Vue lecture

SMT Proves Very Advantageous For AMD Ryzen AI MAX Strix Halo Performance

While Intel opted against implementing Hyper Threading for their latest Arrow Lake and Lunar Lake processors, Simultaneous Multi-Threading (SMT) still proves very effective on the AMD side. Even though the top-end AMD Ryzen AI MAX "Strix Halo" SoCs provide 16 Zen 5 cores, the presence of SMT for 32 threads still proves worthwhile from both a performance and power efficiency perspective. Here is an on/off comparison for SMT with the flagship AMD Ryzen AI MAX+ PRO 395 within the HP ZBook Ultra G1a.
  •  

AMD Upstreams Efficient Malloc Support On GPUs For LLVM libc

AMD compiler engineer Joseph Huber is the one who ported DOOM to run on GPUs atop ROCm + LLVM libc as part of taking standard C/C++ code to run on GPUs and more recently has also been pursuing Flang/Fortran support atop GPUs. The latest in this ongoing quest is implementing efficient malloc support for memory allocation support on GPUs via the LLVM libc library...
  •  

ByoWave Modular Proteus Controller Kit Support Lands In Linux 6.16

Valve engineer Pierre-Loup A. Griffais contributed ByoWave Proteus controller support to the Linux kernel. ByoWave Proteus are some modular controller designs that can be adapted based upon the needs of the gamer and even what title you may be playing at the moment. The triggers and buttons are all repositionable for a very customized controller experience. The only downside is the modular gaming controller kit retails for $299 USD, but at least now works off the mainline Linux kernel...
  •  

Kexec HandOver "KHO" Merged For Linux 6.16

Kexec HandOver "KHO" was merged for the in-development Linux 6.16 kernel as part of all the memory management "MM" changes. Kexec HandOver is providing the basis for some nifty low-level features moving forward...
  •  

Open-Source TPDE Can Compile Code 10-20x Faster Than LLVM

Researchers from the Technical University of Munich (TUM) have announced TPDE as a fast and adaptable compiler back-end framework. The code is now open-source and they are talking up some very wild compile time improvements... Compiling code for x86_64 and AArch64 with TPDE can be ten to twenty times faster than using the LLVM Clang compiler...
  •  

Power & Performance Tuning For The Framework 13 With AMD Ryzen AI 300 Series

In April we looked at the Framework 13 updated for the AMD Ryzen AI 300 series and its great out-of-the-box performance and nice power efficiency of Strix Point. Via ACPI platform profiles the power/performance mode can be tuned if desiring a longer battery life or preferring even greater performance. Today's article is looking at that power/performance impact of the different ACPI platform profiles available on the Framework 13 motherboard with the Ryzen AI 9 HX 370.
  •  

Mesa 25.2 Lands RADV VCN5 Video Encode/Decode Support For RDNA4 GPUs

Posted for the AMDGPU Linux kernel driver in early 2024 was the initial Video Core Next 5.0 IP enablement that was part of their bring-up for the Radeon RX 9000 "RDNA4" graphics processors. It's taken until now though for enabling the VCN5 video encode/decode support within the user-space code for the RADV driver with Mesa 25.2...
  •  

Intel TDX Host Support Merged For KVM With Linux 6.16

While Intel Trust Domain Extensions (TDX) has been around since Sapphire Rapids with select SKUs and with widespread availability since Emerald Rapids in late 2023, only now with the Linux 6.16 kernel debuting in H2'2025 is there going to be mainline kernel support for TDX on the host-side with the Kernel-based Virtual Machine (KVM)...
  •  

Intel Overclocking Watchdog Driver Merged For Linux 6.16

Merged today for the Linux 6.16 kernel were all of the Watchdog subsystem updates for monitoring system health and taking action such as rebooting if the system state goes bad. With the Linux 6.16 is the introduction of the Intel Overclocking Watchdog "OC WDT" driver...
  •