Why Your RTX Laptop GPU Is Capped at 175W — and How to Raise It Safely
NVIDIA caps RTX laptop GPUs at 175W in firmware — even chips that run at 320W or more on desktop. Why the cap exists, what it costs you, and how to raise it safely.
By NvpwrControl Team · TGP unlock · power limit · RTX 4090 laptop · RTX 5090 laptop · 3DMark

If you own a gaming laptop with an RTX 4090, 4080, 5090 or 5080, your GPU is almost certainly capped at 175 W. Not because the chip can't use more — the exact same silicon runs at 320 W or 360 W in a desktop card — but because NVIDIA's firmware says so, and your laptop maker is only allowed to configure it lower. This article explains why the cap exists, what it costs you in measurable performance, and how to raise it safely without flashing your VBIOS.
The cap is a setting, not a hardware limit
Flagship laptop GPUs are built on the same dies as desktop cards. The RTX 4090 Laptop uses the AD103 chip from the desktop RTX 4080, a card rated for 320 W. In a laptop, firmware caps it at 175 W. The current generation repeats the pattern: the RTX 5090 Laptop is the GB203 die — the same one in the desktop RTX 5080, rated at 360 W — capped at 175 W.
| Laptop GPU | Die | Desktop card on the same die | Desktop rating | Laptop cap |
|---|---|---|---|---|
| RTX 5090 Laptop | GB203 | RTX 5080 | 360 W | 175 W |
| RTX 4090 Laptop | AD103 | RTX 4080 | 320 W | 175 W |
| RTX 5080 Laptop | GB203 | RTX 5070 Ti | 300 W | 175 W |
| RTX 5070 Ti Laptop | GB205 | RTX 5070 | 250 W | 140 W |
The cap is also the same in a thin 14-inch laptop and an 18-inch desktop replacement with three fans and a vapour chamber. NVIDIA sets one ceiling for the product line; cooling quality doesn't enter into it. If your laptop's cooling has headroom — and the big machines usually do — the chip could use more power than the cap allows. The power is already in your adapter and your GPU. It's just fenced off.
What the cap costs you, measured
Power is not performance in a straight line, but a GPU held at 175 W when it could sustain 250 W leaves real performance on the table. These are valid 3DMark runs from unlocked laptops, compared with Notebookcheck's database averages for stock laptops with the same GPU:
| GPU | Benchmark | Stock average | Unlocked | Gain |
|---|---|---|---|---|
| RTX 4090 Laptop | Time Spy Graphics | 21,007 | 25,564 | +22% |
| RTX 5070 Ti Laptop | Time Spy Graphics | 17,225 | 20,858 | +21% |
| RTX 5090 Laptop | Steel Nomad | 5,899 | 9,072 | +54% |
| RTX 5080 Laptop | Steel Nomad | 4,854 | 7,338 | +51% |
Across the board, that's around 30% more GPU performance in 3DMark from a limit change alone. The unlocked RTX 5090 and 5080 results also sit above the best stock results on record for those GPUs — headroom no stock configuration can reach, because every stock laptop runs into the same wall. Your own gains will depend on your cooling, your adapter and your silicon, and games vary more than benchmarks do. But the direction is consistent: the cap, not the chip, is the bottleneck.
Why MSI Afterburner can't fix it
The usual first stop is MSI Afterburner, and it genuinely can't help here. Afterburner adjusts clock and memory offsets — it tells the GPU to run faster within the power budget the driver enforces. On a desktop card you can also drag a power-limit slider, because desktop limits are exposed. On a laptop, the ceiling is locked in the driver's power policy: the slider has nowhere to go.
So you can overclock the clocks all you like; the moment the GPU hits 175 W, it throttles back to fit the budget. Raising performance means raising the budget itself — the limit, not the offsets. That's the one thing the mainstream tools don't touch.
How raising the limit works, without flashing VBIOS
The traditional route to a higher laptop power limit was flashing a VBIOS from a higher-wattage model — risky, model-specific, and easy to brick. NvpwrControl takes a different route: it works with the NVIDIA driver directly to change the limit in the driver's power policy. Nothing is flashed, and the app backs up the driver settings before changing anything.
The flow is deliberately cautious:
- Identify. The app checks your GPU, PCI IDs, NVIDIA driver and VBIOS against the live power policy before anything is written. If something doesn't match, power changes stay locked.
- Set MAX. Choose a new maximum limit in 5 W steps, save it and reboot. The driver settings backup happens first.
- Apply CURRENT. After the reboot, set the runtime limit while the GPU is idle. An optional auto-reapply re-sets the limit after MSI Center mode switches, on a kernel event with a 1–10 second delay and no CPU work in between.
Validated ranges on supported GPUs:
| GPU | Stock max | Unlockable to |
|---|---|---|
| RTX 5090 / 5080 Laptop | 175 W | 250 W |
| RTX 5070 Ti Laptop | 140 W | 180 W |
| RTX 5070 / 5060 / 5050 Laptop | 115 W | 140 W |
| RTX 4090 Laptop | 150 W | 250 W |
| RTX 4080 Laptop | 150 W | 225 W |
| RTX 4070 / 4060 Laptop | 120 W | 150 W |
| RTX 4050 Laptop | 115 W | 140 W |
RTX 50 and RTX 40 series are fully supported; RTX 30 (Ampere) and RTX 20 (Turing) support is experimental, meaning ranges haven't been validated across every model yet. The app is portable for Windows 10 and 11 — extract the zip, run NvpwrControl.exe as administrator, no install. (Windows SmartScreen may warn on first run because the app isn't code-signed; choose More info, then Run anyway.) Alongside the limit, it includes clock offsets, experimental V/F curve voltage control, and full telemetry: clocks, voltage, power, temperature, hotspot, P-states and rails.
How to raise it safely
The limit itself is not dangerous; heat is. Everything below is about respecting that.
- Go up in steps. The limit moves in 5 W increments for a reason. Raise it a step or two, benchmark, check temperatures, repeat. Don't jump straight to the maximum.
- Watch the right temperatures. GPU core temperature is the headline number, but hotspot and VRAM matter too. Monitor with HWiNFO or GPU-Z under a sustained load, not a 30-second burst.
- Check your adapter maths. Your power adapter has to supply the GPU and the CPU. A 250 W GPU limit plus a hard-working CPU can exceed a smaller adapter, and the symptoms are throttling or the battery draining while plugged in.
- Know your cooling ceiling. A higher limit only converts to performance if the cooling can hold the clocks. If temperatures pin at the limit and clocks fall back, you've found your real ceiling — back off to where it's stable.
- Keep the exit open. Everything is reversible: Reset all returns the runtime limit to stock and removes the MAX override and autostart. Combined with the driver-settings backup, you can always go home.
Should you do it?
If you have a large laptop with serious cooling, a high-wattage adapter, and a flagship GPU idling under a 175 W cap, you're the person the cap hurts most — the hardware to use more power is already on your desk. If your laptop is thin, runs hot at stock, or its adapter barely covers the stock configuration, the honest answer is that a higher limit buys you little. And whatever your machine, warranty treatment depends on your laptop maker, so read their terms first.
The 175 W limit is a setting. Settings can be changed — carefully, measurably, and reversibly.
Raising power limits increases heat and electrical load; do it at your own risk. NvpwrControl is independent software and is not affiliated with, sponsored by or endorsed by NVIDIA Corporation.
Notes — Questions this entry raises
Is raising my laptop GPU's power limit safe?
It is safe in the narrow sense that the change is software, checked before it is written, and fully reversible — but it is not free of consequences. More power means more heat in the GPU, VRAM, voltage regulators and cooling system. Raise the limit in 5 W steps, watch temperatures in HWiNFO or GPU-Z, and back off if they climb too high. A higher limit only helps if your cooling can keep up.
Will raising the power limit void my warranty?
That depends on your laptop maker. NvpwrControl is software and can be reverted completely — Reset all returns everything to stock and removes the override — but running hardware beyond its stock configuration may affect warranty claims. Check your manufacturer's terms before you start.
Do I need a bigger power adapter?
Possibly. At a higher limit the GPU draws more, so your adapter needs enough wattage for the GPU and CPU together. If it can't keep up, the laptop will throttle or drain the battery while plugged in. Add up your unlocked GPU limit and your CPU's sustained draw, and compare with the wattage printed on your adapter.
Why can't I just use MSI Afterburner?
Afterburner and similar tools adjust clock and memory offsets within the limits the driver allows. On laptops the power ceiling itself is locked, so they can't raise it. NvpwrControl changes the limit itself, then Afterburner-style clock offsets can be applied on top — NvpwrControl also includes core offsets up to ±1000 MHz and memory offsets up to ±3000 MHz through NVAPI.
Does NvpwrControl flash my VBIOS?
No. NvpwrControl never flashes the GPU's VBIOS. It works with the NVIDIA driver directly, checks your GPU, PCI IDs, driver and VBIOS against the live power policy before any write, and backs up the driver settings before changing the MAX limit. If something doesn't match, power changes stay locked.
How much performance will I actually gain?
It depends on your laptop's cooling, power adapter and silicon. In 3DMark, unlocked laptops have scored around 30% higher than stock averages for the same GPU: an RTX 4090 Laptop hit 25,564 in Time Spy Graphics versus a 21,007 stock average, and an RTX 5090 Laptop scored 9,072 in Steel Nomad versus 5,899 stock. Games vary more than benchmarks, and a laptop with weak cooling will gain less.