It should come as little surprise in general but making use of Advanced Vector Extensions (AVX) for multimedia encode/decode is a big deal for performance. While one could hope that by 2020 most open-source encoders were already extensively utilizing AVX(2), that isn't yet the case with the latest being AOM-AV1 picking up another optimization.

Google engineers have added an AVX2-optimized high bit-depth temporal filter for the AOM-AV1 video encoder.

That optimized

It's been widely expected AMD will launch their next-generation RDNA 2 graphics cards and Zen 3 processors in Q4 as they previously reported as well as leaks pointing to October reveal dates. Today the company is sharing actual dates for said announcements.

Via Twitter , AMD shared that "a new journey begins" for Zen 3 on October 8 while for RDNA 2 the date listed is October 28.

Systemd developers are still hoping to introduce systemd-oomd as part of the next release.

Systemd-oomd is the effort to provide better Linux out-of-memory / low-memory handling . Systemd-oomd is being spun from Facebook's out-of-memory daemon and adapted to not only work on Linux servers but desktops as well.

The systemd-oomd daemon polls for OOMD-enabled cgroups to monitor and will kill based on memory pressure or swap usage. The systemd-oomd behavior is configurable via

Along with Intel's SVT-AV1 0.8.5 , another open-source AV1 video encoder seeing a new release this week is the Rust-written rav1e.

Rav1e 0.3.4 is this new release and notably adds support for running on Google's Android and Apple iOS platforms in addition to its existing support on Windows, Linux, and elsewhere.

The Rav1e 0.3.4 release also fixes desync in some 4K content, desync in

Wired up in Mesa 20.3-devel is a new DriConf option override_vram_size for overriding a smaller amount of video memory to report to the program/game being run. This is intended for development/debug purposes.

RADV co-founder Bas Nieuwenhuizen added the "override_vram_size" DriConf tunable for the RADV Vulkan driver as well as Gallium3D drivers.

The intention of this new option is to help in debugging of games that do active memory budget management. The override_

Chinese company Zhaoxin that continues working on x86_64 CPUs based on VIA Centaur Technology is working on supporting their "7" family processors with the Linux kernel.

Since earlier this year we started seeing some Zhaoxin 7-Series patches while a new set was sent out this week.

These patches for the "7" family is for the ZX-F / KX-7000 hardware. This family of processors will reportedly launch formally in 2021 and be manufactured on a 7nm

Proposed last year were a set of patches aiming to improve the Intel virtual GPU "vGPU" performance in para-virtualized cases by having optimizations around the shared memory region between the guest and Intel GVTg code. With optimizing the workload PV submission and PPGTT PV updates, the glxgears performance could improve by 30~50% while for large media/3D workloads was more around a 4% average improvement.

Xiaolin Zhang of Intel has sent out an entirely new patch series now derived from the work done last year

With GNOME 3.38 due to be released next week, the GNOME Shell and Mutter development blog has put out their overview of all the improvements and new features to expect.

- Compositor bypass support for full-screen applications to avoid bypassing of the desktop compositing work when only displaying a full-screen game, media player, or other full-screen software. For GNOME 3.38, this only works with Mutter on Wayland.

- Split frame clock support for improving the multi-monitor experience where the