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
A 70% GPU power limit can cut power use by more than it cuts frame rate—but it does not guarantee a 30% power saving or an imperceptible performance change. In one RTX 4090 test, a 70% limit reduced performance by 4% and power by 19% in a specific game scenario. The result depends on your graphics card, game, settings, and whether the GPU is actually reaching its power limit.
What a 70% GPU power limit changes
A GPU power limit sets a ceiling on the power available to the graphics processor. It does not directly set a frame-rate limit, and 70% does not mean 70% of the usual FPS or 30% less measured power. NVIDIA describes TGP as “the power cap limit for GPU Boost,” which adjusts GPU performance based on available power, temperature, and other factors. NVIDIA’s explanation of TGP and GPU Boost also notes that lighter workloads and CPU bottlenecks can leave a GPU drawing below its power target.
That is why reducing the limit may have little effect in a game where the GPU is underused, but can reduce performance in demanding, GPU-bound scenes. A frame-rate cap is a separate control: it limits output FPS, rather than setting the GPU’s power ceiling. NVIDIA documents Max Frame Rate and Optimal Power settings in its control panel guidance.
The Tool Desk
Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →What benchmark results show—and what they do not
Published results are specific to the card, game, settings, and test method. They illustrate possible trade-offs, not what every player will see.
#1 Best Overall
- Axial-tech fans now feature a smaller fan hub that facilitates longer blades and a barrier ring that increases downward air pressure
- 2.5-slot design allows for greater build compatibility while maintaining cooling performance
- 0dB technology lets you enjoy light gaming in relative silence
- Dual BIOS switch lets you toggle between Quiet and Performance BIOS profiles
- Dual ball fan bearings last up to twice as long as sleeve bearing designs
| Test | Reduced power setting | Reported performance change | Power result |
|---|---|---|---|
| RTX 4090 Founders Edition, A Plague Tale: Requiem with DLSS 3, Tom’s Hardware | 70% of the card’s power limit | 4% lower performance than the test’s base result | 19% less power in that scenario |
| Radeon RX 9070 XT, Cyberpunk 2077, ComputerBase | 20% lower power target | About 6% lower FPS; a 30% lower target cost about 11% FPS versus stock | Power target reductions, not a claim of equal reductions in measured system draw |
| Radeon RX 9070 XT, Doom Eternal, ComputerBase | 10%, 20%, and 30% lower power target | About 3%, 7%, and 10% lower FPS, respectively | Power target reductions, not a claim of equal reductions in measured system draw |
| Radeon RX 5700 XT, Dying Light 2 at 1080p, PCWorld | 100%, 80%, and 60% | 77, 72, and 66 FPS, respectively | Whole-PC consumption: 305 W, 260 W, and 210 W, respectively |
| RTX 3080, Cyberpunk 2077 at 4K, PCWorld | 100%, 80%, and 60% | 42, 40, and 37 FPS, respectively | Whole-PC consumption: 570 W, 480 W, and 410 W, respectively |
The RTX 4090 result is from one game in a multi-game test; it is not a finding that every game loses only 4% at 70%. The RX 9070 XT results likewise vary by game, and the older PCWorld examples use different cards and test conditions. The PCWorld watt figures are total system power, not GPU-only measurements. Tom’s Hardware’s RTX 4090 testing, ComputerBase’s RX 9070 XT results, and PCWorld’s power-limit examples provide the specific test contexts.
How to tell whether the difference matters on your PC
“Almost nothing I would notice” is a personal judgment, not something benchmark averages can establish for every player. Compare your own card at stock and reduced limits using repeatable conditions. Look at frame consistency as well as average FPS: a similar average can conceal worse dips or uneven frame delivery.
Rank #2
- Powered by the NVIDIA Blackwell architecture and DLSS 4
- Powered by GeForce RTX 5070 Ti
- Integrated with 16GB GDDR7 256bit memory interface
- PCIe 5.0
- WINDFORCE cooling system
- Choose a repeatable workload. Use the same game scene, resolution, graphics settings, and upscaling or frame-generation options for each run. A built-in benchmark or a repeatable save point is more useful than comparing unrelated gameplay.
- Measure stock performance first. Record average FPS and, if your monitoring tool provides them, 1% lows, 99th-percentile FPS, or frame-time data. Also record GPU power, temperature, and fan speed or noise.
- Change only the power limit. Set the limit to 70% in the GPU tuning utility you already use, then repeat the same scene. Avoid changing clocks, voltage, fan curves, or game settings at the same time; otherwise, you cannot isolate the power-limit effect.
- Repeat across the games you care about. Include a demanding GPU-bound game and, if relevant, a game that is often CPU-limited or capped to a target frame rate. A result from one title cannot establish how the rest of your library will behave.
- Compare the trade-offs. Check performance retention alongside measured GPU power, temperature, and fan behavior. If you want whole-system power, label it as such: a system reading includes components beyond the graphics card.
If FPS barely changes, check whether the game was already CPU-limited, frame-capped, or otherwise keeping GPU use below its power target. If performance drops more than you find acceptable, raise the limit until the balance suits your games. Cooling design, fan curve, ambient temperature, and case airflow affect temperature and noise, so neither is guaranteed to improve by a particular amount.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
When a frame-rate cap may be a better fit
If your goal is to reduce GPU activity while playing at a frame rate below the card’s uncapped output, a frame-rate cap may be a more direct option. NVIDIA’s documented Max Frame Rate control limits FPS, while its Optimal Power setting is a separate power-management option. These controls are not interchangeable with a GPU power limit: a cap sets an output target, while a power limit constrains the GPU’s power budget. NVIDIA’s instructions for these settings apply to its control panel; available controls can vary by GPU and software version.
Rank #3
- Powered by the NVIDIA Blackwell architecture and DLSS 4
- Powered by GeForce RTX 5060
- Integrated with 8GB GDDR7 128bit memory interface
- PCIe 5.0
- WINDFORCE cooling system
Why 70% is not a universal sweet spot
The cited results show different performance-versus-power curves across cards and games. A lower target can cost a modest share of FPS in one workload and more in another. Undervolting can also change that curve, so figures from a stock-voltage test should not be treated as a prediction for a tuned card. Choose a limit based on your own measurements and the performance you are willing to trade for lower power—not on the percentage alone.
Quick Recap
Best Value
- Axial-tech fans now feature a smaller fan hub that facilitates longer blades and a barrier ring that increases downward air pressure
- Phase-change GPU thermal pad helps ensure optimal heat transfer, lowering GPU temperatures for enhanced performance and reliability
- 2.5-slot design allows for greater build compatibility while maintaining cooling performance
- Dual-ball fan bearings last up to twice as long as standard conventional sleeve bearings designs
- 0dB technology lets you enjoy light gaming in relative silence
Rank #4
- Powered by Radeon RX 9070 XT
- WINDFORCE Cooling System
- Hawk Fan
- Server-grade Thermal Conductive Gel
- RGB Lighting
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.
