Distributions and Operating Systems Leftovers

posted by Roy Schestowitz on Apr 30, 2025

• Distro Watch ☛ DistroWatch Weekly, Issue 1119, 28 April 2025

  Welcome to this year's 17th issue of DistroWatch Weekly! This week we get to share a range of fun, exciting, and interesting developments from across the wider open source community. Our news section this week reports on CachyOS shipping with the OCCT stress testing software while the Debian project enters soft freeze, an early step on the road to Debian 13 being released. Debian held its Project Leader election this week and we share the details. We also share a tutorial on getting NetBSD to run on a Nintendo Wii and even run a web server on the gaming hardware. When we cover this week's new releases, later in the newsletter, we share one of the largest distributions to ship - with a 36GB download. We share details on this and other releases below and list the torrents we are seeding. First though we talk about Ubuntu MATE. Jeff Siegel takes Ubuntu MATE 25.04 for a spin and reports on his experiences with this community edition in our Feature Story. Plus we talk about what is missing from the Linux ecosystem. Sometimes it is harder to notice what is not there rather than what is and we explore this in our Questions and Answers column. What missing feature do you think would benefit the Linux community? Let us know in this week's Opinion Poll or share your thoughts in the comments. Finally, we say a fond thank you to the kind readers who sent us donations. We wish you all a terrific week and happy reading!

• Tom's Hardware ☛ ReactOS now supports 3dfx's Voodoo5 GPUs — open-source backdoored Windows alternative offers near-native performance for retro gamers

  The open-source backdoored Windows alternative ReactOS now has support for 3dfx's Voodoo5 GPU family, with Unreal (1998) successfully tested.

• Barry Kauler ☛ Zarfy updated in Scarthgap and Daedalus

  Zarfy multiple monitor manager is somewhat broken in Easy, and Norman (NNI in the forum) sent me some workarounds.

  The source code is a bit old; it is version 0.1.0 with some patches. I have compiled the latest source from the master branch here: [...]

• BSD

  • The Register UK ☛ OpenBSD 7.7 is out, and so is the second 9Front of 2025

    The cover art for this release, entitled Life of a Fish, is by Tomáš Rodr, also known as analouguenowhere. The style struck us as markedly similar to the release artwork for the new version of Plan 9 from Bell Labs fork 9Front, inscrutably dubbed Clause 15 Common Elements of Maus and Star Type. This is credited to prahou, but it's the same artist, as we confirmed from a Fediverse post.

  • Undeadly ☛ In -current, pkg_add -u no longer advises file removal

    Quieter and more accurate updates - what's not to like?

• Slackware Family

  • TuMFatig ☛ Install and isolate apps using Firejail on Slackware

    Since I started using other Desktop OS than OpenBSD, I’m obsessed with limiting filesystem resources access from applications; especially Web browsers. This already applied to FreeBSD Web browsers here and there . This now has to be applied to Linux.

• Fedora Family / IBM

  • Distro Watch ☛ Development Release: AlmaLinux OS 9.6 Beta

    The AlmaLinux project has published a new beta snapshot. The new development release seeks to improve performance and security for the 9.x series. The release announcement states: [...]

  • Red Hat ☛ How reinforcement learning improves DeepSeek performance

    The rapid advancement of large language models (LLMs) has revolutionized how we interact with and utilize artificial intelligence. From generating creative content to answering complex questions, LLMs have demonstrated remarkable capabilities. However, training LLMs requires vast computational resources and massive datasets, posing significant financial and logistical challenges. Furthermore, aligning these models with ethical values and confirming that their outputs are safe and unbiased remains a complex and ongoing problem.

  • Red Hat ☛ Accelerate model training on OpenShift Hey Hi (AI) with NVIDIA GPUDirect RDMA

    “Attention is all you need.” But you also need efficient GPU peer-to-peer communication! 

    Since that seminal paper was published in 2017, scaling the size of large language models (LLMs) has been a key driver to sustain the pace of innovation and push the frontier of LLM performance while informing the so-called scaling laws of AI.

    While techniques like quantization or compression can be used to lower the amount of memory required to load those large models into a single GPU, it comes with some tradeoffs, and only shifts the threshold beyond which scaling out becomes the only viable option. In the case of model training, scaling-out can also be used to leverage more compute resources and achieve better latency.

• Canonical/Ubuntu Family

  • Kev Quirk ☛ 📝 29 April 2025 at 16:07 - So I've been running #Ubuntu for the last couple days and I have to say, it's been great.

    So I've been running #Ubuntu for the last couple days and I have to say, it's been great. Very stable and no real issues to talk of. Snaps are still a bit shit, but a couple of commands later and I have Flatpak support.