Open Source LLM|Intel

Intel Releases LLM-Scaler 1.0 As Part Of Project Battlematrix

Intel today announced their August 2025 Software Update to Project Battlematrix and the release of the LLM-Scaler 1.0 container for optimized AI inference support on Intel Arc B-Series graphics hardware.

Back in May Intel announced Project Battlematrix for supporting up to eight Intel Arc Pro GPUs for AI inference, SR-IOV support, improved vLLM performance, and a variety of other new features for their graphics stack on Linux. Their goal at the time was to have Q3 availability and full feature enablement by the
Continue reading...

Early Linux 6.17 Tests Show Some AMD Strix Halo Performance Improvements & Regressions

Even prior to the Linux 6.17-rc1 release on Sunday I already had kicked off some Linux 6.17 Git benchmarking in being eager to see how the performance is beginning to shape up for this next kernel release that is set to power the likes of Ubuntu 25.10 and Fedora 43. There is some good news and bad news with my early testing on the ZBook Ultra G1a for AMD Strix Halo .

This weekend I started running some initial benchmarks to see how
Continue reading...

Open Source LLM
更新版包容性語言指南批評「Sanity Check」、「Hung」與「Native Support」的用法

Updated Inclusive Language Guide Calls Out "Sanity Check", "Hung", "Native Support"

The Linux Foundation's Alliance for OpenUSD "AOUSD" and the Academy Software Foundation "ASWF" announced today an updated Inclusive Language Guide.

These projects under the Linux Foundation announced a new Inclusive Language Guide as well as welcoming new members to the AOUSD such as Coca-Cola, Renault, and Accenture.

Continue reading...

Intel Posts Latest Patches For DRM Sharpness Property To Enjoy With Lunar Lake & Beyond

For over a year now Intel has been working on a new DRM sharpness property for making use of Lunar Lake's new adaptive sharpening filter capabilities built into its display engine. This new sharpening filter with Lunar Lake and future SoCs can hep with sharpening blurred or upscaled content and over the past year has gone through several rounds of code review . The latest patches were sent out last week for this DRM sharpness property.

Sadly too late for the Linux 6.17 kernel cycle is this DRM sharpness property but Intel engineers
Continue reading...

LunarG Announces KosmicKrisp As Vulkan-On-Metal Mesa Driver

While there is already MoltenVK for Vulkan implemented over Apple's Metal graphics API, the graphics engineers at LunarG have announced KosmicKrisp as a Mesa-based driver implementing Vulkan over Metal.

LunarG announced KosmicKrisp as a Vulkan driver for Apple hardware built on the Mesa graphics driver framework. This driver is currently close to Vulkan 1.3 conformance. Development on KosmicKrisp appears to be driven in part by Google with wanting to use this Vulkan-on-Metal solution for accelerating the Android emulator on macOS for enhancing the Android developer experience
Continue reading...

AMD EPYC 4545P Achieves 2.24x The Performance At Half The Power Of The First EPYC CPU

Recently we looked at the performance of the AMD EPYC 4545P that is a 16 core 65 Watt processor in the EPYC 4005 "Grado" series. This is quite an interesting processor for those after low-power servers, edge AI deployments, and other purposes with no similar Ryzen 9000 series processor or competition from Intel offering sixteen performance cores at around 65 Watts. Complementing all the performance and power data from that review article, here are some additional tests putting

Continue reading...

Linux Cache-Aware Scheduling / Load Balancing Updated With New Tuning Knob

Intel engineer Chen Yu posted a fresh round of Linux kernel patches working on cache-aware scheduling/load-balancing for this functionality being sought after both by Intel and AMD. The new patches should address some performance regressions observed in the prior patches.

For the past number of months there have been patches floated on the Linux kernel mailing list for cache-aware load balancing that have the potential to help with performance on modern AMD and Intel processors, especially for larger server processors. The focus with the cache aware scheduling is
Continue reading...

NFS Client & Server Feature Updates For Linux 6.17

For those making use of the Network File System (NFS), the in-development Linux 6.17 kernel is bringing some nice improvements both for the NFS server and client code.

During the now-over Linux 6.17 merge window some nice NFS enhancements made it into this next version of the Linux kernel. On the NFSD side there is now the ability to offer write delegations to clients opening files with O_WRONLY . According to the pull request it should open up accelerating some interesting corner cases

Intel

Linux ACPI Idle Driver Being Improved For Hybrid CPUs

For hybrid CPU core designs from recent Intel Core (Ultra) processors to ARM big.LIITTLE, a patch series was posted today in seeking to enhance the generic ACPI processor idle driver around processors with multiple types of CPU cores.

The "acpi_idle" Linux driver manages the CPU power states based on the system's ACPI tables. With the patches posted today the ACPI idle driver would be able to take into account different CPU cores on a system. While Intel and ARM processors are mentioned in the patch cover

Blender 5.0 Will Likely Default To Using OpenGL Rather Than Vulkan

While there was previously talk of Blender 5.0 likely defaulting to using the Vulkan API for rendering but keeping the OpenGL driver around, those plans look like they may be changing. OpenGL-by-default looks to now be on the table for Blender 5.0 due out later this year.

Blender 4.5 recently shipped with good Vulkan rendering support and the hope was that for Blender 5.0 that Vulkan support could be enabled by default on capable systems. But testing is showing some users are running into
Continue reading...