# Linux 7.4 will bring better battery performance for laptops

> The Linux 7.4 kernel adds battery control, thermal profiles, discrete GPU and fan support on ASUS, MSI, HP and Acer laptops and on OneXPlayer handhelds.
> By José Maldonado · Technology · 2026-10-16

The Linux kernel stakes its desktop reputation on details nobody looks at until they fail. The battery that won't stop charging at 80%, the fan that doesn't respond to the performance profile, the discrete GPU that doesn't power down and leaves the laptop running hot while doing nothing. That is why the next kernel version fixes a handful of those specific cases, and the list of machines it covers says a lot about where the installed base of Linux devices is growing.

All these improvements are piling up in the `for-next` branch of the x86 platform drivers, the drawer where the drivers that talk to each laptop's proprietary firmware live. They are nine closed changes, all already in that integration branch, and none of them is spectacular on its own. Together they are the difference between a machine that works and one that works well.

The detail that stands out most is not any of the patches, but the spread. Of the nine devices covered, two are handheld consoles from specialist Asian manufacturers and three are gaming-grade laptops. Office machines appear, but they don't dominate. When the kernel devotes effort to fan curves and tachometers, it is acknowledging that a significant part of its installed base stopped being the corporate laptop long ago.

## Nine changes you notice when you use them

The pattern repeats in every case. A board or model identifier that until now was not in the driver's table, and that therefore left functions the hardware does expose without a response. Adding it doesn't change the machine's performance, but it changes what you can control.

Manufacturer

Model or series

What is enabled

ASUS

ROG Flow X13 (GV302X\\\*) from 2023

Correct detection of tablet mode when the device is folded

ASUS

ProArt laptops (H7606 series)

Alternative to enable and disable the discrete GPU

MSI

Katana 15 B13VFK with recent firmware

Battery charge threshold

MSI

GL65 Leopard 9SCXK

Battery charge threshold

HP

OMEN 17-ck1xxx (board 8A17)

Thermal profiles

HP

Victus 15-fb3xxx (board 8DD0)

General support for the model

TUXEDO (Uniwill)

TUXEDO laptops

Keyboard backlight and lightbar control

Acer

Nitro ANV15-51

Thermal profiles, monitoring and fan control

OneXPlayer

OneXPlayer 3 with Intel

Battery bypass, capacity limits, fan curves, tachometer and turbo button override

AYANEO

AYANEO 2S

Battery charge control

The OneXPlayer 3 entry is the most complete in the batch. The OXPEC driver gets a quirk for the new model with an Intel processor that unlocks five functions at once: battery bypass, which lets the device run directly off the power supply without wearing down the cell; capacity limits; fan curves and the tachometer that makes them measurable; and the ability to override the turbo button. On a handheld console, those five things are the difference between playing for twenty minutes and playing for two hours.

## Handhelds are no longer an odd case

But there are two striking cases on the list. AYANEO and OneXPlayer are not manufacturers that just happen to show up in the kernel, and many will surely be thrilled about it. Both are companies whose main product is a handheld console running [Windows](https://observatorioblockchain.com/en/technology/novedades-windows-11-26h2-la-lista-completa-de-cambios/), and that have discovered that part of their customer base installs Linux and distributions such as Bazzite to get more out of the hardware. That both receive battery and fan support in the same merge window is no coincidence.

After all, the effect of these additions is cumulative. Every console that enters the driver's table no longer needs out-of-tree patches, and that reduces friction for anyone buying the device with Linux in mind. The cost of maintaining a local patch against every new kernel version is high, and removing that friction is the difference between a community that sustains support and one that tires of a product that doesn't work well.

On the laptop side, the ASUS Armoury patch deserves a separate mention. It adds an alternative way to enable and disable the discrete GPU on machines such as the ProArt H7606 series, which are aimed at creative work rather than gaming. On those machines, turning off the dedicated graphics when it isn't needed is what decides whether the battery lasts a day of light work.

## Battery control, the most missed feature

If there is a recurring gap for Linux users on laptops, it is the charge threshold. The feature exists in the firmware of almost every manufacturer and lets you cut charging at 80% to extend the life of the cell, something that matters especially on machines that stay plugged in all day. On Windows the manufacturer's software exposes it. On Linux it depended on the corresponding driver knowing the specific model.

The two MSI machines on the list and the AYANEO 2S come in through that door. The MSI Katana 15 B13VFK and the GL65 Leopard 9SCXK get the threshold through MSI's embedded controller driver, and the AYANEO 2S gets it through its own. It is the kind of patch that doesn't appear in any headline release note and that a user notices on day one.

The TUXEDO case is different and also illustrative. The Uniwill driver enables keyboard backlight and lightbar control on its laptops, which are machines designed and sold with Linux as the primary system. That a manufacturer whose value proposition is precisely Linux has to wait for a merge window for its keyboard to light up points to a bottleneck that isn't at the manufacturer.

## Support that is far from universal

Understanding where these patches come from explains why they don't all arrive at once. A modern laptop exposes much of its firmware functionality through a subsystem called WMI, the interface manufacturers use to publish proprietary methods, and through the embedded controller, the chip that manages battery charging, fan speed and temperature. Neither of those two paths is standardized. Basically, every manufacturer has its own mess, with some exceptions, such as Lenovo, which has had very good support for many years.

That means each manufacturer decides which methods it publishes, under what name and with what parameters, and in many cases changes the board identifier between revisions of the same model. That is why drivers accumulate tables of identifiers, and why the unit of work for the kernel in this area is the quirk (known by few, hated by many): a specific exception for a specific machine. The manufacturer's driver doesn't know what to do with the 8A17 board of the OMEN 17-ck1xxx until someone adds that entry.

The detail that makes the list useful is that most of these functions already exist in the hardware. There is no reverse-engineering discovery behind most of the nine cases: there is an interface that already responds, and a driver that doesn't ask it yet. That explains why the effort pays off so much per patch, and also why coverage depends on someone with that specific machine showing up willing to write the submission.

That is why two of the nine changes refer to specific MSI models rather than the whole range. The Katana 15 B13VFK entry only applies to recent firmware for that model, and the GL65 Leopard 9SCXK gets its own because its combination of board and firmware is different. For the user, that means something practical: two laptops from the same family and the same year can have different features if the firmware doesn't match.

## What about Linux 7.3 in the meantime?

It is worth placing these dates because the numbering moves fast. The seventh release candidate of Linux 7.3 was published on October 12, 2026 and, if nothing goes wrong, the stable version will come out at the end of the month. Ubuntu 26.10 has already been shipping with Linux 7.3 since October 14, barely a couple of days later, something strange considering that the branch is still considered unstable.

The Linux 7.4 merge window opens at the end of October 2026, when 7.3 is already closed. The stable 7.4 kernel will foreseeably arrive at the end of December, and from there another count begins: each distribution decides when it adopts it. Ubuntu will do so in its April 2027 release, which will also be a long-support one. Anyone using a continuous-cycle or rolling release distribution will see it sooner.

That lag explains why these patches matter even if they don't arrive tomorrow. An Arch, [Fedora](https://observatorioblockchain.com/en/technology/fedora-linux-explicado-distro-linus-torvalds/) or CachyOS user will try the AYANEO 2S charge threshold within weeks. A user of the current LTS may wait half a year or more, and it will depend on whether their distribution backports the specific feature. x86 platform patches are often carried over to earlier versions when they are small and isolated, but it isn't guaranteed.

## The kernel as a consumer product

For years, the promise of Linux on the desktop rested on a reasonable idea: if the hardware works, the system works. The list piling up for Linux 7.4 shows that the second half of that sentence is the hard part. The hardware works, and the system works, but communication between the two depends on a table of identifiers that someone has to maintain by hand.

What changes with this batch of patches is the proportion of machines where that communication is already resolved out of the box. And there is a second-order consequence worth watching: the more support relies on model-specific platform drivers, the more fragile the result will be when the manufacturer releases a board revision with a new identifier. The alternative, forcing manufacturers to expose standard interfaces, moves much more slowly than the list of quirks.

For the user buying a handheld console or a gaming laptop with Linux in mind, the practical signal is that support does arrive, but it arrives late and it arrives by accumulation. Checking whether your exact model is in the driver's table before buying is still more useful than any general compatibility promise.
