If your laptop or desktop PC sounds like a jet engine taking off every time you play games or render, you're probably experiencing the effects of excessively high CPU or GPU voltage. Undervolting is precisely the technique that allows you to lower that voltage to reduce temperature, power consumption, and noise , while maintaining almost no performance if done correctly.
In modern gaming laptops like the Lenovo Legion, desktops with powerful processors such as the Core i9 or Ryzen 9, and power-hungry graphics cards, manufacturers tend to err on the side of caution by slightly increasing the voltage. The good news is that this extra margin can usually be reduced quite safely by following a step-by-step method , either through the BIOS/UEFI or using specialized Windows software when the hardware allows it.
What is undervolting and how does it differ from underclocking?
When we talk about undervolting, we're referring exclusively to reducing the operating voltage of a processor (CPU or GPU) while trying to maintain its stock frequencies. You're not "capping" the chip by design, but rather fine-tuning how much power it needs to perform the same as before while generating less heat.
In any modern processor, when the maximum frequency it can reach is increased, the power management system itself raises the voltage to ensure that electrical signals travel fast enough and without errors. The problem is that power consumption and heat increase much faster than the FPS or the seconds you shave off in a render , because power is highly dependent on voltage.
That's why it makes sense to "tighten" the voltage in reverse: if the manufacturer has provided a safety margin, you can lower the voltage by millivolts and still maintain complete stability . This way, the CPU or GPU runs cooler, the system is quieter, and in some cases, it even maintains boost speeds better because it doesn't reach its thermal limit as quickly.
It's important not to mix terms, because many people confuse undervolting with other tuning techniques:
- UnderclockYou reduce the operating frequency (MHz or GHz). This lowers power consumption and temperature, but You lose performance directly and visibly.
- Undervolt: You cut the voltage trying to maintain the same frequency or a very similar one. If you do it right, you maintain almost the same performance, gain efficiency, and avoid throttling..
Unlike overclocking, undervolting doesn't force the chip beyond its specifications , so the risk of physically damaging the hardware is very low. The real danger lies in instability: random crashes, game closures, restarts, or blue screens if you undervolt too much.
What do you get by undervolting your CPU?
When you undervolt your CPU, you're configuring it to run at a lower voltage than its default setting for the same frequency . The most immediate effect is that the processor becomes more energy efficient, which is vital for Legion laptops, ultrabooks, and powerful desktops with limited cooling.
Reducing the voltage lowers the CPU's power consumption, and this lower consumption also reduces the amount of heat the chip generates. With proper tuning, the temperature can drop several degrees without any noticeable performance loss , especially during extended workloads like Cinebench, rendering, video compression, or demanding games.
Another very interesting side effect is the noise. If the CPU runs cooler, the fans don't need to increase their RPMs to keep the temperature under control. This is especially noticeable in gaming laptops, where every couple of degrees less can make the difference between a tolerable machine and an unbearable one.
Note that this isn't an automatic power reduction. If the processor can maintain its turbo frequencies with lower voltage and without errors, performance remains the same or even improves by avoiding thermal throttling. A Ryzen 7 5800X, for example, can better handle its high clock speeds in games if it's not constantly hitting 90°C.
The flip side is that if you cut too much, the chip won't have enough electrical headroom to operate at a certain frequency, and symptoms of instability will start to appear, ranging from blue screens to subtle errors in heavy applications.
Risks, limits, and why undervolting requires patience
The main real risk of undervolting isn't "burning out" the hardware, but rather living with an unstable system without realizing it . When you cross the line, the following issues typically arise:
- Random crashes of demanding games or applications.
- System freezes that force you to restart.
- Blue Screenshots (BSOD) or spontaneous restarts under load.
The most treacherous aspect is that many of these errors don't appear immediately after applying the new voltage . You can run a short test, think everything is fine, and then the problem will surface after two hours of gameplay or in the middle of a nighttime render. That's why it's essential to perform long and varied stability tests.
Furthermore, to strictly speak of undervolting , the other key CPU parameters must remain constant: multiplier, BCLK, and power limits . If you start adjusting the base frequency, multiplier, or perform aggressive underclocking, you enter a different category of adjustments that complicate troubleshooting.
There is software that allows you to undervolt from within Windows (Intel XTU, Throttlestop, and some manufacturers' own tools), which is very tempting for convenience, especially on laptops. However, for fine-tuning and permanent adjustments, it's always recommended, whenever possible, to do it from the BIOS/UEFI , where control over the Vcore and voltage curves is more direct and predictable.
For all these reasons, undervolting is a technique that works best in the hands of someone with some basic hardware knowledge and, above all, a lot of patience . It's not about pressing a magic button, but rather about testing step by step until you find the sweet spot between temperature, noise, and stability.
What is Vdroop, how does it relate to Load Line Calibration, and why should you care?
On any modern motherboard, when the CPU goes from idle to full load, the voltage it actually receives doesn't remain constant. This controlled voltage drop under load is called Vdroop.
At idle, you might see a relatively high voltage in monitoring tools, but when you start a stress test like Prime95 or Cinebench, that value drops slightly . The key to successful undervolting is understanding how your processor and motherboard react to this phenomenon, because cutting voltage without considering it can leave you with no headroom when you need it most.
Accurately measuring Vdroop would involve using a multimeter connected to the board's physical reading points, which is only recommended for advanced users. The theoretical procedure would be:
- Identify the nominal CPU voltage in the specifications or in the BIOS itself.
- Connect the multimeter to the CPU power line (if the board offers test points, even better).
- Measure the voltage at idle and then under a heavy load at 100% CPU usage.
- Calculate the difference between both readings to find the actual drop..
In practice, most users simply use software readings ( HWInfo and CPU-Z ) and play around with the Load Line Calibration (LLC) option in BIOS , which defines how the CPU voltage responds to Vdroop.
Increasing the LLC setting significantly reduces voltage drops under CPU load , but can lead to dangerous spikes when the load is released. Setting the LLC setting too low has the opposite effect: a pronounced Vdroop which, combined with aggressive undervolting, can severely compromise stability as soon as you launch a demanding game.
How to undervolt Intel CPUs step by step from BIOS
On current Intel processors (modern Core series), the most reliable way to undervolt is to enter the BIOS/UEFI and directly adjust the core voltage (Vcore) and, if applicable, the cache voltage . On ASUS ROG motherboards for LGA 1151 socket or later, for example, you'll see parameters similar to:
- CPU Core/Cache Voltage (Sometimes the kernel and cache go together, other times they go separately).
- Modes: Auto, Manual, Offset, Adaptive.
- Power options: PL1, PL2, Tau, etc.
Auto mode delegates everything to the motherboard and tends to be quite conservative with the voltage. Manual mode sets a constant voltage at idle and under load , which is not recommended for a computer used daily because it causes high idle temperatures and doesn't take advantage of power-saving capabilities.
Offset and Adaptive Modes
For undervolting, the most interesting modes are Offset and Adaptive. On many modern Intel platforms, you can use Adaptive for finer adjustments, but since not all models behave the same, negative Offset mode is the most universal and easiest option to understand.
- Voltage type: Offset.
- Offset direction: negative (subtract voltage).
- Initial value: something prudent, for example -0,020 V (i.e., -20 mV).
In addition to voltage, you need to pay attention to the power limits (PL1 and PL2). PL1 is usually the sustained power limit, while PL2 marks the maximum consumption during short peaks . On many motherboards for recent K-series processors, these values ​​are either "unlocked" or very high from the factory to maximize performance.
If your cooling isn't particularly powerful or your case has poor airflow, you might want to lower PL1 and/or PL2 to something more reasonable (for example, 125-200W depending on your CPU model and cooler). This will prevent your CPU from spiking in power consumption and temperature, although you'll see "Power Limit Throttling" messages in tools like Intel XTU when the processor demands more power than you're allowing.
For a PC you use daily, perhaps for work, it's recommended that each "candidate" configuration undergo 8 to 24 hours of stability testing . It might sound excessive, but it's the only way to minimize unpleasant surprises when the system has been running for many hours.
How to undervolt AMD (Ryzen) CPUs from BIOS
In modern AMD CPUs like the Zen 2-based Ryzen 3000 series or the Ryzen 5000 series, the approach is similar, but the menu names are different. AMD tends to use deliberately high voltages in certain scenarios , especially to trigger very aggressive boosts, which generated considerable discussion with the arrival of the Ryzen 3000 series because we saw high voltages even at idle.
In the BIOS of many ASUS AM4/AM5 motherboards, you'll find the CPU voltage setting labeled VDDCR CPU Voltage . Again, the typical modes will be Auto, Manual, and Offset, and most consumer motherboards won't have an Adaptive mode equivalent to Intel's , so the best approach is once again to use a negative Offset setting.
The basic procedure would be:
- Change VDDCR CPU Voltage from Auto to Offset mode.
- Select a moderate initial negative offset, for example -0,025 V.
- Save changes, start up and check temperature and stability.
A key difference with Intel is that, in Ryzen, boost performance is closely tied to temperature and the power budget allocated to the CPU. A well-executed undervolt on a Ryzen 7 5800X or a Ryzen 9 can achieve several degree lower temperatures while also better maintaining high frequencies in games and Cinebench , especially if your liquid cooling or stock cooler is barely adequate.
Just like with Intel, it's advisable to accompany each adjustment with rigorous stress tests . Tools like RealBench, AIDA64, or even OCCT for CPUs are great allies because they combine CPU, memory, and even GPU loads to simulate quite demanding real-world usage conditions.
Undervolting laptops (Legion, gaming, and similar) without touching the BIOS
Many gaming laptops, especially some Legion models and those from other manufacturers, have very restricted BIOS settings that don't allow free voltage adjustment. In these cases, the solution is to use Windows tools like Throttlestop, Intel XTU, or the manufacturer's own utilities (Lenovo Vantage, Armoury Crate, etc.) , provided the processor doesn't have undervolting blocked via firmware.
For years, Intel XTU and Throttlestop allowed Intel laptops to apply a negative CPU voltage offset directly from Windows, similar to how you do it in the BIOS. A significant change was the Plundervolt vulnerability, which led many manufacturers to block undervolting at the microcode or BIOS level , so on many recent systems it's no longer possible to adjust Vcore via software.
If your Legion or similar system still allows software undervolting, the idea is very similar:
- Install the tool (for example Throttlestop) and familiarize yourself with its interface.
- Apply a small negative offset to the CPU voltage (e.g., -50 mV).
- Test stability with demanding games and programs for hours., in addition to stress tests such as Cinebench or Prime95.
- Slow down gradually if everything is stable, always being careful not to be too aggressive.
In parallel, many of the manufacturer's own suites (such as Lenovo Vantage) allow you to select performance modes , fan profiles, and sometimes power limits.
If you can't adjust Vcore from either the BIOS or Windows due to manufacturer restrictions, the alternative is to experiment with power limits, thermal profiles, and ultimately, use cooling pads and fan settings . It's not as refined or elegant as a classic undervolt, but it can make a noticeable difference in the user experience.
GPU Undervolting: Lower Temperature and Noise in Modern Graphics Cards
Undervolting also applies to the GPU, and in many cases the results are even more spectacular than with the CPU . Current mid-range and high-end graphics cards have very high power consumption, forcing them to operate at their thermal limits, generating a lot of heat inside the case and making the fans work at maximum speed.
In GPUs, the goal is the same: to reduce the voltage associated with a given frequency to achieve the same graphics power with less heat and noise . The trick is finding a frequency/voltage combination that your specific chip can sustain reliably in the games or applications you use most.
ADM Adrenalin
For AMD graphics cards, the most recommended tool is the official AMD Adrenalin software , which includes advanced panels for adjusting frequency, voltage, power limits, and fan curves without having to access the card's BIOS. For Nvidia and third-party cards, MSI Afterburner is the classic tool for these tasks.
The usual approach in AMD Adrenalin would be:
- Install the latest official drivers in full mode to have Adrenalin available.
- Open the Performance tab → Settings and activate Manual or Custom mode.
- Enable advanced controls for GPU, memory, fan, and power.
- Increase the Power Limit to the maximum allowed (+20% is usually typical) to prevent that limit from throttling the GPU when you lower the voltage.
It might seem odd to raise the power limit when you're trying to achieve the opposite, but you're not actually forcing the graphics card to draw more power; you're simply removing an artificial ceiling so it can better maintain its frequency with less voltage . The savings will come from the undervolting itself, not from a power cap.
100 MHz Rule
A common trick with Radeon is to apply the so-called "100 MHz rule" : bringing the minimum and maximum frequencies defined by Adrenalin closer together, leaving a difference of about 100 MHz. This reduces frequency fluctuations and results in more stable FPS rates and better low 1%s , which is especially noticeable in competitive games.
The adjustment process usually follows this pattern:
- Measure temperatures, power consumption and frequencies with the factory settings for a few minutes under stress (demanding game or test such as OCCT 3D).
- Gradually reduce the GPU voltage in 10-20 mV steps from the Adrenalin panel or from Afterburner.
- After each change, subject the card to severe stress (OCCT in aggressive mode, for example) and watch for artifacts, errors, or driver restarts.
- When the test starts to fail or the controller freezes, you will have reached the limit and it is time to increase it slightly (10-20 mV) to leave a safety margin.
The typical benefits seen with a well-executed GPU undervolt are reductions of 8-10°C or more compared to stock values , with a significantly lower noise level and, sometimes, even a small performance improvement in long sessions because the card stops throttling its frequency due to reaching the thermal limit.
One user shared their experience, for example, by setting a reasonable frequency (around 1835 MHz) and finding the minimum stable voltage to maintain it, and their GPU supported that MHz at 875 mV. Depending on the silicon you get, another identical unit might stabilize at 825 mV or require 900 mV ; hence, there are no universal values, and everything depends on testing.
Is undervolting always worthwhile, or are there better alternatives?
Before you start changing voltages, it's important to consider exactly what you're trying to fix. Undervolting usually comes into play when you're experiencing overheating, noise, or excessive power consumption , whether it's in a laptop that's overheating or a desktop computer with a power-hungry graphics card in a small case.
However, it's not always the only or best solution. Often, improving the computer's physical cooling resolves the problems without needing to modify the BIOS . Some typical measures include:
- Install a better CPU cooler if you're using the stock one or a very low-end one.
- Ensure good airflow in the case: front intake, rear exhaust and, if possible, top exhaust.
- Reorganize cables and components so they do not block airflow.
- Install higher quality fans and position them in the correct orientation.
On desktop PCs, switching to a decent liquid cooler or a mid-range air cooler makes a huge difference in the temperatures of a Ryzen 7 5800X, a Core i7, or a Core i9 , reducing the need to adjust voltages. On laptops, a good cooling pad and thorough internal cleaning (removing dust and dried thermal paste) can also be key.
Ultimately, it's about understanding that undervolting is just another tool in the arsenal for taming your PC or laptop: it's neither mandatory nor a panacea, but when applied correctly, it allows very powerful machines to run cooler, quieter, and more efficiently —something especially valuable in the era of increasingly faster… and more power-hungry CPUs and GPUs.
