Rust 1.93 is out today as the first feature release for this programming lanugage of 2026.

Yesterday along with releasing ROCm 7.2 there was also the release of AOMP 22.0-2 as the newest version of their open-source downstream of LLVM/Clang/Flang that is focused on offering the best OpenMP/OpenACC offloading support to Instinct/Radeon hardware.

James Brodman worked for the last 15 years at Intel on their ISPC SIMD compiler and then in more recent years on the Intel DPC++ compiler and SYCL support as part of Intel's oneAPI initiative. Rather interestingly, this compiler expert has now joined AMD.

The ReactOS project is celebrating today that it marks 30 years since their first code commit in the ReactOS source tree. During the past 30 years now the project has seen more than 88k commits from more than 300 developers as it seeks to be a robust open-source Windows implementation. In their 30 year birthday blog post they also provide a look ahead at what they're working on.

While the Linux kernel has been seeing preparations from NVIDIA for 1.6 Tb/s networking in preparing for next-generation super-computing, the kernel has still retained support to now for the High Performance Parallel Interface. HIPPI was the standard for connecting supercomputers in the late 1980s and a portion of the 1990s with being the first networking standard for near-Gigabit connectivity at 800 Mb/s over distances up to 25 meters. But HIPPI looks like it will be retired from the mainline kernel with Linux 7.0.

Sent out to the Linux kernel mailing list this afternoon were a set of 19 patches in preparing for some new CPU features presumably to be found with AMD's next-generation EPYC "Venice" processors.

Back at CES earlier this month AMD talked up features of the ROCm 7.2 release. ROCm 7.2 though wasn't actually released then, at least not for Linux. That ROCm 7.2.0 release though was pushed out today as the latest improvement to this open-source AMD GPU compute stack and officially extending the support to more Radeon graphics cards.

Eric Engestrom just released Mesa 26.0-rc1 with the code for this quarter's Mesa feature release now branched and under a feature freeze leading up to the stable release in February.

PyTorch 2.10 is out today as the latest feature update to this widely-used deep learning library. The new PyTorch release continues improving support for Intel GPUs as well as for the AMD ROCm compute stack along with still driving more enhancements for NVIDIA CUDA.

XDG-Desktop-Portal 1.21 is now available for testing with the latest features for this portal frontend service to Flatpak.

With the Dell Pro Max GB10 testing at Phoronix we have been focused on the AI performance with its Blackwell GPU as the GB10 superchip was designed for meeting the needs of AI. Many Phoronix readers have also been curious about the GB10's CPU performance in more traditional Linux workloads. So for those curious about the GB10 CPU performance, here are some Linux benchmarks focused today on the CPU performance and going up against the AMD Ryzen AI Max+ 395 "Strix Halo" within the Framework Desktop.

Ahead of the Linux 6.20~7.0 cycle kicking off next month, the Apple Silicon Device Tree updates have been sent out for queuing ahead of that next merge window. Notable this round are the Device Tree additions for rounding out the USB 2.0/3.x support with the USB-C ports.

A new patch posted to the Linux kernel mailing list aims to address the high wake-up latency experienced on modern Intel Xeon server platforms. With Sapphire Rapids and newer, "excessive" wakeup latencies with the Linux menu governor and NOHZ_FULL configuration can negatively impair Xeon CPUs for latency-sensitive workloads but a 16 line patch aims to better improve the situation. That is, changing one line of actual code and the rest being code comments.

Prominent Intel Linux engineer H. Peter Anvin has posted a new patch series working to clean-up the Linux x86/x86_64 kernel boot code. Besides cleaning up the code, the kernel configuration would drop options around EFI stub mode and relocatable kernels in making those features now always enabled.

For those using the powerful PHPStan tool for static analysis on PHP code, this week's PHPStan 2.1.34 is promoting optimized performance with projects seeing around 25% to 40% faster analysis times.

Mesa 26.0 was due to be branched last week and in turn start its feature freeze but ended up being pushed back to tomorrow (21 January) to allow some lingering features to land. It's been beneficial for the Radeon Vulkan driver "RADV" with several interesting merge requests having landed in time for Mesa 26.0.

Intel Linux engineers have been working on enhancing the NVMe storage performance with today's high core count processors. Due to situations where multiple CPUs could end up sharing the same NVMe IRQ(s), performance penalties can arise if the IRQ affinity and the CPU's cluster do not align. There is a pending patch to address this situation. A 15% performance improvement was reported with the pending patch.

It's typically rare these days for the ATA subsystem updates in the Linux kernel to contain anything really noteworthy. But today some important fixes were merged for the ATA code to deal with a reported power management regression affecting the past number of Linux kernel releases over the last year. ATAPI devices with dummy ports weren't hitting their low-power state and in turn preventing the CPU from reaching low-power C-states but thankfully that is now resolved with this code.

Following the December launch of Pop!_OS 24.04 LTS and the first major COSMIC desktop release, System76 software engineers have continued making improvements to their Rust-based desktop environment.

Deploying vLLM for LLM inference and serving on NVIDIA hardware can be as easy as pip3 install vllm. Beautifully simple just as many of the AI/LLM Python libraries can deploy straight-away and typically "just work" on NVIDIA. Running vLLM atop AMD Radeon/Instinct hardware though has traditionally meant either compiling vLLM from source yourself or AMD's recommended approach of using Docker containers that contain pre-built versions of vLLM. Finally there is now a blessed Python wheel for making it easier to install vLLM without Docker and leveraging ROCm.