Phoronix

---

Intel today posted a set of patches for bringing SoundWire support in the upstream Intel AVS driver for Raptor Lake. While Intel SoundWire activities are not new, I was curious to see them focusing on Raptor Lake improvements in 2026 for what will then go into a 2027 Linux kernel version.

Intel is now upstreaming support for SoundWire on Raptor Lake P and Raptor Lake M platforms. This is done as SoundWire is being utilized for audio streaming on these Raptor Lake M/P platforms in the automotive segment.

Intel's "Malibou Lake" is using Raptor Lake SoCs for in-vehicle computing, digital cockpits, camera processing, etc. This is their first-generation Software Defined Vehicle (SDV) platform using Raptor Lake, but to be succeeded by Raptor Lake and Nova Lake based automotive platforms too.

It looks like Intel has been relying on downstream kernel flavors for their automotive work with the mention in today's patch series that it's the first time bringing this SoundWire work for Raptor Lake to the "public domain", a.k.a. upstream Linux kernel intent:

"The RaptorLake-P and RaptorLake-M based platform used in Automotive area do utilize SoundWire for streaming audio. This patchset brings such support to the public domain."

The patch series is out for review. Hopefully we continue to see more Intel automotive / software defined vehicle work heading up into the mainline Linux kernel.

---

AerynOS 2026.11 was released today along with a lengthy blog post highlighting the recent development work on this original Linux distribution.

AerynOS 2026.11 is now available as their newest release that also introduces lichen-installer as a new alternative installer. Lichen-installer hasn't yet replaced their current default installer but is now on the ISO for easier testing purposes. Lichen-installer should help in making the installation more robust.

AerynOS has also finally landed Fwupd within its repository for easy Linux firmware updating.

AerynOS is now offering the Budgie desktop as an option for AerynOS early adopters, including as an install-time option.

Some recent package updates for AerynOS include GNOME 51.1, KDE Plasma 6.8.2, COSMIC 1.11, Budgie 10.10.4, Mesa 26.2.6, QEMU 11.1.3, Wine 11.21, and many others. AerynOS 2026.11 is making use of the Linux 7.2.14 kernel.

More details on these recent AerynOS initiatives and the new 2026.11 ISO refresh can be found at AerynOS.com.

---

The Mesa RADV driver that provides Vulkan API support for AMD Radeon GPUs on Linux has adopted its own policy around AI / large language models (LLMs) for code contributions.

Beyond Mesa having its own AI/LLM policy, the Radeon RADV driver has instituted its own loose policy around AI/LLMs. Samuel Pitoiset of Valve's Linux graphics team and one of the prolific contributors to the RADV driver is the one that took the initiative and moved forward in adding this RADV-specific policy.

Ultimately it's a loose one in not outright forbidding AI/LLM-based contributions but discourages its use and leaves the door open to closing merge requests that were AI generated. The new policy is:

"If you find a bug or shortcoming in RADV and cannot address it yourself, file an issue on the tracker. LLM-assisted MRs require a good understanding of the codebase, and reviewing them has a real cost. Reviewers/maintainers may close such MRs at their discretion.

LLM-written issues are discouraged: they tend to be verbose and hard to read. Please describe the issue in your own words. LLM-generated reproducers are fine as long as you have reviewed them and they demonstrate the bug."

Pitoiset rationalized the need for this RADV policy due to beginning to see more AI-generated patches and wanting to get out ahead of them before the developers become overwhelmed. He wrote in the merge request for this policy:

"We are starting to get slightly more LLM MRs by external contributors these days for RADV and it seems better to state this now before getting flooded."

The policy was merged today to Mesa Git.

---

The FreeRDP 3.33 software released today as the newest version of this open-source Microsoft Remote Desktop Protocol (RDP) implementation.

FreeRDP has been working toward EmScripten and WebAssembly (WASM) support the past few years. With FreeRDP 3.33 they are now ready with their WebAssembly support. This new version is advertised as being able to run within your web browser when compiling it to WebAssembly.

FreeRDP 3.33 has also improved its SDL client, added experimental UDP support for direct RDP connections, Unicode keyboard input pipe is now supported by all clients, experimental Windows shadow server support, macOS file clipboard support without relying on FUSE, and initial support for translations of messages.

FreeRDP 3.33 was also impacted by the onslaught of AI/LLM activity... They note in this morning's release announcement:

"The influx of pull requests and feedback for bugs/regressions/... was huge, so we will do a patch release with all the papercuts and bug still in the backlog soon."

This includes FreeRDP 3.33 shipping with 13 known security fixes.

Downloads and more details on FreeRDP 3.33 via GitHub.

---

KDE developer Nate Graham published his monthly blog post on Sunday to outline the current state of KDE Linux and some of the changes made to this in-house Linux distribution over the course of October. Additionally, KDE Linux is nearing its beta milestone after beinbg in development for over one year as a reference Linux distribution for KDE.

For those installing KDE Linux for the first time, one of the improvements made in October is that KDE Linux's installer now has a basic slideshow. The KDE Linux installer slideshow is similar to the operating system installers for other Linux distributions.

On the kernel side, KDE Linux now enables the NTSYNC kernel module by default. With NTSYNC by default it enhances the out-of-the-box experience and performance for those gaming on Linux via Wine/Proton.

On the desktop side, wayland-protocols is added to the KDE Linux base image to broaden the OS support for some KDE projects depending upon it.

KDE Linux is nearing its beta milestone and is roughly 85% of the way there as of the start of November. The milestones page shows work still needing to be done on the swap support, improving the update experience within Discover, KDE infrastructure changes, improved dual booting experience, better handling Flatpak crashes, and other improvements.

More details on these recent KDE Linux changes via the October 2026 report.

---

Raspberry Pi announced their latest round of price hikes as a result of the RAM shortages and increased pricing continuing to affect the industry.

Today's round of price changes only affect the Raspberry Pi products with 16GB of RAM, but it's a significant one. The Raspberry Pi 5 16GB is increasing in price by $80, now to a price of $385 USD. How the RAM pricing has affected the industry when thinking about the days when most Raspberry Pi products were $35 ~ $60.

The Raspberry Pi Compute Module 5 16GB is facing a $40 increase.

Eben Upton announced today's Raspberry Pi 16GB product price hikes on the Raspberry Pi blog.

---

Valve just published their Steam Survey results for October and point to a small uptick in Linux gaming use on a relative basis.

According to today's published Steam Survey figures, for October 2026 the Steam on Linux use was up 0.05% and taking it to 3.95% overall. This is quite a bit different from the original Steam Survey figures for September that showed Linux diving down to 3.05% as a 0.85% decline over August. But as noted in that article last month, it was closely associated with a strong shift in Simplified Chinese use. When that occurs in the Steam Survey, Microsoft Windows typically sees large upward swings against Linux.

In any case Valve revised their September numbers to 3.9% and now for October are reporting 3.95% for their overall Linux customer base. Apple's macOS is at 2.35% and Windows at 93.7%. In comparison, this time last year we were celebrating Steam on Linux finally cracking 3% during October 2025.

Among the Linux distributions, Arch Linux based CachyOS remains most popular at 16.3% and even outpacing SteamOS at 13.89% in second place.

AMD CPU use by Linux gamers remains at just under 70%, driven in part by the Steam Deck and Steam Machine using custom AMD SoCs.

All the Steam Survey results in full for October can be found at SteamPowered.com.

---

Even with Linus Torvalds having been on holiday, Linux 7.4-rc1 is out today on-schedule for capping off the Linux 7.4 merge window.

Linux 7.4 brings many exciting changes. With v7.4-rc1 out today on schedule without the possible week delay Linus Torvalds talked about in the v7.3 announcement, it still stands good chances of releasing before the end of December. With that, Linux 7.4 will likely end up being this year's Long Term Support (LTS) kernel version. That is assuming no frightening delays toward the end of the kernel cycle that would drag it out. As it stands now, if Linux 7.4 releases on time it will likely be here on 20 December.

Among the many changes in Linux 7.4 is the new AMD PerfOpt feature for Radeon iGPUs to help enhance AI/LLM performance, AMD Ryzen Zen 6 support getting all squared away ahead of those next-gen client CPUs launching, a lot of laptop quirks/additions, AMDGPU HDMI 2.1 enabled by default, Intel IPU8 support for upcoming Nova Lake laptops, more Qualcomm Snapdragon X2 laptop enablement, optional buffered IOmap support for EXT4, basic Apple M4 support enablement albeit not yet ready for end-users, Google Tensor G5 SoC and Pixel 10 device support with the mainline kernel, faster kernel build times, and much more.

In the next day or two I will be out with my usual feature overview to summarize all the interesting Linux 7.4 kernel features/changes, beyond my dozens of individual articles in recent weeks. This is a BIG kernel with the most commits ever during the merge window; AI is in large part driving that huge upward trajectory in commit/patch count.

Stay tuned for the Linux 7.4 feature overview on Phoronix followed by Linux 7.4 kernel benchmarking getting underway.

Update: Linus Torvalds is now out with his customary release announcement:

"This is a *BIG* merge window. I think the reason is the by now obvious one, but I'm actually taking it as a good sign: I'm *much* happier with the flood of AI triggered work going into the merge window than it showing up mainly as extra fixes during the -rc period. So I'm hoping that this merge window being larger than usual is a sign that the AI fixes are becoming more part of the "normal development" than a flood of fixes that go into the whole stabilization phase. I think that's a much healthier model. But we'll see how the -rc's look.

That said, a small part of the "this is a big merge window" may also be that people are expecting 7.4 to be the next LTS release, which is another independent reason why some releases tend larger than others. So it may not *just* be the flood of AI patches.

Whatever the reason(s) for a big release, it's huge both in commits - almost 20k commits vs the more common ~15k - and as a diff. Although the diff is large partly for the other traditional reason: yet another AMD hardware register dump into huge headers. A full third of the patch is just AMD GPU register offsets and bit masks. We've been here before, and I'm sure we'll see it many times again in the future. Just ignore the noise."

---

A new KDE terminal emulator is in development as a modern alternative to Konsole. Kwi is this new terminal app being built atop the Ghostty library.

KDE developer Eike Hein announced the development today of Kwi for a fast and modern terminal that complies with the KDE HIG and UI conventions. Kwi is employing Ghostty's libghostty-vt library and other mature KDE Frameworks. In turn the Kwi terminal supports GPU acceleration via OpenGL or Vulkan with the Qt Scene Graph.

"3x–35x faster overall performance, depending on the benchmark E.g. 870 fps in DOOM-fire-zig on a test system, vs. 23 fps in Konsole"

Kwi is self-described as offering 3x to 35x faster overall performance than Konsole, complies with the latest KDE HIG standards, and a more modern architecture over Konsole.

Kwi for now is considered experimental and not yet proposed as a replacement to Konsole. The current state is summed up as:

"Kwi is well underway and has seen limited testing by a small circle of friends, but it's also so very hot off the presses there's not even a tagged v0.1.0 yet. As such, please greatly temper your expectations and expect obvious and immediate bugs :-). It is in no way a match for the 29 years of continuous development work on Konsole, used by millions every day."

Those wishing to learn more about the Kwi terminal emulator can do so via the KDE.org blog.

---

After writing this morning with some stats about the size of the Linux 7.4 merge window and the kernel's source tree approaching 42 million lines, I was curious to run some numbers for putting the size of the merge window into perspective against the major kernel releases of the past 15 years. This shows the definite recent uptick due to AI/LLM activity.

The Linux kernel development pace has been on a definite upward trajectory over the past year. While, yes, Linux popularity is likely at an all-time high from consumer devices to massive growth in data centers, etc, a lot of the uptick in kernel patches is due to AI/LLMs. With Linux kernel maintainers being "completely overwhelmed" due to both AI generated bug reports and patches/fixes, the Git commit activity has been surging as a result.

For visualizing the impact, here is a look at the number of Git commits per Linux kernel merge window each release going back to Linux 3.1 -- released in 2011:

The Linux 7.4 merge window by far is the largest merge window ever in terms of commits/patches. Nearly 20k! The upward trend has continued building the past several kernel releases. Typically the merge windows saw 13~14k commits or less and now we are pushing 20k. Here's all the raw numbers for each merge window for those curious:

Linux 3.1: 7620
Linux 3.2: 10730
Linux 3.3: 9460
Linux 3.4: 9859
Linux 3.5: 10069
Linux 3.6: 9161
Linux 3.7: 11084
Linux 3.8: 11738
Linux 3.9: 10942
Linux 3.10: 12715
Linux 3.11: 10161
Linux 3.12: 10153
Linux 3.13: 11356
Linux 3.14: 11367
Linux 3.15: 12826
Linux 3.16: 12078
Linux 3.17: 11529
Linux 3.18: 10363
Linux 3.19: 12142
Linux 4.0: 9591
Linux 4.1: 11410
Linux 4.2: 12809
Linux 4.3: 11431
Linux 4.4: 12226
Linux 4.5: 11032
Linux 4.6: 12929
Linux 4.7: 11518
Linux 4.8: 12423
Linux 4.9: 15095
Linux 4.10: 12302
Linux 4.11: 11747
Linux 4.12: 13729
Linux 4.13: 12051
Linux 4.14: 12352
Linux 4.15: 13489
Linux 4.16: 12654
Linux 4.17: 12633
Linux 4.18: 12373
Linux 4.19: 13108
Linux 4.20: 12883
Linux 5.0: 11551
Linux 5.1: 12032
Linux 5.2: 12792
Linux 5.3: 13390
Linux 5.4: 13286
Linux 5.5: 13452
Linux 5.6: 11454
Linux 5.7: 12748
Linux 5.8: 15086
Linux 5.9: 13681
Linux 5.10: 14770
Linux 5.11: 13261
Linux 5.12: 11749
Linux 5.13: 15069
Linux 5.14: 13764
Linux 5.15: 11224
Linux 5.16: 13137
Linux 5.17: 11812
Linux 5.18: 14115
Linux 5.19: 13973
Linux 6.0: 14378
Linux 6.1: 12288
Linux 6.2: 14595
Linux 6.3: 13565
Linux 6.4: 13871
Linux 6.5: 12492
Linux 6.6: 13161
Linux 6.7: 16203
Linux 6.8: 13052
Linux 6.9: 13242
Linux 6.10: 12402
Linux 6.11: 12954
Linux 6.12: 12115
Linux 6.13: 12142
Linux 6.14: 10034
Linux 6.15: 13568
Linux 6.16: 13805
Linux 6.17: 12289
Linux 6.18: 12910
Linux 6.19: 13229
Linux 7.0: 12468
Linux 7.1: 13963
Linux 7.2: 14395
Linux 7.3: 16267
Linux 7.4: 19671

Granted, we could still see a few more trickle in today for Linux 7.4 ahead of the expected Linux 7.4-rc1 kernel release later in the day.

With much of the AI/LLM activity being around fixes, the surge in patches don't stop at the merge window. Here is a look at the number of Git commits for each kernel cycle in full from prior release to the tagged stable version:

Linux 7.3 was the first time breaching 20k commits for an entire kernel cycle (20,260 commits from v7.2 to v7.3). Now with Linux 7.4 we're seeing nearly that much just during the merge window.

---

Merged this weekend for Linux 7.4 and also marked for back-porting to current stable kernel versions are a number of fixes to the OverlayFS union mount file-system.

OverlayFS changes for Linux 7.4 just consist of many fixes, including some with security implications. These fixes are marked for back-porting to the stable Linux kernel versions too.

Chris Mason of Btrfs notoriety and who has recently been focused on AI/LLM driven code review for the Linux kernel ended up uncovering a number of OverlayFS bugs. Amir Goldstein explained in the pull request of the OverlayFS updates:

"- Fixes for bugs found by Chris Mason's AI review, several of them reachable by unprivileged users: an info leak, use-after-free and NULL deref bugs, an RCU walk stall with remote layers and data loss with nfs_export

- Fix copy up of files with an empty security.capability xattr, which any user could set to block copy up

- Fix a regression from v6.16 with the override_creds mount option

- Two casefold fixes"

There ended up being more than a dozen patches to OverlayFS in the past week from Chris Mason after uncovering these issues in the code. With the information leak, UAF, and related bugs reachable by unprivileged users there are security considerations.

These OverlayFS fixes merged ahead of Linux 7.4-rc1 and to be back-ported to stable branches in the coming days can be found via this Git merge.

---

October brought the releases of Linux 7.3, Ubuntu 26.10, the KDE Plasma 6.8 desktop, and a wealth of other open-source software releases. The expanding use of AI/LLMs within Linux kernel development also made for interesting discussions in October. Microsoft bringing their Defender software to the Linux desktop was perhaps the biggest surprise out of October. Here is a look back at the past month.

During October on Phoronix were 308 original news articles and 16 Linux hardware reviews / multi-page featured benchmark articles. A lot of interesting technical content on Phoronix each and every day. Before getting to the most popular articles of the month, the usual monthly public service announcement... If you enjoy all of this content on Phoronix over the past 22+ years, please consider showing your support by joining Phoronix Premium to help support the site while enjoying ad-free viewing, multi-page articles on a single page, native dark mode, and other benefits. Tips via Stripe and PayPal are also much appreciated. Or at the least, please turn off any pesky ad-blocker when viewing this site. Due to the unfortunate state of the ad industry and squeeze among independent online publishers, operations remain very difficult.

The most popular news for October 2026 included:

Several Linux Kernel Vulnerabilities Accidentally Posted To Public Kernel Mailing List
Seemingly as an accident, overnight several new Linux kernel vulnerability reports were posted to the public Linux kernel mailing list rather than the private security mailing list. The vulnerabilities have yet to be patched on mainline but at least in some cases there are patches undergoing testing or queued in "-next" branches. The CVSS scores are as high as 9.0.

Linus Torvalds Working On A Hefty Linux Kernel Improvement With Claude Code
Linux creator Linus Torvalds has been known to use AI to help with kernel debugging and outside of the kernel tree has also done some vibe coding with Google's Antigravity. He has previously said he views AI as a useful tool. It turns out he's even now using Claude Code for working on new Linux kernel features.

The Amazing Work By Valve's Timur Kristóf On Improving Old AMD GPUs On Linux
Over the past year Timur Kristóf of Valve's Linux graphics driver team has made multiple very nice improvements to the AMDGPU kernel driver for enhancing support for old (GCN 1.0/1.1 era from a decade ago) graphics cards so that they can better handle Linux gaming and other tasks. This week in Toronto, Kristóf presented on this AMDGPU work that he initially began as a kernel driver development exercise after initially spending years in user-space focused on the Mesa 3D driver code.

Linus Torvalds "Did What I Preach", Used AI To Make Code Decipherable & Uncovered More Bugs
After using Claude Code to draft code to help make a fundamental improvement to the Linux kernel, Linux creator Linus Torvalds continued using AI this weekend and talking up its benefits. Beyond generating initial patches and a test program for that noted I/O enhancement, he separately talked up using AI/LLMs for making patches more decipherable to understand in the process the AI noted more long-standing bugs within the Linux kernel.

Microsoft Releases Microsoft Defender For The Linux Desktop
Microsoft's newest software release for Linux is... Microsoft Defender for the Linux desktop. Microsoft Defender has been ported to Linux to "secure Linux desktops" with Microsoft acknowledging that "more developers begin to use Linux desktops" and thus bringing their security application and threat-protection product to Linux desktops to complement their prior ill-used Linux server support.

Siemens Slams The Door Shut On Promising Open-Source Radioss Project
As a major disappointment to the open-source community, Siemens has decided to end -- and close-up access -- to the OpenRadioss project as the open-source version opened up by Altair Engineering under the GNU AGPL License back in 2022. This was an open-source version of the industry-leading finite element solver that Altair developed. Siemens acquired Altair Engineering last year and have now completely shut the door on OpenRadioss.

Linus Torvalds' AI Code Improved: Huge Performance Wins With Reworked Code
Last week I wrote about Linus Torvalds leveraging AI to come up with a hefty Linux kernel improvement using Claude Code. The code is for reading from the page cache without folio references to help workloads of small reads from a single file across many CPU threads. Meta engineer Kiryl Shutsemau has taken that code and reworked it and showing extremely promising performance results.

NVIDIA's Results Of Conducting Wayland vs. X.Org Latency Testing
In addition to NVIDIA announcing their Display Config Server project at XDC 2026 in Toronto, at this annual developer's conference they also presented their findings on scientific testing of X.Org vs. Wayland latency.

Meta Developing Compressed RAM "CRAM" For Linux: Better Than ZRAM & Zswap
With today's memory shortages, high prices, and ever increasing memory capacity needs, Meta engineers have been working on Compressed RAM "CRAM" for Linux. This hardware-offloaded compression still allows cacheline/byte access to compressed memory and in turn a better implementation than the likes of ZRAM and Zswap. Very promising is that there is near-native performance to raw DRAM speeds with CRAM.

FAT32 File-System Seeing As Much As 265x Improvement With Linux 7.4
A change merged for Linux 7.4 is showing some mighty big speed-ups for the FAT32 file-system.

AI Makes "Open-Source, Clean Room" Implementations Of Adobe Photoshop & Premier In Rust
As an interesting demonstrator for the power of AI and potentially pressing legal challenges, Claude Code and AI agents have created "open-source, clean-room reimplementation" versions of Adobe Photoshop, Adobe Premier, and Adobe Lightroom. These new creations from the ArtCraft project are implemented in the Rust programming language.

Linux's Early Boot-Time Physical Memory Allocator Sees Big Speed-Up: ~268s To ~50ms
Sent out today for the Linux 7.4 kernel were a set of memblock improvements. Most notable is a performance optimization within mm/memblock.c that implements the kernel's early boot-time physical memory allocator that is used prior to the bring-up of the full page allocator.

Linus Torvalds Eliminates Unsafe __get_user() & __put_user() Functions In Linux 7.4
Linus Torvalds continues with his recent uptick in kernel patches himself, assisted in part by AI with Claude Code. Merged today was removing the historical and unsafe __get_user() and __put_user() kernel functions.

Claude Code / AI Shows Potential For Reverse-Engineering Hardware For Linux Support
AI/LLMs like Claude Code are beginning to prove capable of helping to reverse engineer hardware for Linux support, especially when it comes to the tedious reverse engineering tasks.

Linux 7.4 To Introduce Initial Device Trees For Apple M4, A18 Pro For MacBook Neo
Asahi Linux developer Sven Peter today sent out the pull requests of the Apple SoC Device Tree changes they are ready to upstream for the Linux 7.4 merge window happening later this month. Most notable is the initial Device Tree for Apple systems using the base M4 SoC model as well as for the MacBook Neo with the A18 Pro SoC.

Today's Steam Beta Makes The Linux Client A 64-bit App
Valve just dropped a new Steam client beta and it's a big one for Linux gamers.

New Wayland Protocol Proposed For VNC/Remote Software
Prominent Wayland developer Simon Ser sent out a merge request today for introducing the new ext-transient-output protocol. With ext-transient-output, it's intent is for helping remoting software like VNC servers in a Wayland world.

Linux Kernel Developers Plan To Remove "Assisted-By" Tag Requirement For AI/LLM Usage
Currently there is a policy requirement for Linux kernel patches where AI tools contributed, they should be documented with an "Assisted-by:" tag to make clear AI was used in the creation of said patch. But Linux kernel developers are now looking at dropping this requirement amid the rampant use of AI/LLMs.

Fedora 45 Hits A Second NO-GO: Not Releasing Until November
Fedora 45 wasn't released this week as its ideal early release target due to not having a release candidate build in time for last week's Go/No-Go meeting. At today's Go/No-Go meeting, Fedora 45 was delayed by at least one more week due to open blocker bugs.

Omarchy Gets The Web Browser Launching Faster By Avoiding XDG Overhead
Omarchy 4.0.5 released this weekend as the newest version of this Arch Linux based distribution built around the Hyprland desktop. Omarchy 4.0.5 pulls in the latest Linux 7.2 stable kernel with BORE scheduler, updates against the Qt 6.12 toolkit, and a variety of smaller improvements.

And the most popular Linux hardware reviews of last month:

CachyOS vs. Fedora 45 vs. Ubuntu 26.10 Performance On AMD Ryzen 9
With the recently released Ubuntu 26.10 and the nearly-released Fedora Workstation 45, Phoronix readers have begun inquiring about fresh benchmarks looking at how they stack up against the performance-optimized, Arch Linux based CachyOS. Up today are our first CachyOS vs. Fedora 45 vs. Ubuntu 26.10 benchmarks on an AMD Ryzen 9 9900X3D2 desktop system.

Snapdragon X2 Performance On Linux Exceeding Expectations In Initial Benchmarks
For your viewing pleasure today are the first Linux benchmarks of the Qualcomm Snapdragon X2 Elite Extreme on Linux using the new Ubuntu 26.10 X1E Concept ISO for Snapdragon X laptop support on Linux. The Snapdragon X2 Elite Extreme flagship SoC is tested within the ASUS Zenbook A16 UX3607OA laptop.

Fedora 45 Beta Performance: The Satisfyingly Boring Benchmarks
While there have been a number of Ubuntu 26.10 development benchmarks in recent weeks and more on the way, some Phoronix readers have asked about the lack of any early Fedora 45 tests given its recent beta. For good reason, it's largely boring but here are some to illustrate.

NVIDIA Vera HPC Performance Benchmarks Take On AMD EPYC, Intel Xeon
While most everyone understandably likes to talk about AI these days, one of the areas of NVIDIA's Vera CPU I have been most eager to test has been for conventional high performance computing (HPC) workloads. HPC was talked up as one of Vera's strengths thanks to its 1.2TB/s LPDDR5x memory bandwidth and custom Olympus cores. Vera is seeing deployments at HPC labs like Los Alamos. With recently having unfettered remote access to a dual NVIDIA Vera server, I ran through a variety of HPC workloads and similar performance benchmarks. Here is a look at how NVIDIA Vera performs against AMD EPYC and Intel Xeon, including with CPU power and frequency metrics.

Star Labs StarFighter Ultra: A Modern High-End Linux Laptop With Coreboot
When it comes to Linux pre-loaded laptops, I have looked at hardware from most of these vendors over the past 22+ years of Phoronix. One vendor I haven't had the chance to personally test until now is the UK-based Star Labs. That changed recently and over the past few weeks I have been testing out their new StarFighter laptop that offers great Linux support and comes pre-loaded with Coreboot and streamlined firmware updates via LVFS. For those after a high-end mobile workstation, the Star Labs StarFighter offers a lot of performance potential while being open-source friendly and with Coreboot in tow.

AMD ROCm vs. Vulkan Performance For Lemonade Local AI Server With Llama.cpp
A Phoronix Premium reader recently inquired about seeing some new benchmarks of AI performance between Vulkan and AMD ROCm back-ends to see how the situation plays out today. Here are some fresh benchmarks on two separate systems with AMD Radeon graphics and using the Lemonade local AI server while exploring the difference between the ROCm and Vulkan Llama.cpp back-ends.

Framework Laptop 12 Offers Improved Performance & Efficiency With Intel Core 3 Wildcat Lake
Last year Framework Computer launched the Framework Laptop 12 as a convertible, 2-in-1 laptop that was friendly with Linux and upgradeable inline with the beloved Framework Laptop 13 and 16 models. The initial Framework Laptop 12 was nice but the use of an Intel Raptor Lake U SoC in 2025 wasn't all that exciting. Now though Framework Laptop 12 is available with Intel Core 3 "Wildcat Lake" for offering better performance and power efficiency.

LPDDR5 5467 vs. 6400 vs. 7467 MT/s Memory Performance Benchmarks For Laptops
With the recently reviewed Star Labs StarFighter Ultra Linux laptop its BIOS allows for configurable memory speeds unlike many other laptop vendor BIOS implementations these days. Star Labs also made the default design decision to run the LPDDR5-7467 rated memory at LPDDR5-6400 speeds in the name of a balanced user experience for power/performance. So I was curious to run some benchmarks quantifying that power and performance impact of LPDDR5-6400 vs. LPDDR5-7467 plus tossed in LPDDR5-5467 testing as well as the third choice offered with this StarFighter Ultra laptop.

ROCm 10 OpenCL vs. Mesa 26.3-dev Rusticl Performance On AMD Strix Halo
With all the recent progress on Mesa's Rusticl driver as a Rust-based OpenCL driver working across the different Mesa Gallium3D hardware drivers, a Phoronix Premium supporter inquired about seeing some fresh benchmarks of this open-source OpenCL solution. Here is a look at how the Mesa 26.3 Rusticl driver is looking against AMD's official OpenCL driver within ROCm 10.

NVIDIA Vera Single vs. Dual Socket CPU Performance
For those curious how NVIDIA Vera is scaling between single and dual socket deployments, here are some benchmarks I carried out a while back exploring the power and performance from one to two Vera CPUs.

---

Merged this weekend to Linux 7.4 Git were all of the perf subsystem tooling updates. Most notable to these tooling updates for dealing with performance counters and more on Linux is an overhaul to the Python code.

With Linux 7.4, all of perf's Python-based scripts have been adapted to using the Python perf module and supporting standalone execution. The Python module has also been improved with lazy sample callchains and event copies, GIL handling, better session/thread lookup, new treport, flamegraph and sqlite/postgres exporters.

The Python perf script conversion to using the module was previously covered on Phoronix within Linux's Perf Script Command Much Faster With Overhaul As Python Module. Google's Ian Rogers created a proper perf module as a Python C extension and leading to much faster performance than the prior code.

The perf tool updates for Linux 7.4 additionally bring instruction tracking on ARM64 for data type profiling, resolving stack, global, per-CPU and 'current' variable accesses. This allows perf annotate --data-type to behave similarly on ARM64 to Linux x86/x86_64.

There is also an assortment of other fixes and minor improvements with the perf tools as outlined in the pull request.

---

Linux 7.3 started out just under 41 million lines in total in the kernel source tree. In the end the Linux v7.3 tag came in at 41,019,555 lines as measured by cloc while now with Linux 7.4 we are inching ever closer to 42 million lines. On a Git commit basis, the Linux 7.4 merge window has been the busiest on record.

Linux 7.4-rc1 will likely be released later today. The kernel merge window is typically two weeks but in the Linux 7.3 announcement, Torvalds did note the possibility of an extra week to the merge window since this cycle collided with his holidays. However, as he appears caught-up with all of the merge requests thus far, he will presumably go ahead and release Linux 7.4-rc1 later today rather than dragging it out an extra week for little reason.

In any event, I was curious how Linux 7.4 is sizing up after the very busy merge window. With Linux 7.3 it came in at 41.019 million lines between 5,138,701 blank lines in code files, 4,918,548 lines of comments, and then 30,962,306 lines of detected actual code by cloc.

As of this morning in Linux Git ahead of v7.4-rc1, the kernel source tree is up to 41,605,120 lines. That's from 5,193,578 blank lines, 4,993,604 lines of comments, and 31,417,938 lines of detected code. Or nearly half a million lines of code added during the Linux 7.4 merge window, ignoring comments and blank lines.

The kernel source tree continues to grow and its likely that by the time of Linux 7.5 in the new year it will be past 42 million lines. Just some food for thought for those curious about the size of this merge window and how big the kernel codebase is up to, especially in this era of AI/LLM driven activity.

Or in looking at Git commits rather than L.O.C., since Linux 7.3 there have been 19,671 commits. In comparison from Linux v7.2 to v7.3-rc1 was 16,267 commits. Or from Linux v7.1 to v7.2-rc1 was 14,395 commits or 13,963 commits back during the v7.1 merge window. The Linux 7.4 merge window has seen significantly more commits than typical for a Linux kernel merge window period.

Update: Follow-up with more insight in The Linux 7.4 Merge Window Is Easily The Largest In Kernel History.

---

Just a friendly heads up that Phoronix.com and associated sites will be down briefly at around midnight PST / 3AM EST on 1 November for a failed drive in the NVMe SSD RAID array.

Unfortunately the drive can't be hot-swapped so our friends at HiVelocity will need to power down the server. Downtime will hopefully be minimal.

It must be all the AI scraping bots and similar that have been hammering the site especially hard this year... The failed drive was just replaced one year ago.

Update: All complete and back online!

---

OpenShot is closing out the month of October by releasing OpenShot 4.1 as the latest major feature release for this useful open-source video editing solution.

OpenShot 4.1 finally brings native text editing to its video editor to allow adding and styling text directly without requiring any external editor. Another big addition is Auto-Optimize to provide small, fast optimized copies of heavy videos for instant previews while the export is maintained at the higher 4K+ quality.

One exciting me is CLI automation. The intent here is for scripts or AI agents like Claude Code and Codex to be able to edit videos with OpenShot. I'm excited about OpenShot's CLI Automation with scripting support for hopefully being able to make an automated performance benchmark of OpenShot.

OpenShot 4.1 also brings improvements to exports, keyboard editing and preview enhancements, faster editing and better audio support, and a variety of other improvements. On the packaging side, OpenShot 4.1 adds ARM64 builds to complement their x86_64 binaries.

OpenShot 4.1 downloads and more details on the new version at OpenShot.org.

---

Enlightenment 0.27 released at the beginning of 2025 and since then there's been no major feature update, falling off their roughly 12 month release cadence. But out this Halloween is Enlightenment 0.28 now with the better part of two years worth of changes.

Carsten Haitzler released Enlightenment 0.28 today and it comes with initial NetBSD operating system support, a NetworkManager module for wired and wireless networking management along with VPNs, and various other network integration improvements.

Enlightenment 0.28 also comes with native PipeWire mixer support while still retaining support for PulseAudio.

There is also a variety of other audio-related enhancements to this Enlightenment update.Some of the other changes include new window actions, improved client side decoration handling, improved display mode sorting, improved backlight filtering, improved window siwtching, and a wide variety of fixes and code cleanups. On the low-level side is also improved pixmap handling, X connection recovery, and nested Wayland compositor shutdown.

Enlightenment 0.28 continues to boast both X11 and Wayland support. Downloads and more details on this Enlightenment 0.28 release via Enlightenment.org.

---

Linux Mint is out with their October 2026 status report to outline various initiatives they have underway in their lead-up to the Linux Mint 27 stable release happening on/around Christmas.

Linux Mint developers have been working on bringing back hibernation support to Linux Mint after upstream Ubuntu had removed hibernation from the supported sleep modes. Via Linux Mint's System Administration tool it will be possible to enable hibernation support for those wanting it as an alternative to suspend.

Linux Mint's Nemo file manager added a preview pane for images as another desktop usability enhancement of Linux Mint 27.

A brand new app being developed for Linux Mint is Superlight. The Superlight app is for system monitoring with CPU, RAM, GPU, network, and storage monitoring along with the likes of temperatures and fan speeds. Yet another system monitoring tool for the Linux desktop.

Linux Mint is also developing Velocity as a new terminal emulator.

Lastly is Unfold as a small archive manager.

More details on the hibernation support and new apps via the Linux Mint blog.

---

A new feature merged last week for Linux 7.4 to help reduce AI/LLM generated noise from invalid/irrelevant testing configurations is now reverted due to a "crazy beast" in the VFIO subsystem.

Greg Kroah-Hartman back in September crafted the TAINT_FORCED_BIND option to taint the running kernel if Linux's bind/unbind sysfs interface is used. This interface allows arbitrarily binding/unbinding hardware to a given driver. In the era of AI/LLM, this interface was abused by Syzbot and friends for trying to bind hardware to drivers not relevant to one another. This resulted in noise from irrelevant bug reports from these tools/bots that don't match any legitimate kernel use-case.

So with the TAINT_FORCED_BIND, it would be clear to kernel developers in bug reports that sysfs bind/unbind was used and so helping quickly deduce it may not be a legitimate bug report.

The feature was merged last week for Linux 7.4 for what Greg KH characterized as a recent major fuzzing "attack" against the kernel.

Only after TAINT_FORCED_BIND was it realized VFIO code is relying on the bind/unbind behavior in legitimate practice. In the revert patch, Greg described VFIO as a "crazy beast" in its usage:

"Turns out that vfio is a crazy beast and people have been "relying" on the bind/unbind debugging feature as a way to handle passing through devices to guest instances. Because of this (https://xkcd.com/1172/), let's revert the feature for now until we can figure out a more [agreeable] way to handle this such that normal workflows do not taint the kernel.

Odds are something like what Uwe proposes at:
https://lore.kernel.org/lkml/[email protected]/
will be the best way forward and will be revisited after for after 7.4.

The feature was originally intended to stem the flow of unnecessary fuzzing reports of "broken" drivers where random devices were being bound to random drivers and of course things were going wrong. The main syzbot tool is now not doing this anymore, and hopefully no one else starts it up again (but in this madness of LLMs these days, you can't be sure.)"

VFIO relies on bind/unbind for detaching PCIe devices from a Linux host driver and attaching to vfio-pci for use within VMs.

The revert was merged on Friday night.

Uwe Kleine-König's proposal is for using TAINT_DRIVER_OVERRIDE just in the usage of driver_override and disabling driver_override by default while allowing Linux driver the option of allowing themselves into it. We'll see if that ends up being adopted for Linux v7.5 but is now too late for v7.4 with the merge window expected to close tomorrow.

---

I love a good performance optimization and there are several this week in the KDE space with Plasma 6.9 development heating up.

This Week in Plasma is out with its latest issue to highlight the newest KDE changes as we close out the month of October.

For those using KWin's zoom effect, with Plasma 6.9 the zoom effect will have a much lower performance impact. The zoom plugin will now skip processing low-contrast pixels like for flat and solid areas, which is leading to significant speed-ups:

Plasma 6.9 is also now clearing stale entries for desktop file and folder positions in its configuration file, to avoid it growing massively over time and slowing down Plasma overall. This stems from a 2023 bug report over the screen mapping entry in the config file growing infinitely enough that it slows down the Plasma shell and could crash it.

With Plasma 6.9, the Discover software center will now allow copying and sharing of appstream://URLs for locally-installed Flatpak packages.

Meanwhile this week's Plasma 6.8.2 release brought some crash fixes and also addressed other regressions. With Plasma 6.8.2 it also included a regression that could degrade gaming performance due in KWin, which is fixed by disabling post-blend offloading if not supported by the DRM color pipelines. Plasma 6.8.3 will bring some additional crash/regression fixes in November.

Also exciting this week on the fixes front is KDE Frameworks now fixing a 15 year old bug over file completion to USB storage devices.

More details on these changes via This Week in Plasma.

---

The Ceph kernel client changes have been submitted for Linux 7.4 ahead of its merge window expected to wrap up on Sunday.

Ceph has supported an ioctl for enabling lazy I/O on a per-file basis while the new code for Linux 7.4 is adding the "lazyio" mount option if wanting to enable lazy I/O globally. Ceph's lazy I/O mode will enable it for all regular files at open time. The mode allows for buffered I/O via the page cache, even if multiple clients have the file open for write. That scenario is intended for HPC workloads and similar that can cope with relaxed cache coherency.

In addition to the "lazyio" mount option for Ceph, the inine data feature that has been around for more than a decade as an experimental feature and deprecated since 2020 is in the process of being removed. For Linux 7.4 the kernel adds a CEPH_FS_INLINE_DATA Kconfig option that defaults to off. So those still relying on Ceph inline data support will now need to ensure that CEPH_FS_INLINE_DATA=Y is set to continue using it. But in a future release that plan to remove the deprecated code outright. Ceoh inline data support was added in Linux 3.19 back in 2014 for allowing small files to be stored directly inside the file's inode or metadata data structures rather than separate RADOS objects.

The Ceph pull request also brings a number of bug fixes and other code clean-ups for Linux 7.4.

---

In recent weeks GNOME developers have been discussing a formal Request For Comments "RFC" process for solving major technical questions and changes to GNOME. This formalized process for succeeding their ad-hoc processes up to now has reached consensus.

GNOME stakeholders have reached a consensus on this RFC process led by Sophie Herold. This will help formalize and ease the discussions moving forward for major project questions/changes. Those wanting to view GNOME's RFC process can read it in full via the GNOME RFCs repo.

This week the GNOME project also saw enhancements to its Nautilus "Files" file manager, GNOME Software is now more efficient with its icon handling and loading of remote assets, and GNOME Calendar also continues to be improved upon. Lower down the stack, GLib now has support for XDG Intents for letting applications register themselves in supporting opening commands in a terminal and handling certain URI schemes.

GNOME OS meanwhile now enables the LDAC codec for their PipeWire builds, Zswap fixes have been merged, the extra-codecs feature is now enabled by default, and there have been NVIDIA improvements to make NVIDIA graphics work with Snap applications and adapting their firmware for the NVIDIA R615 driver.

More details on these latest GNOME and GNOME OS enhancements via This Week in GNOME.

---

DXVK-NVAPI 0.9.3 is out this weekend as a new feature update to this NVIDIA NVAPI implementation used by Valve as part of Steam Play with Proton for nicely integrating NVIDIA technologies into various Windows games running on Linux.

With DXVK-NVAPI 0.9.3 is now support for Opacity Micromaps "OMM" as part of Microsoft's DirectX Raytracing (DXR 1.2+) via NVAPI. With the NVIDIA R615 Linux driver or newer and VKD3D-Proton 3.1, this allows leveraging the Opacity Micromaps feature of DXR for speeding up ray-tracing through alpha-tested and semi-transparent geometry.

Those curious about Opacity Micromaps with DXR can learn more via this Microsoft dev blog.

DXVK-NVAPI 0.9.3 also brings an alternative NVIDIA Reflex implementation for Vulkan games using the VK_NV_low_latency revision 2 extension. This also benefits those on the NVIDIA R615 driver or newer and using this implementation makes the DXVK-NVAPI Vulkan layer obsolete.

The updated DXVK-NVAPI also supports new driver settings found in NVIDIA R610 and R615 drivers, improves reporting of available/free VRAM, and supports compiling for ARM64.

More details on DXVK-NVAPI 0.9.3 via GitHub.

---

NVIDIA software engineers released CUDA Oxide 0.3 for closing out October. CUDA Oxide as a reminder is their experimental Rust to CUDA compiler that has been seeing a lot of activity in recent months.

With CUDA Oxide 0.3, cuda-oxide is now using the official stable Rust version, currently Rust 1.99.1. In prior versions of cuda-oxide, running Rust nightly builds were required. The new release also unifies the host runtime with cutile-rs in the cuda-rust monorepo.

Another big step forward for cuda-oxide 0.3 is shipping an experimental version of "cute-rs" as the CUTLASS CuTe in Rust. With cute-rs it's able to run end-to-end Qwen3-8B at BF16 + NVFP4 along with Qwen3-TTS. Those models with cute-rs are being tested on NVIDIA Blackwell GPUs.

For improving the debugging experience, v0.3 also has full cuda-gdb DWARF debugging support. Plus there is now in-kernel event tracing, register and spill diagnostics, and improved testing.

The new release also brings performance and compiler optimizations for Blackwell and Hopper GPUs.

Downloads and more details for CUDA Oxide 0.3 via GitHub.

---

Sculpt OS as the general purpose operating system project born out of the Genode initiative is now using the project's own microkernel rather than a third-party microkernel.

Sculpt OS 26.10 debuted on Friday and notable is now using the custom Genode microkernel:

"Sculpt 26.10 is the first version based on Genode's custom microkernel instead of relying on a 3rd-party kernel. This kernel - designed from the ground up for the needs of Genode and Sculpt OS in particular - is not only more "micro" than any other of the microkernels supported by Genode, it also solves long-standing deficiencies concerning CPU scheduling and kernel-memory management. Sculpt OS finally stands on Genode's own footing!"

Sculpt OS 26.10 also now includes VirtualBox 7 components for running in VMs rather than VirtualBox 6. Additionally this OS update brings better web camera handling, new support for UK keyboards, and other improvements.

Downloads and more details on Sculpt OS 26.10 via Genode.org.

---

Wine 11.21 is out as the newest bi-weekly development release of Wine and as we near Wine 12.0 stable release happenings getting underway in about one month with the official v12.0 release expected early in the new year.

Most exciting with Wine 11.21 is having merged support for the MESA_map_buffer_client_pointer OpenGL extension to improve WoW64 32-bit OpenGL. This should end up improving the performance for those using OpenGL with a WoW64 setup.

Wine 11.21 also brings an initial version of the Voice Capture DSP library. Rounding out the release is improved support for user accounts.

Wine 11.21 also has 20 known bug fixes, including for Shift + Ctrl + Tab not working, issues with Cinebench 2026, and game fixes like for Skyrim SE and World of Tanks.

The full list of changes with Wine 11.21 can be found at WineHQ.org.

---

The PostgreSQL database server typically sees its annual feature release every September~October, but it's not coming until at least November if not longer this year. Due to the increased developer burden due to increased bug reporting and patches via AI/LLM activity, the PostgreSQL developers have been overwhelmed.

PostgreSQL developer Robert Haas outlined in a blog post today how PostgreSQL 19 is running behind schedule and AI is at fault. The developers have been dealing with a surge of bug reports from AI/LLM agents and also needing to deal with security/bug back-ports to prior PostgreSQL versions.

Haas commented:

"And I think that’s also why we had so many late reverts, and why the release itself is delayed. Two factors are at play here.

First, because so many more bugs are being found now, it takes more time to fix them. LLMs can draft fixes, but those fixes aren’t always correct or compliant with PostgreSQL standards. So, they still need to be reviewed and committed by humans, and those humans have not gotten any smarter between 2025 and 2026.

Second, time that would have been spent reviewing what went into the release got diverted into addressing security issues and other bugs in the back-branches, and so it took longer to figure out what needed to be reverted before shipping v19."

More details on this AI-driven delay to PostgreSQL 19 can be found via this blog post.

---

Following this month's release of Ubuntu 26.10, Ubuntu 27.04 development formally got underway today.

Ubuntu 27.04 is being developed under the "Tenacious T-Rex" codename. The T-Rex repository is now open and the auto-sync from Debian will be enabled soon, which will lead to many new packages being pulled in.

The Ubuntu 27.04 schedule puts the feature freeze for 25 February, the UI freeze for 18 March, and the beta freeze for 29 March. The Ubuntu 27.04 beta is expected to happen on 1 April while the official release of Ubuntu 27.04 is expected on 22 April.

For Ubuntu 27.04 we are expected to see more Rust system/infrastructure transition work with plans for ntpd-rs by default, renewed development of Ubuntu Core Desktop for the immutable flavor, and more AI work on the desktop.

Ubuntu 27.04 Tenacious T-Rex now being open for development was announced today on the ubuntu-devel list.

---

The x86 platform driver changes were merged this week to the Linux 7.4 kernel ahead of the merge window closing this weekend.

As usual there are a lot of new laptop support improvements with Linux 7.4 plus other platform driver work with AMD Ryzen Zen 6 client additions standing out this cycle. Plus we continue to see more laptop driver activity for ARM-based Qualcomm Snapdragon X laptops heading to the kernel through this x86 platform drivers tree.

The AMD Host System Management Port (HSMP) driver with Linux 7.4 is extended to support upcoming AMD Ryzen Zen 6 client platforms. The AMD Platform Management Controller (PMC) also adds support for upcoming Zen 6 platforms. Lastly, the AMD Platform Management Framework (PMF) driver for Ryzen laptops and desktops was extended for a new "AMDI0112" ID that presumably is also for Zen 6 systems.

Linux 7.4 is looking like it will be in nice shape for upcoming Ryzen Zen 6 hardware.

Over on the Intel side, the Intel PMC and Uncore Freq drivers added IDs for the Rugged Panther Lake SoCs.

The Acer WMI driver added hardware monitoring support for the Aspire A315-58G. Meanwhile the Acer Nitro ANV15-51 has not only HWMON b ut also fan control and ACPI platform profile handling too.

Another Acer improvement with Linux 7.4 is introducing the acer-wmi-battery driver for supporting battery/charge controls on Acer laptops.

The Alienware-WMI-WMAX driver now supports the Alienware Area-51 AAT2250.

With the MSI EC driver there is now support for the MSI Creator Z16 A11UE, MSI GL65 Leopard 9SCXK, and MSI Katana 15 B13VFK models.

For HP systems, their WMI driver added support for several more models: OMEN 16-am0xxx, OMEN 16-am2xxx, OMEN 17-ck1xxx, OMEN MAX 16-ah0xxx, OMEN Slim 16t-an000, OMEN Transcend 14-fb1xxx, HP Laptop 15-fd1xxx, Victus 15-fb3xxx, Victus 16-d1xxx (8A26), and Victus 16-e0xxx.

Over on the ASUS side, the ASUS WMI driver added support for mute LED support, the ASUS Armoury driver adds a discrete GPU power state attribute, and there are tunables now enabled for more ASUS devices.

The maturing Uniwill laptop driver with Linux 7.4 adds support for the PCSpecialist Recoil 16 AMD, LAPKC71E Intel NUC x15, and incorporating various fixes.

On the handheld device side, the Ayaneo EC driver added support for the AYANEO 2S. The OXPEC driver for OneXPlayer hardware added support for the OneXPlayer 3.

For Qualcomm Snapdragon X laptops, the x86 platform drivers pull request adds the ASUS Glymur EC support code. There is also the lenovo-yoga-slim7x-gen11-ec as another Snapdragon X laptop improvement.

The full list of x86 platform driver changes merged for Linux 7.4 can be found via this pull request.

---

With today's AI/LLM capabilities, there is almost a new operating system heard about each week "coded from scratch" predominantly via AI agents. This week was no exception with another project hitting our email inbox. Though most of them are open-source, this one is closed-source in the name of security.

NULLos is a "operating system built entirely from scratch" from its own UEFI bootloader to kernel and even its own C compiler that is self-hosting-capable.

While many of these vibe coded operating systems are open-source projects, this one is the unusual route of being closed-source. The NULLos developer argues in an email to Phoronix that it's for a security benefit:

"The source code stays closed, and the system is developed by a single person. I see this as a security decision: one person who cannot be infiltrated, instead of a team from which code can leak, or where vulnerabilities can be found and exploited before they are fixed."

Take it as you wish.

For those curious about this latest creation, NULLos is hosted at nullos.de.

---

For those curious how NVIDIA Vera is scaling between single and dual socket deployments, here are some benchmarks I carried out a while back exploring the power and performance from one to two Vera CPUs.

A Phoronix Premium supporter recently inquired about 1P vs. 2P NVIDIA Vera performance and I happened to have some data collected during the recent NVIDIA Vera benchmarking when also carrying out the NVIDIA Vera SMT performance evaluation and also looking at NVIDIA Vera performance for HPC against AMD EPYC and Intel Xeon.

This article is quite straight-forward and is a standalone look at how the performance and power align between one and two socket Vera configurations. A variety of workloads known to scale to high core/thread counts were used that are also multi-socket NUMA friendly. Due to time constraints and Xeon and EPYC having not exact 88 core product equivalents, these benchmarks are just standalone for NVIDIA Vera 1P vs. 2P.

Each NVIDIA Vera CPU provides 88 Olympus cores with SMT to provide 176 threads, or 172 cores / 352 threads for the full NVIDIA dual Vera configuration.

These NVIDIA Vera benchmarks were done remotely on Ubuntu 24.04 LTS with the Linux 6.17 64K kernel.

---

With the recently released Ubuntu 26.10 and the nearly-released Fedora Workstation 45, Phoronix readers have begun inquiring about fresh benchmarks looking at how they stack up against the performance-optimized, Arch Linux based CachyOS. Up today are our first CachyOS vs. Fedora 45 vs. Ubuntu 26.10 benchmarks on an AMD Ryzen 9 9900X3D2 desktop system.

Both Ubuntu 26.10 and CachyOS are currently on the Linux 7.3 kernel as the latest stable release while Fedora 45 is still on Linux 7.2 but with then rolling out Linux 7.3 as a post-release update. The Mesa 26.2 graphics drivers are present across all of the tested distributions. Both Fedora 45 and CachyOS are using GCC 16.2 by default where as Ubuntu 26.10 stuck to using GCC 15.

The AMD Ryzen 9 9900X3D2 with ASRock X870E Taichi motherboard, 2 x 16GB DDR5-6000 memory, and WD_BLACK SN8100 NVMe SSD were used for testing. Graphics tests were omitted this round of testing because the Radeon RX 9070 GRE in use had to fallback to using LLVMpipe on Ubuntu 26.10. On this particular system, when the AMDGPU driver loaded with the RX 9070 GRE at boot the display would stop working as a regression compared to Ubuntu 26.04 LTS. So for the Ubuntu 26.10 run on this system "nomodeset" was used for getting the display to successfully light up.

Dozens of different CPU/system benchmarks were carried out on this AMD Ryzen 9 desktop for seeing how Ubuntu 26.10 and Fedora Workstation 45 are competing with CachyOS that tends to be on the leading-edge of out-of-the-box performance optimizations.