on the card and workload.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Address airflow if the card is thermally constrained
If temperatures remain high or the GPU is thermally throttling, first check that the card’s fans are operating and that case intake and exhaust paths are unobstructed. NVIDIA says a GPU that reaches its maximum specified temperature will have its performance throttled; if temperature continues rising, the system may shut down to protect the card. NVIDIA also identifies improved system cooling, including an added case fan, as a possible way to reduce temperatures. NVIDIA Support temperature guidance
Rank #3
- Model : PLD09220S12H, 12V, 0.55A .This is OEM fan, not the cheap fake fan , please do not buy the fake fan to damage your GPU
- Suitable for EVGA RTX 3070 FTW3 ULTRA GAMING Video Card
- Suitable for EVGA RTX 3080 Ti FTW3 ULTRA GAMING Graphics Card
- Suitable for EVGA RTX 3080 FTW3 ULTRA GAMING Graphics Card
- Suitable for EVGA RTX 3090 FTW3 ULTRA GAMING Graphics Card
Before adding a fan, check available case mounts and whether the airflow direction will help move air through the case. A cooling change depends on case compatibility; compare the result using the same workload and measurements as your baseline.
Compare tuning options by the outcome that matters
| Option | What to compare | Trade-off |
|---|---|---|
| Software power cap | Sustained board power, temperatures, task duration or throughput, stability, and noise | Reversible where supported; reaching the cap can lower clocks and performance. |
| Case airflow improvement | GPU and reported memory temperatures, throttling, stability, and noise | May help cooling, but depends on case mounts and airflow needs. |
A 2024 study, “Sustainable Supercomputing for AI: GPU Power Capping at HPC Scale,” estimated about 10.1 W lower GPU power draw and about 4.8°C lower GPU temperature on average for a treatment effect in GPU jobs at a research supercomputing center. Those figures describe that study’s environment, not a controlled RTX 3090 desktop benchmark or a prediction for an individual card. Sustainable Supercomputing for AI: GPU Power Capping at HPC Scale
Rank #4
- Model:CF9010U12D / Voltage: DC 12V / Current: 0.45AMP
- Fan Dimension (L x W x H): approx. 88mm * 88mm * 12mm
- For GeForce RTX 3060 Ti 8GB TUF OC
- For GeForce RTX 3070 Ti 8GB TUF OC
- For GeForce RTX 3080 Ti 12GB TUF
Use measurements, not universal undervolting or fan settings
No universal RTX 3090 power cap, voltage-frequency curve, or fan setting is established for AI workloads. Treat any such setting as card- and workload-specific: make one change, rerun the same task, and retain it only if measured results meet your goals without unacceptable performance loss or instability.
Quick Recap
Best Value
- Compatibility: Compatible with Dell and Compatible with Lenovo RTX 3060, 3060Ti, 3070, 3080, 3090 graphics cards.
- Optimal Cooling: Efficiently dissipates heat to maintain GPU performance during intensive use.
- Easy Replacement: Direct-fit design allows quick installation without modifications.
- Quiet Operation: Low-noise fan keeps your system cool without disturbing your workspace.
- Upgrade & Maintenance: Ideal for replacing worn-out or failing GPU fans to restore cooling efficiency.
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.
