Vue lecture

Benchmarks: OpenCL Kernel Latency ~76x Lower For Intel Lunar Lake With Updated Compute Runtime

This week Intel released the Compute Runtime 25.18.33578.6 release for Windows and Linux. This updated open-source GPU compute stack for OpenCL and Level Zero brings the latest work on Ultra Low Latency Scheduling (ULLS) for Xe2 Lunar Lake graphics and other ongoing Xe2 improvements along with further preparations for next-gen Xe3 hardware. This new Intel Compute Runtime release is clocking in around 76x lower OpenCL kernel latency and other nice wins for those with current-generation Intel Lunar Lake hardware.
  •  

Mesa Will Stop Building Gallium-XA By Default

Back in April Mesa deprecated Gallium Nine alongside the XA state tracker used for X.Org Server acceleration. Gallium-XA was developed by VMware for use with their DDX driver for accelerating X11 within their virtualized environments. XA isn't really used these days and the generic GLAMOR code is in much better shape. Thus Mesa is now no longer building the XA code as the next step toward its removal...
  •  

Maximizing The Performance & Power Efficiency Of AMD Ryzen AI Max+ 395 With Platform Profiles

Last week I began posting Linux benchmarks of the flagship "Strix Halo" SoC, the AMD Ryzen AI Max+ PRO 395 with powerful Radeon 8060S graphics using the HP ZBook Ultra G1a laptop. Both the CPU and GPU results for the data published thus far have been mighty impressive and done at the Ubuntu Linux defaults... But the Ryzen AI Max+ 395 within the ZBook Ultra G1a can be pushed even further by leveraging the ACPI Platform Profiles capabilities. In this article is showing how much further performance -- and the power costs -- can be achieved out of this AMD Strix Halo SoC when leveraging the HP Power Profiles as well as the power efficiency for those wanting to run the SoC instead in the "power saver" / "low power" configuration.
  •  

AMD Releases ROCm 6.4.1 With RDNA4 GPU Support

Well, this is a pleasant surprise... Less than 24 hours after the AMD Computex 2025 keynote where better/more-timely ROCm support for client GPUs was brought up, AMD ROCm 6.4.1 is now officially released. With ROCm 6.4.1 there is formal support for RDNA4 GPUs, including the Radeon RX 9000 series consumer graphics cards...
  •  

libinput Preparing To Introduce A Lua-Based Plugin System For Modifying Devices/Events

The libinput input handling library that's used by both X11 and Wayland based environments on the Linux desktop is preparing to introduce a Lua-based plug-in system. Via Lua scripts it will be possible to modify evdev input events / input device behavior to deal with quirky/broken input devices and better workaround other problems that aren't currently easily addressable...
  •