Along with getting the NVIDIA GSP fully booted and initialized for Ampere GPUs with the Nova driver code coming to Linux 6.19, this next kernel version is also beginning to make preparations for eyeing next-generation NVIDIA GPU support on this open-source driver.

Rusticl as a modern OpenCL implementation for Mesa Gallium3D drivers has turned out remarkably well. Rusticl performance has evolved quite well for this Rust-based OpenCL driver and it continues tacking on new features / OpenCL extensions as well as working gracefully with more Mesa drivers. Rusticl lead developer Karol Herbst presented on some of the recent accomplishments for this driver back at XDC2025.

The latest feature enablement work happening by Intel for the Linux kernel with next-generation Diamond Rapids server processors are the adjustments to the Kernel-based Virtual Machine (KVM) for readying the new CPU ISA capabilities for a virtualized world.

A patch series sent out today on the Linux kernel mailing list proposes "MEMINSPECT" as a new mechanism for helping with memory analysis.

Initially upstreamed into the Linux 6.18 kernel is Tyr as a Rust-based GPU kernel driver for Arm Mali hardware. This is in effect a Rust alternative to the Panthor DRM kernel driver for newer Arm Mali GPUs with the Command Stream Firmware (CSF). With the latest development code for Tyr, it's moved onto running the GNOME desktop and basic games like SuperTuxKart.

Xen 4.21 is out today as the newest feature release for this open-source hypervisor backed by AMD, Arm, AWS, and other organizations. Plus with Xen's use within automotive environments, Ford and Honda too.

This week marks two years since the debut of the Ryzen Threadripper 7000 series processors. Given the occasion, I decided to revisit the Linux performance of the Threadripper 7980X compared to original benchmarks from November 2023 to see how the latest Linux software stack performs for these Zen 4 HEDT processors.

Beginning yesterday and continuing today are several patch series beginning to lay the foundation in the AMDGPU kernel graphics driver for enabling some next-generation graphics IP. Due to the AMD graphics driver block by block enablement strategy and IP-based discovery adopted by their driver over the past few years, it's not clear what this new hardware enablement is for whether it's RDNA5 / UDNA or some RDNA4 refresh. In any event, the Linux driver enablement has begun.

Intel engineers today posted Linux kernel patches for plumbing a brand new Error Detection And Correction "EDAC" driver for the next-generation memory controller design debuting with Xeon Diamond Rapids.

Back in September the Qualcomm X2 Elite SoCs were announced for next-gen Windows 11 on Arm laptops. Since then some initial X2 Elite enablement patches for the Linux kernel have arrived and for the upcoming Linux 6.19 kernel more of that work will reach mainline. Excitingly, Linux 6.19 is now bringing GPU and display support for the Adreno X2-85 found within the Snapdragon X2 Elite SoC.

Microsoft announced the Cobalt 200 processor as their next-generation cloud-native CPU for the Azure Cloud. The Cobalt 200 will feature 132 Arm Neoverse-V3 based cores.

When it comes to GPU virtualization we have seen AMD engineers carry out a lot of work in recent years around the Xen hypervisor even when it hasn't seen as much interest from other vendors. We found out that much of their interest in Xen for GPU virtualization is due to automotive / in-vehicle infotainment demands and it remains that way. They continue cooking some new features and they say "the best is yet to come" in a new presentation on their Xen virtualization efforts.

Alice Ryhl of Google sent out the main set of Rust language code changes for the Direct Rendering Manager (DRM) graphics/display driver subsystem ahead of Linux 6.19. Notable is continued DRM core infrastructure work for Rust plus the open-source NVIDIA "Nova" driver continues taking shape albeit isn't yet ready for end-user usage.

Intel's ISA documentation was updated last week to confirm Nova Lake processors will support AVX10.2 and APX extensions after they were not officially acknowledged in prior versions of the spec and the initial open-source compiler enablement with -march=novalake also left them without those prominent ISA capabilities. Following that documentation update, a few days ago LLVM Clang updated their Nova Lake compiler support for the new ISA capabilities and now the GCC compiler has received similar treatment.

Sent out today was likely the last batch of HID subsystem fixes ahead of the Linux 6.18 kernel releasing as stable around the end of the month. With it are some new device-specific quirks for fixing hardware support for a mouse and keyboard.

Last week's Vulkan 1.4.333 brought a new ray-tracing extension with VK_EXT_ray_tracing_invocation_reorder that was derived from a prior NVIDIA vendor extension (VK_NV_ray_tracing_invocation_reorder). This new extension for Shader Execution Reordering "SER" is showing to deliver some nice performance potential for Vulkan ray-tracing performance.

For the past several months a Uniwill laptop driver for the Linux kernel has been in development to expose extra platform capabilities for laptops manufactured by this Taiwanese OEM/ODM manufacturer. Assuming no last minute issues, this driver is now set to premiere in Linux 6.19 for helping Uniwill laptops and hardware from other brands relying on Uniwill as the device manufacturer.

Intel's LASS functionality was queued today into tip/tip.git's "x86/cpu" Git branch. With LASS now making it into a TIP branch, it looks like it will be submitted for the upcoming Linux 6.19 merge window barring any last minute issues or objections from Linus Torvalds.

It's the Blender 5.0 release day! Blender 5.0 is a big step forward for this open-source 3D modeling software with better Vulkan viewport support across different GPUs/drivers, HDR support when using Vulkan and Wayland on Linux, and other very nice refinements for this popular cross-platform software package.

In the past we have seen Llama.cpp with Vulkan outperforming AMD's ROCm compute stack in some of the large language model (LLM) AI benchmarks. Curious if anything has changed given the recent ROCm 7.1 release, I ran some benchmarks of an up-to-date Llama.cpp using the AMD ROCm back-end compared to the Vulkan back-end with the latest RADV driver. For this round of testing the Radeon AI PRO R9700 graphics card was used.