Lisa Su at AMD Advancing AI 2026 just launched the new Instinct MI455X GPUs and the Helios rack.

Instinct MI455X is AMD's new very potent accelerator offering with 432GB of HBM4 memory and manufactured on a 2nm process. The peak MXFP8 and MXFP4 come in at up to 4x the Instinct MI355X for a very loft generational upgrade. With 50% more HBM capacity is support for larger AI models.

The FreeBSD project issued a statement today over the FreeBSD Ports repository having been frozen for the past two days. The freeze is not due to a malicious compromise but rather a very large commit breaking things.

Since 21 July the FreeBSD Ports repository has been frozen and as of writing remains that way. The status of the FreeBSD Ports freeze is being communicated via this web page . A mailing list post elaborates on the freeze, which comes down to a 150MB file having been committed and that breaking mirroring to

One of the latest change proposals filed for Fedora Linux 45 is for supporting web-based remote installations of Fedora Atomic Desktop images.

The change still needs to be approved by the Fedora Engineering and Steering Committee (FESCo), but the plan is beginning with Fedora 45 to allow for the Anaconda WebUI to be supported for remote installations of Fedora Atomic Desktop images. The Fedora Anaconda WebUI itself has been in the works for a while and is now about it exposing it to the Fedora Atomic Desktop world as a new

Ahead of Lisa Su's keynote this morning at the AMD Advancing AI event, tagged on GitHub just now was the Ryzen AI Software 1.8 release for helping to deploy AI models on Ryzen AI PCs.

Going along with the recent Lemonade local AI server updates and other AMD open-source AI software releases, today brings the release of Ryzen AI Software 1.8. This Ryzen AI "RAI" software release primarily is about supporting newer models and shipping more optimizations. A variety of new large language models are

Following the recent code branching for the LLVM 23 series, LLVM 23.1-rc1 was released today in working toward the next stable release of this open-source compiler stack.

LLVM 23.1.0 stable is working toward a stable release by the end of August and with that comes many changes for this half-year update to the LLVM/Clang compiler. LLVM 23 brings x86 Lightweight Fault Isolation "LFI" target support, initial support for the NVIDIA Rigel core to

While the VideoLAN developers are busy these days working on dav2d for AV2 video decoding, they haven't let up work on dav1d and recently released dav1d 1.5.4 for continuing to enhance this leading open-source, CPU-based AV1 decoder.

With dav1d 1.5.4 they have introduced support foir OS/2. Dav1d to the OS/2 operating system ended up being a fairly easy port with just Meson build system and header

Just one week after releasing Lemonade 11.0 , the Lemonade 11.5 local AI server was released today for this open-source AMD backed project during their AMD Advancing AI event.

The Lemonade local AI server allows easily deploying LLM workloads across Ryzen AI NPUs, Radeon GPUs, and CPUs with ease under both Windows and Linux. The most significant change with Lemonade 11.5 is the completion of the Lemonade Router that can be used for automatically routing queries to relevant models based on defined policies --

Sam Lantinga of Valve and SDL fame has landed HDR10 rendering support into the Simple DirectMedia Library.

When using SDL's Direct3D 11, Direct3D 12, or Vulkan renderers, HDR10 display support is now in place for SDL3 Git. SDL 3.0 previously introduced colorspace support and other HDR preparations like Wayland color management . It's great seeing HDR support continue to expand throughout the Linux and open-source ecosystem.