Linux Kernel 7.0-rc1 arrives packed with hardware support and performance work

Home » Python » Linux Kernel 7.0-rc1 arrives packed with hardware support and performance work

The Linux ecosystem is moving into a new phase with the arrival of Linux kernel 7.0-rc1, the first release candidate of the next major series. Even though the jump from 6.x to 7.0 is, as Linus Torvalds himself admits, mostly a matter of keeping version numbers manageable, this cycle happens to be particularly dense in terms of changes and groundwork for future hardware.

Far from being just a cosmetic version bump, Linux 7.0 is shaping up as a pivotal release for upcoming distributions and platforms. It is expected to become the default kernel for prominent releases such as Ubuntu 26.04 LTS and Fedora 44, meaning that the decisions and optimizations landing now are likely to define the Linux experience on desktops, servers, and laptops for years to come.

Linux 7.0-rc1 closes the merge window

With the release of Linux 7.0-rc1, the kernel merge window for this cycle is officially closed. As usual, the two‑week integration period has been followed by a freeze on major new features so that the code can stabilize ahead of the final 7.0 release.

Torvalds was characteristically direct in his mailing list announcement, noting that the new major version number is not tied to a dramatic architectural shift but rather to his preference for avoiding unwieldy minor versions once they near x.19. The result, however, is that this “coincidental” 7.0 ends up being one of the more feature‑heavy kernels in recent memory.

The source for Linux 7.0‑rc1 is already available and can be cloned from the official kernel.org Git repository. Over the coming weeks, subsequent release candidates will primarily focus on bug fixing, regression hunting, and polishing the large body of changes that made it in during the merge window.

As is customary, detailed feature overviews and in‑depth benchmarks are starting to roll out, providing an early look at how Linux 7.0 behaves under real workloads compared to the previous stable series. Early testing indicates that while some areas show promising gains, others still need tuning before final release.

Targeting next‑generation Intel and AMD platforms

One of the most striking aspects of this cycle is the sheer volume of work aimed at future Intel and AMD platforms. Linux 7.0 lays down extensive support for Intel’s Nova Lake and Diamond Rapids processors, as well as AMD’s upcoming Zen 6 architecture, ensuring that these chips will be ready to run Linux effectively as soon as they hit the market.

On the Intel side, the kernel introduces Nova Lake enablement across several subsystems. Notably, Nova Lake S platforms now have their identifiers wired into the Intel LPSS (Low Power Subsystem) driver, which handles interfaces like SPI and HS‑UART. Interestingly, this support mainly required adding new device IDs, suggesting that the existing driver model already fit the new hardware fairly well.

Diamond Rapids Xeon processors also receive focused attention, including support for NTB (Non‑Transparent Bridge) drivers and performance event monitoring. These changes should help system administrators and developers more accurately profile and manage these next‑generation server CPUs once they become available.

For AMD, Linux 7.0 brings additional Zen 6 performance events and metrics support, enhancing observability and fine‑grained tuning via performance counters. There is also new support for address translation features on Zen 5 within the CLX subsystem, indicating that the kernel developers are not just targeting upcoming architectures but also refining support for current generations.

Beyond x86, the kernel widens its scope with Atomic LS64/LS64V instruction support for ARM64 CPUs and user‑space CFI (Control Flow Integrity) features for RISC‑V. Additionally, mainline support has been added for the SpacemiT K3 RVA 23 SoC, continuing Linux’s trend of embracing a broad array of vendors and form factors.

DSA 3.0 and accelerators for data movement

Linux 7.0 also takes a significant step forward in accelerator support by merging updated patches for Intel’s Data Streaming Accelerator (DSA) 3.0. This hardware engine is designed to offload data movement and transformation tasks from the CPU, which can be particularly useful in data centers running workloads involving large‑scale copying, analytics, or streaming operations, especialmente en entornos que emplean tecnologías de contenedorización.

The new DSA 3.0 code introduces fresh sysfs interfaces exposing capability registers, allowing user‑space software to understand and leverage the additional options offered by the latest accelerator IP. Among the notable additions is Max SGL Size support, an important piece for operations such as Gather copy and Gather reduce, where scatter‑gather lists must be sized correctly before user applications can safely use them.

One interesting detail is that the DSA 3.0 sysfs interface bends the usual kernel conventions by placing three values in a single sysfs file, whereas the standard practice is typically one value per file. While this is technically an exception to the rule, it was accepted as part of the DMA engine pull for Linux 7.0, underlining how hardware complexity sometimes nudges the boundaries of long‑standing conventions.

These accelerator‑related changes are expected to pay off most notably in future Diamond Rapids‑based servers, where DSA 3.0 is presumed to appear. However, the benefits will only fully materialize once user‑space software stacks and frameworks are updated to detect and exploit these new capabilities.

Graphics, laptops and broader hardware enablement

While CPU and accelerator enablement dominate the headlines, Linux 7.0 also includes notable graphics and laptop‑related updates. On the GPU front, the kernel adds support for upcoming AMD graphics hardware, continuing the pattern of having kernel‑level plumbing ready ahead of new GPU launches.

Intel’s integrated roadmap is not left behind: Nova Lake display support for the iGPU enters the tree, setting the stage for future laptops and desktops powered by this architecture to have working display pipelines from day one. These display changes go hand in hand with ongoing updates to the Intel Xe graphics driver, which continues to mature around the newer Xe3 architecture.

Beyond pure GPU features, Linux 7.0 contains numerous laptop driver improvements and sensor monitoring additions, including support for reading sensors on more ASUS motherboards. This kind of incremental work is less eye‑catching than big architecture announcements but tends to have a very visible impact on daily usability, especially when it comes to thermals, fan control, and battery‑conscious performance modes.

Support for Apple hardware also advances. The kernel now wires up the RTC, HWMON, and input sub‑devices for the Apple System Management Controller (MACSMC) driver, and adds Apple USB Type‑C PHY support. Together, these elements gradually improve the experience of running Linux on recent Apple machines, even though such systems still require a fair amount of specialized work.

Rounding out the hardware enablement story, the multi‑function device (MFD) pull for this cycle introduces support for components such as the ROHM BD72720 and BD73900 PMICs, Rockchip RK801 PMIC, and additional network and storage‑related controllers like the Delta Networks TN48M and a TS133 variant for QNAP MCUs.

File systems, storage and I/O improvements

As usual, a large chunk of visible user impact comes from file‑system and I/O work. Linux 7.0 brings enhancements across several widely used file systems, with an eye on both performance and robustness.

Among the more user‑facing gains are better sequential read performance for exFAT and various updates to F2FS, which is popular on flash‑based storage. EXT4, one of the most common default file systems on Linux distributions, gains improvements related to concurrent direct I/O writes, aiming to reduce contention and improve behavior under heavy parallel workloads.

Beyond file systems, Linux 7.0 includes standardized generic I/O error reporting, which should help tools and applications diagnose storage issues more consistently. Multi‑lane SPI support and Octal DTR capabilities for SPI NAND devices are also part of this cycle, targeting embedded systems and storage solutions that rely on high‑throughput serial interfaces.

Other subtle but useful additions include non‑blocking timestamps, which can reduce contention in timekeeping‑sensitive paths, and various low‑level optimizations throughout the block and I/O stack. Many of these changes are incremental, but collectively they contribute to smoother behavior under mixed or demanding workloads.

On top of that, there are continued performance and stability refinements in the storage and memory subsystems. These may not come with flashy names or marketing slogans, yet they often matter more for day‑to‑day reliability than headline features do.

Performance tuning: wins, regressions and gaming‑relevant work

Performance is a recurring theme in Linux 7.0, with work spanning databases, schedulers, memory management and graphics. One highlight is notable PostgreSQL performance gains on AMD EPYC platforms, where targeted kernel improvements yield measurable throughput increases in database workloads.

There are also memory management optimizations and scheduler scalability updates that should benefit multi‑core and many‑core systems alike. Combined with improvements in various file systems, these changes set the stage for better performance under both server workloads and heavy desktop usage.

From a gaming and graphics standpoint, Linux 7.0 brings back large page support for Nouveau, the open‑source driver for NVIDIA GPUs. This is particularly relevant for the NVK Vulkan driver, which can make use of large pages to reduce overhead and potentially improve frame times and consistency in games and 3D applications.

Intel TSX (Transactional Synchronization Extensions) is now set to auto mode by default on supported CPUs. While TSX is a niche feature for many users, applications designed to exploit transactional memory can see benefits from a more adaptive, kernel‑managed configuration that takes advantage of hardware capabilities without requiring manual tuning.

At the same time, early benchmarks on Intel Core Ultra Series 3 Panther Lake systems paint a mixed picture. Tests comparing Linux 7.0 development kernels to Linux 6.19 stable on an MSI Prestige 14 laptop with a Core Ultra X7 358H and Arc B390 graphics indicate that performance in some scenarios is regressing rather than improving.

These measurements were carried out with the same toolchain, consistent “performance” power profile, and an almost identical kernel configuration apart from new options added in v7.0. The ongoing work now is to determine whether these regressions are Panther Lake‑specific or symptomatic of wider issues introduced during the merge window. Further cross‑platform benchmarking is underway, and any problems uncovered are likely to be targeted during the release‑candidate phase.

Developers and users interested in gaming performance have reasons to keep an eye on this cycle: recent kernel development has been emphasizing task scheduling, memory management, and graphics driver maturity, all of which are crucial for consistent frame pacing and latency‑sensitive workloads. While real‑world gains will depend on the game engine, driver stack, and distribution, the direction of travel is clearly oriented toward better end‑user experience.

Rust, tooling updates and community recognition

Beyond pure hardware and performance work, Linux 7.0 continues to evolve in how the kernel is written and maintained. A key symbolic step this cycle is the formal acknowledgment that Rust support is here to stay, effectively putting an end to the idea of Rust being just an experiment within the kernel.

Rust‑based components are still a small fraction of the overall codebase, but their presence is gradually expanding. The language’s focus on memory safety and modern tooling is seen by many developers as a useful complement to C, especially in areas such as drivers where bugs can have far‑reaching consequences. The 7.0 release reinforces that Rust is now a permanent part of the kernel’s long‑term evolution.

On the diagnostics side, existing tools have been enhanced as well. For example, turbostat now reports L2 cache statistics on newer Intel CPUs, giving power users and performance engineers deeper insight into how these processors behave under load. Being able to inspect more cache‑level metrics directly from a familiar tool can be helpful for debugging performance anomalies or tuning workloads.

This release also includes a more human element: the kernel’s CREDITS file now formally recognizes Stephen Rothwell’s long‑standing stewardship of Linux‑Next. After managing Linux‑Next since 2008, Rothwell stepped down in mid‑January, passing the baton to Mark Brown.

Linux‑Next acts as a staging area where subsystem branches and topic trees are integrated ahead of each merge window. It allows developers to test cross‑subsystem changes earlier and provides adventurous users with access to bleeding‑edge patches without having to manually assemble multiple repositories. The credit entry is a small but meaningful recognition of nearly two decades of work that has quietly underpinned the kernel’s day‑to‑day development process.

Looking across all these areas – from future‑proofing for Nova Lake, Diamond Rapids, and Zen 6, through to file‑system tuning, Rust integration, and community acknowledgments – Linux 7.0 stands out less for any single headline feature and more for the breadth of incremental, interconnected changes. With 7.0‑rc1 now available, the focus shifts to stabilization and performance fine‑tuning, but it is already clear that this kernel will serve as a foundational release for the next wave of Linux distributions and hardware platforms.