Intel News Archives

Intel's open-source Linux engineers continue to be quite busy bringing up CXL interconnect support within the mainline kernel. For the in-development Linux 5.16 is another batch of code landing.

With the embargo lifted following this morning's Intel Core i5 12600K + Core i9 12900K Linux review, I've begun uploading more public test data to OpenBenchmarking.org and making my earlier test results public. With that and initial data flowing in from others in the community, here is some more data to poke through if interested in Alder Lake on Linux.

After going through a number of rounds of patch revisions over the past year, Intel's kernel-side changes for supporting Advanced Matrix Extensions (AMX) with next-gen Xeon Scalable "Sapphire Rapids" processors has landed for Linux 5.16!

Along with announcing Alder Lake and other hardware advancements, Intel is using their new Innovation event kicking off today to also talk more about their vast collection of software... This follows Intel CEO Pat Gelsinger's comments from earlier in the week around a bias towards open-source and a pledge to openness. As part of this, Intel is today announcing a new and unified Developer Zone.

It's been over one year since Intel disclosed Advanced Matrix Extensions and began posting patches for bringing up AMX support under Linux in anticipation of Xeon Scalable "Sapphire Rapids" processors. While the compiler-side work to GCC and LLVM/Clang has been landing, finally with the forthcoming Linux 5.16 cycle that AMX support appears ready for landing.

Ahead of Intel's inaugural Intel Innovation event taking place virtually later this week, Intel CEO Pat Gelsinger published an open letter to an open ecosystem.

While for many years we have been accustomed to seeing Intel land their new hardware feature enablement work in the Linux kernel and related components well ahead of products shipping, occasionally there are lapses due to various internal and external timings. The launch of Sapphire Rapids is quickly approaching and one of the major additions is Advanced Matrix Extensions with its Linux support still being in the works.

Intel's "parallel submission" user-space API for their i915 kernel graphics driver has been queued into DRM-Next today ahead of the Linux 5.16 kernel cycle.

Last year Intel announced ControlFlag as a machine learning tool for helping to uncover bugs within code. ControlFlag promised impressive results after being trained on more than one billion lines of code and at the end of 2020 was already being used internally on Intel's code-bases from firmware to software applications. We hadn't heard anything more about ControlFlag this year... Until today. Intel has now made ControlFlag open-source for helping to autonomously detect more programming bugs.

The drama around DMA-BUF code for the Habana Labs AI driver appears to be wrapping up with the Linux 5.16 cycle that is coming up.

Intel's open-source engineers have shipped Compute-Runtime 21.41.21220 as the newest version of this Linux compute stack enabling OpenCL and Level Zero support with their graphics processors.

A new feature for upcoming XeHPG graphics that was merged into Intel's open-source Linux OpenGL and Vulkan drivers today was for tessellation distribution and geometry distribution across the graphics hardware.

Intel has contributed AVX-512 optimizations to upstream Numpy. For those using Numpy as this leading Python library for numerical computing, newer Intel CPUs with AVX-512 capabilities can enjoy major speed-ups in the range of 14~32x faster.

It's been one year now that Intel has been posting Linux kernel patches to enable AMX support for upcoming Sapphire Rapids processors. Over the past year their Linux kernel patches for enabling Advanced Matrix Extensions has gone through 11 rounds of review but that journey isn't over yet.

While Linux 5.15 brings very early bits around DG2/Alchemist graphics card support, further work is needed to bring it into usable shape for end-users. The latest new patch series to be posted came out today with more driver changes needed around local device memory handling for DG2.

Sent in this morning were an urgent set of x86 updates for the Linux 5.15-rc5 kernel due out later today.

Intel has been preparing Xe HP bring-up for many months already including fundamental work around their discrete graphics/accelerator support for their Linux graphics driver stack going back quite a while. On the Xe HP front, Friday afternoon brought an important patch series posted for the first time: initial work around multi-tile support.

A new batch of Intel kernel graphics driver code was mailed out today to DRM-Next for staging ahead of next month's Linux 5.16 kernel merge window. Lots of notable changes in this pull!

With the Linux 5.15 kernel there is a patch to benefit tiered memory systems with a focus on servers having persistent memory. That patch is demoting pages during page reclamation to slower tiers of memory such as Optane DC persistent memory. Intel continues building on that and other persistent memory kernel work for plumbing the kernel with optimized memory placement for these modern servers.

Intel's CM Compiler is out with its first big update since earlier in the year for advancing their "C For Metal" GPU programming language effort.

Somewhat of a surprising change with Intel Gen12.5 graphics is that they have removed the hardware supporting Adaptive Scalable Texture Compression (ASTC). Intel's Linux graphics driver has now been updated to address Gen12.5+ foregoing hardware support for ASTC texture compression.

In Linux 5.14 Intel introduced initial Alder Lake P enablement driver support including around the new "XeLPD" display block. With Linux 5.15 there was the initial enablement around DG2/Alchemist graphics. Now for Linux 5.16 is a significant amount of new driver code for actually getting the display support into shape for both DG2 and ADL-P.

While Intel Xeon "Sapphire Rapids" processors with Advanced Matrix Extensions are set for a Q2'22 ramp in production, one of the key new features that has yet to be properly plumbed in the mainline Linux kernel is for supporting AMX.

Intel's newest weekly Compute-Runtime update providing open-source OpenCL and oneAPI Level Zero support for their graphics hardware is now reader with broader support for upcoming Alder Lake S processors.

Intel on Friday formally released their oneAPI Toolkits 2021.4 release as the latest collection of their various software components for a multi-vendor, multi-architecture software platform across CPUs and XPUs (GPUs / accelerators).

Intel's open-source Linux graphics driver engineers are evaluating possible improvements to the Linux kernel for accommodating CPU and GPU synchronized priority scheduling.

Intel has some new announcements around their neuromorphic computing research.

There has been talk of Intel moving to offer more license-able/opt-in features for hardware capabilities found within a given processor as an upgrade. We are now seeing the Linux signs of that support coming with a driver for "Intel Software Defined Silicon" to allow for the secure activation of such features baked into the processor's silicon but only available as an up-charge option.

Earlier this month Intel engineers posted their initial Linux kernel enablement around x86 User Interrupts with this feature premiering with Xeon "Sapphire Rapids" CPUs. As implied by the name, the User Interrupt functionality allows for interrupts to bypass the kernel for more efficient, low-latency, low-utilization interrupts being received by other user-space tasks. Intel talked more about User Interrupts this week at LPC2021.

While Intel is normally quite good with their new hardware support being in good shape well ahead of launch, their new code for supporting the ENQCMD functionality for the Data Streaming Accelerator (DSA) with Xeon "Sapphire Rapids" has been an exception. This summer the mainline Linux kernel disabled ENQCMD support since the code was "broken beyond repair" while now Intel engineers have sent out a new series looking to get it re-enabled.

Along with bringing up DG2/Alchemist graphics card support on Linux, Intel engineers have been working to square away their support for the DG1 developer graphics card. This week thanks to XDC2021 is a fresh status update about what is working with this initial Intel graphics card on their open-source driver and what remains in the works.

It looks like Intel's ISHTP_ECLITE driver will be ready for mainlining in Linux 5.16 as a driver for newer systems skipping out on a traditional embedded control (EC) and instead using this EC-like IP as part of their Programmable Service Engine subsystem.

Prominent Intel open-source Vulkan Linux driver developer Jason Ekstrand presented at today's X.Org Developers Conference (XDC2021) about their work on enabling Vulkan ray-tracing support.

Intel has released a new version of their loader for oneAPI Level Zero for loading the Level Zero software driver components.

"Intel Seamless Update" is a forthcoming feature for Intel platforms seemingly first being exposed by their new Linux kernel patches working on the functionality... Intel is working on being able to carry out system firmware upates such as UEFI updates but doing so at run-time and being able to avoid the reboot in the process.

In addition to the big Advanced Matrix Extensions support still being in flux and the kernel-side AMX code not yet being merged, another feature of next year's Xeon "Sapphire Rapids" that we are only now seeing in early published form for the Linux kernel is handling of x86 user interrupts.

While much of Intel's next-gen Alder Lake processor support appears to be in good shape for Linux 5.14, some remaining items are landing for the current Linux 5.15 cycle. The latest Alder Lake support hitting the kernel is for Intel's TCC cooling driver.

Intel-owned Habana Labs now has the most open software stack among AI accelerators! While Habana Labs has long provided an open-source, upstream kernel driver for their Gaudi AI training and Goya AI inference accelerators, the user-space portions including their code compiler and run-time library have been closed-source. This has been a thorn for upstream kernel developers and their standards, but now Habana Labs has open-sourced their user-space components too.

Last month Intel's open-source Linux graphics driver engineers began posting patches working on DisplayPort 2.0 support for their driver with DG2/Alchemist now set to be Intel's first GPU supporting the newest DP standard. DP 2.0 enablement work continues with Panel Replay being the latest feature being worked on for their Linux driver.

Intel Compute Runtime 21.35.20826 is available today with initial support for oneAPI Level Zero v1.2.

Landing via Andrew Morton's patch series today in the Linux 5.15 kernel is handling for demoting pages during memory reclaim, which can be used for punting cold pages off to slower, tiered memory devices (like Intel persistent memory) when under system memory pressure.

Intel's i965 classic DRI driver is still the default within Mesa for i965 through Haswell generations of Intel integrated graphics, but the new "Crocus" Gallium3D driver has been added to the default driver build list so it's now at least building by default on x86/x86_64 systems and thus trivial after that to override.

While the Linux 5.15 merge window opening is imminent, merged today to net-next were the latest batch of wireless driver updates for this next kernel version. Notable to this batch of WiFi driver updates was the new Intel material.

In preparation for the support within their Vulkan driver, the Intel-led effort for preparing mesh shader support within Mesa's NIR and SPIR-V code has now been merged.

The Linux kernel has already sported SM4 cipher algorithm implementation optimized for AES-NI and AVX while now an Alibaba engineer has contributed an AVX2 optimized variant for even greater performance.

Going back to June of last year there has been work on Intel bringing up Advanced Matrix Extension (AMX) that will debut with next-gen Xeon "Sapphire Rapids" processors as a new programming paradigm. Over the past year they have published patches for the Linux kernel and open-source toolchains with GCC and LLVM Clang. One year later, the AMX kernel patches are up to their tenth revision but will miss out on the imminent Linux 5.15 merge window.

Shortly after Intel's TDX whitepaper was made public last year for better protecting virtual machines, Intel open-source engineers began posting support patches for bringing up Trust Domain Extensions under Linux. That work remains ongoing and now further Linux kernel infrastructure work is pending to better deal with the notion of guest private memory afforded by TDX.

One year ago to the day Intel announced OSPray Studio as a scene graph application for rendering glTF assets and other 3D models. OSPray Studio is built off their OSPray ray-tracing engine that they've been working on for years. These Intel efforts are all part of their oneAPI initiative and today happens to mark a shiny new feature release.

Going on three years now there have been proposed patches for allowing per-client GPU engine statistics for being able to show on a per-game/application level how many resources across 3D/blitting/video engines are being consumed. The patches continue to be revised but sadly will be missing out on the imminent Linux 5.15 kernel merge window.

Since the start of July we've seen Intel beginning Linux support patches for their DG2 graphics card that is now known by the "Alchemist" codename. There's been several rounds of DG2 patches since they started publicly pushing out the code -- including some notable work like DisplayPort 2.0 bring-up -- while sent out this Sunday is another important piece of the puzzle: getting the device memory (the dedicated vRAM) actually working with the open-source driver.