Intel News Archives

As some more exciting news for upcoming Xe2 graphics with Lunar Lake integrated graphics and Battlemage discrete GPUs, the latest open-source driver activity for Linux has confirmed Xe2 supporting native 64-bit integer arithmetic.

It's not too common for Intel to publish new CPU microcode updates outside of their "Patch Tuesday" regiment but that happened yesterday with a Friday night release of new CPU microcode although this time is limited to the Celeron and Pentium Silver families.

Intel engineer Noah Goldstein has landed another nice performance optimization in the GNU C Library "glibc" for benefiting newer Intel processors.

The Intel Battlemage discrete graphics support is beginning to come together for the open-source Linux graphics driver stack as the successor to DG2/Alchemist. In addition to all the Xe2 work for what's found with Lunar Lake, more Battlemage Linux kernel and user-space driver work has been appearing recently. The milestone crossed today is the initial Battlemage "BMG" platform support being merged for the Mesa 24.2 OpenGL/Vulkan drivers.

As part of Intel's Flexible Return Event Delivery (FRED), Intel open-source software engineers are now working on improving Non-Maskable Interrupt (NMI) source reporting for the Linux kernel.

The past year there's been a big Linux kernel patch series in the work by Intel to improve Sub-NUMA Clustering "SNC" support so it behaves well with Intel Resource Director Technology (RDT) on modern Intel hardware. Hopefully that work will soon be ready for mainlining in the Linux kernel while this week brought the 19th revision to those patches.

Intel Performance Limit Reasons (PLR) can indicate why your CPU is downclocking / limited to a lower performance limit for a given core or die. With some Windows utilities this information this Intel CPU feature has been available there while now Intel is bringing PLR support to Linux too.

The Intel Implicit SPMD Program Compiler (ISPC) is out with a new version today for this C programming language variant that features SPMD programming extensions. Intel ISPC aims to make it easy to take advantage of SIMD capabilities on their modern processors as well as GPUs.

Intel's open-source NPU "iVPU" Linux kernel driver for supporting their Neural Processing Unit beginning with Core Ultra "Meteor Lake" processors is already seeing a code refactoring. The refactoring of this Intel accelerator driver is intended for allowing more versatile CPU and NPU combinations moving forward.

Intel software engineers did a late Friday night release of the Intel NPU Acceleration Library v1.1, their Python library for tapping into the Intel Neural Processing Unit (NPU) found on the new Core Ultra (Meteor Lake) processors. This Python library makes it easier to interface with the Intel VPU/NPU kernel driver and in turn enjoying accelerated operations for AI.

Intel previously indicated that Lunar Lake processors would launch by the end of 2024 and leading to anticipation of a Q4 launch... Intel today announced that Lunar Lake will actually launch in Q3.

Last week I wrote about Intel aiming to remove Xeon Phi support in GCC 15 with the products being end-of-life and deprecated in GCC 14. While some openly wondered whether the open-source community would allow it given the Xeon Phi accelerators were available to buy just a few years ago and at some very low prices going back years so some potentially finding use still out of them especially during this AI boom (and still readily available to buy used for around ~$50 USD), today the Intel Xeon Phi support was indeed removed.

Intel Compute Runtime 24.17.29377.6 is now available as the latest routine update to this open-source GPU compute stack used by the company's integrated and discrete graphics products for providing OpenCL and oneAPI Level Zero compute capabilities.

The performance events updates were submitted today for the ongoing Linux 6.10 kernel merge window. This pull adds support for Advanced Performance Extensions (APX) and other new Intel CPU instructions to the x86 instruction decoder.

For the GCC 14 compiler release is the deprecation of the Xeon Phi targets. With Intel Knights Landing and Knights Mill being end-of-life at Intel, they are working to do away with the GNU Compiler Collection support. A patch has been posted to drop the Xeon Phi ISAs with GCC 15.

Coming as a surprise, longtime Linux developer Oded Gabbay announced he's left Intel / Habana Labs and is therefore stepping down from the maintainer role of the Linux kernel drivers for the Intel Xe DRM driver and more notably the Habana Labs accelerator driver that he's maintained from the start.

Intel's Image Processing Unit (IPU) IP has been a cause for concern in recent years as the lack of proper upstream open-source driver support has led Linux users running into troubles making use of MIPI camera sensors on modern laptops. Finally with Linux 6.10 the Intel IPU6 driver is being upstreamed into the media subsystem.

One of the capabilities of newer Intel Xeon Scalable processors is support for Trust Domain Extensions (TDX) as a way of providing for confidential virtual machines. Intel TDX allows for "isolation, confidentiality, and integrity at the VM level" which is good from the security perspective but the dependence on signed binaries is causing mixed feelings within the Fedora camp at the broader open-source community.

Along with the various Intel Xe2 driver changes that are ongoing for the Intel Linux graphics driver stack, over in the sound subsystem the in-development Linux 6.10 kernel is bringing HDMI audio support for upcoming Intel Battlemage graphics cards.

Intel just published a new set of CPU microcode files for updating Alder lake and newer as well as Xeon Scalable 4th Gen and 5th Gen in order to address three security issues plus take care of various functional issues.

Intel, HPE, and Argonne National Laboratory have announced at ISC High Performance 2024 that the Aurora Supercomputer has broken the Excascale barrier and is now the fastest AI supercomputer currently in existence.

With the new Intel "Xe" Direct Rendering Manager kernel driver that's been in development one of the touted benefits of the clean sheet driver design is that it would enable using Intel discrete GPUs on non-x86 CPU architectures. The long-used "i915" DRM kernel graphics driver has many x86'isms in the code-base built up over the many years of Intel integrated graphics that were only ever found within their x86/x86_64 processors. But now in the era of Intel discrete graphics, there's been issues in trying to run Intel Arc Graphics on say ARM, POWER9, and RISC-V, among others. The experimental Intel Xe driver was recently successfully demonstrated in running on ARM using an Ampere Altra workstation.

While Intel can be praised for their dozens (or likely by now, hundreds) of open-source projects they maintain and countless other existing open-source software projects they actively contribute to and are covered by Phoronix on a near-daily basis, not everything there is open-source. Intel is a wonderful and leading open-source promoter but occasionally there are closed-source blobs or questionable moves such as today: Intel is taking their Hyperscan library development from BSD-licensed open-source software to now the Intel Proprietary License moving forward.

Intel has released their Intel Extension for PyTorch v2.3 to succeed their earlier v2.1 derived extension. With this updated extension targeting PyTorch 2.3, Intel is rolling out more optimizations around Large Language Models (LLMs).

The Intel iVPU accelerator driver changes for the upcoming Linux 6.10 merge window have been submitted for advancing the Neural Processing Unit (NPU) support found since the launch of Meteor Lake with Intel Core Ultra notebook CPUs. For this iVPU/NPU driver in Linux 6.10 are a few notable new features.

Since last year Intel's open-source software engineers have been working on a PCIe bandwidth controller driver for the Linux kernel to avoid thermal issues by being able to automatically reduce the PCIe link speed when needed. This driver still isn't over the finish line but today brought the fifth iteration of these patches.

Intel's power management lead Rafael Wysocki posted a set of patches recently for working out asymmetic CPU capacity on hybrid Core x86 systems.

Intel software engineer Victor Rodriguez presented at the Open-Source Summit North America last month on their open-source compiler toolchain work for enabling ISA capabilities of upcoming Intel CPUs as well as using simulation tools for helping to test compiler enhancements/optimizations moving forward.

Intel engineers have been reworking Intel CPU model handling for Linux after using "Family 6" since the mid-90's with the P6 micro-architecture and continuing to rev the model ID only with new micro-architectural generations. It's an end of the era for Family 6 coming up and thus there's a lot of Linux patches being worked on to address assumptions within the kernel code that was only checking for an Intel CPU's model ID and not for any family ID differences.

Following the big set of Xe DRM driver updates for Linux 6.10 and earlier Adaptive Snyc SDP, Lunar Lake display support, and more DG2 PCI IDs for i915 pulls sent in over weeks prior for this next kernel version, the drm-intel-gt-next pull request was submitted today for last minute Intel graphics driver feature changes aiming for Linux 6.10.

Intel engineers have just released OpenVINO 2024.1, the newest feature release for this excellent open-source AI toolkit that continues expanding its features and capabilities particularly around Generative AI "GenAI" and Large Language Models (LLMs).

Following yesterday's Mesa 24.1 feature branching, Mesa 24.2-devel is now open for the Mesa Git mainline code and some early feature work has begun for that Q3 release series.

Intel today sent out more than one hundred new feature patches to DRM-Next of new "Xe" kernel graphics driver material they have readied for the upcoming Linux 6.10 kernel merge window.

Intel's open-source Linux graphics driver engineers have been busy working to enable the display support for the upcoming Battlemage graphics cards as the successor to DG2/Alchemist.

The Intel Media Driver 2024Q1 release is now available that serves as the company's modern Video Acceleration API (VA-API) driver for Linux systems. The Intel media driver allows for iGPU/dGPU-based video encode/decode for HEVC, VP9, AV1, and other formats supported by the respective graphics hardware.

As the first new release to Intel's open-source Compute Runtime stack in about one month for this OpenCL and Level Zero compute support, Intel Compute Runtime 24.13.29138.7 was released this morning with much improved OpenCL/OpenGL sharing and interoperability on Linux, out-of-the-box support for the Xe kernel graphics driver, new optimizations, and many other changes.

Overnight Intel released their oneVPL GPU Runtime 2024Q1 release for this media stack component to their oneAPI software collection.

On Wednesday the latest round of drm-intel-next material was submitted to DRM-Next ahead of the Linux 6.10 kernel merge window. Intel's open-source engineers remain very busy working on the i915 and Xe kernel graphics drivers with new display features, expanding hardware support, and other functionality.

In addition to Vulkan explicit sync under X11, another merge request hitting Mesa 24.1 overnight that's worth mentioning is the open-source Intel "ANV" Vulkan driver now supporting VK_EXT_image_compression_control.

With the new security mitigation for the "Native BHI" Spectre vulnerability affecting even the recent Intel processors, a number of Phoronix readers have been curious about the performance impact of the mitigation. Over the past week I've been running some benchmarks on recent Intel CPUs to better look into any performance implications.

Intel's software team is today sharing their newest innovation for achieving greater performance on Linux systems: the Thin Layout Optimizer. Intel's Thin Layout Optimizer is inspired by the likes of the Meta/LLVM BOLT optimizer and Google's Propeller but aims to be much easier to use while still delivering measurable performance gains for optimized binaries.

The turbostat utility is useful on Linux systems for reporting idle/power-state statistics, temperatures, and other useful metrics for modern CPUs. It's also able to dive deeper and provide various MSR values and counters and other intriguing CPU bits. For much of these features root access is required and thus turbostat has bailed out up to this point if not running as root. But as a number of the metrics can still be obtained without root access, turbostat is finally being adapted to handle running better as a non-root user.

Intel's FFmpeg Cartwheel is where they maintain their various yet-to-be-upstreamed patches for the FFmpeg multimedia library either to enhance/enable new Intel graphics hardware support or improve/add extra functionality to this widely-used open-source library. With the Intel FFmpeg Cartwheel 2024Q1 release they are shipping a new filter for dealing with older content as well as several other new features.

For several years Intel has been developing the OpenCL Intercept Layer to assist in debugging OpenCL software. It's been nearly two years since the last release of this open-source OpenCL interception layer while today brings v3.0.4 with a number of optimizations and new features.

Following the Intel Linux kernel graphics driver patches last month adding two new DG2/Alchemist PCI IDs that when digging through the Intel Compute Runtime sources were confirmed as the Arc Graphics A580E and A750E, the Mesa OpenGL/Vulkan drivers have now added support for these new graphics processors.

Queued up recently into the crypto subsystem's development branch ahead of the Linux 6.10 merge window is support for VFIO live migration with Intel's QuickAssist Technology (QAT) driver.

Intel's Hardware Feedback Interface (HFI) driver will act more efficiently come the Linux 6.10 kernel this summer.

Along with AMD's upcoming hardware enablement and other Q1 code contributions, Intel's open-source engineers remained very busy this quarter as well. Intel continues upstreaming a lot of new code not only for upcoming hardware but also a lot of exciting Linux kernel features in general, various optimizations and improvements to countless open-source user-space software projects, and their other great open-source efforts in general.

Intel is ending out the month and quarter with the latest update to its open-source Compute Runtime and Intel Graphics Compiler (IGC) code that enables OpenCL and Level Zero support on Linux systems and is also used by their Windows driver too.

Since the mid-90's with the P6 micro-architecture for the Pentium Pro as the sixth-generation x86 microarchitecture, Intel has relied on the "Family 6" CPU ID. From there Intel has just revved the Model number within Family 6 for each new microarchitecture/core. For example, Meteor Lake is Family 6 Model 170 and Emerald Rapids is Family 6 Model 207. This CPU ID identification is used within the Linux kernel and other operating systems for identifying CPU generations for correct handling, etc. But Intel Linux engineers today disclosed that Family 6 is coming to an end "soon-ish".