The first release candidate of LLVM 17 is now available for testing as what will be the next half-year update to this innovative open-source compiler stack...
Sent out today was a batch of "new stuff" for the AMDGPU and AMDKFD kernel graphics drivers for queuing in DRM-Next ahead of the Linux 6.6 merge window opening in about one month...
The new AMD EPYC Bergamo and Genoa-X processors have been very fascinating in the lab from the performance angle and the many different features and knobs provided by these new high-end server processors focused on dense cloud and energy-efficient deployments and HPC/AI, respectively. With Bergamo the flagship AMD EPYC 9754 provides 128 cores with SMT and the Zen 4C cores still boast AVX-512. Another nifty aspect on this high core count CPU catering to cloud service providers is the adjustable TDP from 320 to Watts. Prior Phoronix benchmarks have looked at the default 360 Watt performance and the 400W at the high-end with power determinism mode while today's article is looking at the efficiency gains made possible by pulling back to a 320W cTDP.
It appears the OpenZFS 2.2 file-system driver for Linux and FreeBSD systems will see its release very soon while out today is the third release candidate...
Announced in early June by Intel-owned Codeplay Software was the oneAPI Construction Kit for helping to bring SYCL codebases to new processor/accelerator architectures with an emphasis on AI and HPC. Today marks the release already of the oneAPI Construction Kit 3.0...
Richard Hughes is the Red Hat developer who is most prominently known for leading the Linux Vendor Firmware Service (LVFS) and Fwupd development as well as formerly being behind the ColorHug monitor color calibration hardware effort and PackageKit, among other open-source software. He's recently been developing a new software project called Passim that today he announced to the world...
Going along with LLVM's recent additions around supporting new Intel instructions coming with future generation Core CPUs, the LLVM 18 Git development code has now landed support for actually honoring -march=arrowlake, -march=arrowlake-s, and -march=lunarlake targets...
AMD lifted the embargo this evening on the Ryzen 9 7945HX3D, their first mobile processor sporting 3D V-Cache technology for boosting gaming performance and other cache-happy workloads...
This afternoon AMD announced the availability of the HIP SDK for Microsoft Windows as a portion of their ROCm computing platform with support for various professional and consumer GPUs...
For over a year Fedora / Red Hat has been planning for major package management changes with DNF5. The hope for months has been to use DNF5 by default for Fedora 39 but that is no longer going to work out... FESCo has decided to reject DNF5 for Fedora 39 and then due to Red Hat Enterprise Linux 10 branching with Fedora 40, this means DNF5 isn't expected by default until at least Fedora 41 in late 2024...
FreeBSD developers have published their Q2-2023 status report where they outlined various technical milestones and software accomplishments for this leading BSD operating system...
Going along with Intel adding Lunar Lake and Arrow Lake support to the GCC compiler, Intel has also now contributed the new ISA extensions for these future processors to the GNU Assembler "Gas" support as part of their early compiler toolchain enablement...
Earlier this month Canonical asserted control over the LXD project. As another step in tightening up control over this container management extension for Linux Containers (LXC) is now apparently limiting LXD maintainership rights to only Canonical employees...
Following the recent KDE Akademy developer conference, prominent KDE developer Nate Graham has provided more insight into some of the features being removed with the in-development Plasma 6 desktop...
Released back in April was GCC 13.1 as the first stable release in the GCC 13 series that brought Modula-2 language support, more C++23/C23 features, and other new CPU targets supported from Arm to Intel. Debuting today is GCC 13.2 as the first point release in the series to ship dozens of bug fixes...
While this week was the surprise announcement of Intel AVX10 and with that taking the super-set of AVX-512 to both E and P core processors in the future, for next year's Xeon "Sierra Forest" server processors at up to 144 cores, it appears they will lack AVX-512/AVX10. Intel's AVX10 announcement noted initial support with Granite Rapids processors that will debut next year but no mention of the E-core-only Sierra Forest. With the AVX10 only coming to P/E core client processors after Granite Rapids, it would appear the high density Sierra Forest generation will miss out on AVX10/AVX-512 and not appear until Clearwater Forest. Meanwhile with the 128-core AMD EPYC "Bergamo" processors now shipping, there is AVX-512 with the Zen 4C cores. Here are some benchmarks looking at the AVX-512 impact for Bergamo.
Mold 2.0 is out today as a major update to this high performance linker developed by Rui Ueyama. Mold has consistently shown to outperform GNU's Gold and LLVM's LLD linkers while today is making another shift with it now turning to MIT licensing...
NVK as the open-source NVIDIA Vulkan driver being developed for Mesa has to this point been developed out-of-tree as it's been in its early stages, depends upon Nouveau DRM kernel driver improvements, and ultimately isn't too useful until the Nouveau GSP/re-clocking situation is sorted out upstream. But overnight the merge request was opened to introduce NVK to mainline Mesa...
There's been talk of new Radeon RX 7000 series graphics cards this quarter and adding some weight to that is AMD publishing several new firmware files for different intellectual property blocks of IP versions previously not seeing firmware binaries in the linux-firmware.git repository...
Thanks to work carried out by Sony engineers and then offered for upstream, over the past few weeks support for a unified LTO bitcode front-end has materialized within the LLVM codebase...
For a few years now Canonical has partnered with Intel to offer Ubuntu images optimized for their hardware like with IoT initiatives while the latest now are new Ubuntu real-time images optimized for Intel Core usage...
Two weeks ago a bug report was opened for Mesa that when using Radeon RX 7900 XT or Radeon Pro WX 9100 graphics, Blender's Eevee shader node trees are unusably slow. A fix has now been merged in reducing that shader compilation time from around 251 seconds to now getting done in just about 9 seconds...
Just days after covering a recent ~10% speed-up for the Intel Arc Graphics on Linux, another optimization has landed in Mesa 23.3-devel for further enhancing Intel's latest graphics wares with their open-source driver stack.
For processors supporting CPU performance boosting with higher performance states available beyond the base states, Linux allows toggling the boosting on a per-CPU basis. However, a new patch proposed this week would allow per-policy performance boosting where capable...
As part of AMD's recent Linux graphics driver development approach of enabling new GPU support gradually on a IP block-by-block basis rather than big monolithic patch series marked by colorful fishy codenames, it's worked out well for getting new hardware support rolling into the kernel early and without revealing any combined details on yet-to-be-released graphics processors. This week has seen some new IP block patches surface...
Loongson Binary Translation (LBT) support for the Chinese LoongArch CPU architecture is slated for introduction in the upcoming Linux 6.6 cycle. Loongson Binary Translation aims to help speed-up and handle ARM / x86 / MIPS binary translation on LoongArch more efficiently with capable LoongArch processors...
While the EROFS Linux read-only file-system already supports LZ4 and microLZMA support, Zlib DEFLATE support is also being worked on and could be introduced in the next Linux kernel cycle...
The first beta is out of FreeRDP 3.0, the open-source Remote Desktop Protocol (RDP) implementation of that proprietary Microsoft protocol for remote desktop/GUI handling...
Along with detailing Advanced Performance Extensions (APX), Intel as effectively a footnote to that also disclosed another exciting addition to find with future Intel CPUs: AVX10. Most notably for consumer use is that AVX10 will enable AVX-512 capabilities across both Performance and Efficient core designs with hybrid processors...
Following Advanced Vector Extensions (AVX) and more recently Advanced Matrix Extensions (AMX) for furthering the x86_64 CPU compute potential, Intel has now published initial details on APX: Advanced Performance Extensions...
Last week in the AMD EPYC 9684X review were many benchmarks looking at how this flagship Genoa-X processor compares to various AMD EPYC and Intel Xeon processors. The EPYC 9684X delivers terrific generational uplift compared to Milan-X, offers significant advantages over the EPYC 9004 Genoa processors thanks to the 1.1GB of L3 cache per CPU that proves very beneficial in HPC and AI workloads, and the 96-core / AVX-512 / 3D V-Cache combination far surpassed the Intel performance in the vast majority of benchmarks. As some follow-up benchmarks, today is looking precisely at the performance difference caused by the 3D V-Cache presence by looking at the EPYC 9684X performance when the 3D V-Cache feature was enabled and then the tests repeated when disabled.
It looks like the updated Family 17h microcode this morning isin relation to a new Zen 2 CPU security vulnerability being disclosed. The Linux kernel has also just received a patch for this "Zenbleed" vulnerability for older AMD CPUs...
Qiang Yu continues leading the charge on integrating the ACO compiler back-end into the RadeonSI Gallium3D driver as an optional alternative to the AMDGPU LLVM shader compiler back-end...
Merged back in 2021 with Linux 5.16 was the FUTEX2 code to help with Linux gaming needs particularly around Steam Play. There were plans to further extend FUTEX2 and now two years later there's been recent patches working out more enhancements to this interface...
The most popular topic among the emails I received this weekend weren't of direct technical nature but the number of people pointing out Twitter's new "X" logo and the similarities to the X.Org logo...
Inkscape 1.3 is now available as the newest feature release ot this open-source software focused on being a vector graphics editor that can rival the likes of Adobe Illustrator...
MPV 0.36 is out today as the newest version of this open-source media player that was originally forked from the MPlayer/mplayer2 code and leveraging the FFmpeg library...
Debian 13 "Trixie" has been aiming for official RISC-V support and indeed it will happen: RISC-V has now been promoted to an official Debian CPU architecture...
Automatic IBRS is a new feature with AMD Zen 4 processors akin to Intel's Enhanced IBRS functionality. Linux 6.3 added Auto IBRS support but it turns out when that was being enabled an oversight was made...
Intel Linux engineer Peter Zijlstra's EEVDF CPU scheduler code to replace the existing Completely Fair Scheduler "CFS" code looks like it will attempt to land with the upcoming Linux 6.6 merge window...
Mesa's LLVMpipe software driver is now exposing system Shared Virtual Memory (SVM) support with the necessary API bits being in place for the modern Rusticl OpenCL driver as well as the older Clover code. Plus with being a CPU-based driver there isn't any added work or complications around shared virtual memory...
Red Hat engineers are looking at making it more evident on Fedora IoT, CoreOS, and Server editions when firmware updates become available for the hardware in use...