↩ Accueil

Vue lecture

Mesa's "Present Timing" Vulkan Driver Support Now Feature Complete

Last month the Vulkan VK_EXT_present_timing extension was merged after 5+ years in development. VK_EXT_present_timing ended up debuting at the end of November within the Vulkan 1.4.335 spec update to much excitement for providing functionality to obtain information on the presentation engine's display for accurate timing information and to assist in scheduling a present to happen no earlier than a desired time. This is a big win for helping avoid game stuttering and more while now the Mesa support for it is nearly complete and could be merged soon...
  •  

Intel llm-scaler-vllm Beta 1.2 Brings Support For New AI Models On Arc Graphics

Following yesterday's release of a new llm-scaler-omni beta there is now a new beta feature release of llm-scaler-vllm that provides the Intel-optimized version of vLLM within a Docker container that is set and ready to go for AI on modern Arc Graphics hardware. With today's llm-scaler-vllm 1.2 beta release there is support for a variety of additional large language models (LLMs) and other improvements...
  •  

Linux 6.19 Networking Delivers 4x Improvement For Heavy Transfer Workloads, New Hardware

The big set of networking subsystem updates was recently merged for the ongoing Linux 6.19 merge window. There are some enticing core networking improvements like a big performance improvement for heavy transfer workloads, Bluetooth PAST enablement, and more. Plus a lot of wired and wireless networking driver activity and new hardware enablement...
  •  

Intel's Vulkan Linux Driver Merges Shader VMA Allocator For Ray-Tracing Capture/Replay

Merged today to the Intel open-source "ANV" Vulkan driver in Mesa 26.0 is introducing a shader VMA allocator. Long story short this new allocator steps toward enabling Vulkan ray-tracing capture/replay support, which can come in hand for debugging issues with Vulkan ray-tracing on Intel graphics hardware under Linux and similarly to assist in optimizing for better performance...
  •  

Qt Toolkit Lands IO_uring Abstraction

The newest feature to land in the cross-platform Qt toolkit is QIORing as an abstraction for Linux's IO_uring interface. This QIORing may also end up supporting Microsoft's Windows IORing implementation as well...
  •  

FreeBSD 15.0 vs. Ubuntu Linux For AMD EPYC Server Performance

Given the recent release of FreeBSD 15, I started off my testing in looking at how FreeBSD 15.0 improves performance versus FreeBSD 14.3. Now it's onto the next important question: how is FreeBSD 15.0 performing relative to Linux on servers? Here are some benchmarks exploring that topic today.
  •  

Updated Intel LLM-Scaler-Omni Improves ComfyUI Performance For Arc Graphics

The past several months Intel software engineers have been quite busy with LLM-Scaler as part of Project Battlematrix. LLM-Scaler is a Docker-based solution for AI workloads on Intel graphics hardware to ship an optimized vLLM stack and other AI frameworks. Out today is a new LLM-Scaler-Omni release to help enhance ComfyUI performance on Intel hardware...
  •  

Linux 6.19 Gets Rid Of The Kernel's "Genocide" Function

While the Linux kernel has inclusive terminology guidelines for the past five years to replace phrases like master/slave and blacklist/whitelist, there has surprisingly been a "genocide" function within the kernel that was questioned when it was first submitted for inclusion but now removed in Linux 6.19...
  •  

Canonical To Distribute AMD ROCm Libraries With Ubuntu 26.04 LTS

AMD previously talked of simplifying the in-box Linux support for ROCm during the second half of 2025. So far we haven't seen any groundbreaking changes from that initiative besides AMD working on various package archives/repositories to make it easier to install the latest ROCm on different Linux distributions. But today a big announcement is now public that Canonical with next year's Ubuntu 26.04 LTS release will provide official ROCm packages along with other libraries...
  •  

Scheduler Woes: Bisecting Early Performance Regressions Found In Linux 6.19

Yesterday I noted some early performance regressions I've found on the Linux 6.19 kernel compared to Linux 6.18 LTS stable. Those initial benchmarks were on an AMD EPYC server. Since then I've seen many of the same workloads regressing similarly on an AMD Ryzen Threadripper workstation between Linux 6.18 and Linux 6.19 Git. Given the significant impact and AMD Threadripper processors always helping out to speed-up Linux kernel build times to make for a quicker and more manageable kernel bisecting experience, here is a look at some of the results for the Linux 6.19 performance regressions.
  •