Language Selection

English French German Italian Portuguese Spanish

Graphics/Benchmarks

Graphics: Mesa 20.1.8 Released and More

Filed under
Graphics/Benchmarks
  • mesa 20.1.8
    Hi all,
    
    I'd like to announce Mesa 20.1.8, the eighth bugfix release for the 20.1 branch.
    
    The next bugfix release is planned for 2 weeks from now, on 2020-09-30.
    
    Cheers,
    Eric
    
  • Mesa 20.1.8 Released While Waiting For Mesa 20.2

    Mesa 20.2 (or 20.2-RC5) didn't debut last week as intended, but for the interim the Mesa 20.1.x release cycle brought 20.1.8 on Wednesday and now it's been extended to having at least a ninth point release to allow more time until not only Mesa 20.2.0 to ship but Mesa 20.2.1 alignment.

  • Ray-Tracing Support For AMDGPU LLVM Back-End Lands For RDNA 2

    AMD previously confirmed it would be supporting real-time ray-tracing with their next-generation GPUs while now one month out from the Radeon RX 6000 series debut are the first signs of the open-source driver work around GPU ray-tracing.

    One day after spotting the patches for AV1 video decode with VCN 3.0, the latest open-source Radeon driver work to point out is the fundamentals around their ray-tracing introduction.

  • NVIDIA 455.23.04 Linux Beta Released With GeForce RTX 3080/3090 Support

    NVIDIA has once again managed to provide launch-day Linux driver support for their next-generation graphics processors. Today the NVIDIA 455.23.04 beta driver is shipping for Linux support with the GeForce RTX 3080 and RTX 3090 "Ampere" graphics cards.

  • RADV's "ACO" Shader Backend Still Pursuing RadeonSI, Early Work On RDNA 2

    Valve developer Timur Kristóf who has been spending the past year working on the AMD Compiler "ACO" back-end for the Mesa Radeon Vulkan driver "RADV" as well as beginning to port this shader compiler back-end to RadeonSI Gallium3D. This alternative to the AMDGPU LLVM back-end has made incredible progress over the past year -- enough so that it's been the default for Mesa's RADV driver. During XDC2020 Day 2, Timur provided an update on ACO.

  • Cache Creator Tool Proposed For AMDVLK Vulkan Driver

    Google engineer Steven Perron has laid out their proposal for an XGL cache creator tool for AMD's official Vulkan Linux driver, AMDVLK.

    As part of their work on relocatable shaders and supporting the offline compilation of Vulkan/SPIR-V shaders, they are working on "xgl_cache_creator" as a tool to take precompiled shaders and construct a file that can be redistributed and passed as the initial data into the Vulkan pipeline cache.

  • Arm Is Now Backing Panfrost Gallium3D As Open-Source Mali Graphics Driver

    Most information presented during the annual X.Org Developers' Conference doesn't tend to be very surprising or ushering in breaking news, but during today's XDC2020 it was subtly dropped that Arm Holdings appears to now be backing the open-source Panfrost Gallium3D driver.

  • Microsoft Has A Large Presence At This Year's X.Org Conference [Ed: Microsoft is now interjecting Windows and DirectX into conferences about Linux]

    Years ago if saying Microsoft would have multiple developers presenting at the annual X.Org Developers' Conference (XDC) as well as being a sponsor, you'd probably raise some laughs. But this year for XDC2020 Gdansk (albeit virtual due to COVID-19), Microsoft engineers gave not just one talk but three on the opening day.

    [...]

    Jesse Natalie and Steve Pronovost both of Microsoft kicked off XDC2020 by talking about the WSL graphics architecture in a pre-recorded, well-edited video presentation. That was followed by Pronovost talking about X11/Wayland application support under WSL and then the third and final Microsoft talk of the day was Jesse talking about their Mesa Direct3D 12 mapping layers for getting OpenCL/OpenGL over D3D12.

Graphics: XDC2020, Vulkan and NVIDIA GeForce NOW

Filed under
Graphics/Benchmarks
  • X.Org Developers Conference 2020 Starts With Many Interesting Talks

    XDC2020 as the annual gathering of X.Org developers was due to take place in Poland this year but the COVID-19 pandemic has caused it to be yet another conference happening exclusively online. Day 1 of XDC2020 has just begun.

    XDC2020 is organized by Intel's graphics engineering team in Gdansk, Poland. The virtual event is leveraging existing Linux Plumbers Conference infrastructure as well as YouTube for video streaming. Intel is the platinum sponsor of XDC2020 while Google and NVIDIA are the gold sponsors.

  • After Reaching Vulkan 1.0 Conformance, V3DV Raspberry Pi Vulkan To Pursue Mainline Mesa

    The V3DV driver for providing Vulkan support on the Raspberry Pi 4 is very close to Vulkan 1.0 conformance and once squaring that away along with other lingering bits they will be pursuing the upstreaming of this driver within Mesa.

    Upstreaming the V3DV driver in Mesa will be a huge help for those wanting to see this Vulkan driver readily available on the many Linux distributions shipping Raspberry Pi spins and sticking to just upstream/mainline code. V3DV is very close to Vulkan 1.0 conformance with Quake III Vulkan, vkOpenArena, Vulkan-powered emulators, and various demos now running well on the Raspberry Pi 4 with this driver developed by Igalia under cooperation with the Raspberry Pi Foundation. Thus once upstream in 2021 we are looking at Linux distributions shipping this driver with their upstream Mesa.

  • XDC 2020 conference is today - Vulkan Ray Tracing and Vulkan for Raspberry Pi 4

    Today the X.Org Developers Conference (XDC 2020) begins and there's a couple of interesting talks worth checking out, especially if you like to follow OpenGL and Vulkan. What is the event? The X.Org Developers Conference is the event for developers working on all things open graphics related including in the Linux kernel, Mesa, DRM, Wayland, X11 and so on.

    Due to the ongoing COVID19 pandemic, it's going to be quite a streamlined virtual event spread across three days from September 16 to September 18. There's still going to be quite a few talks and lots of them sound really interesting.

  • NVIDIA GeForce NOW on Linux can run without user agent spoofing in a browser

    Looks like NVIDIA might be ready for the next push in cloud gaming with their GeForce NOW service, as it's even easier to run it on Linux. What is it? GeForce NOW allows you to play games you already digitally own on other platforms, on whatever device you have available. It hooks in with Steam, EA / Origin, Epic Games and more.

    Back in August, NVIDIA officially opened it up to Chromebooks by letting people playing with it in the browser. For everyone else, it needed you to spoof your user agent string to act like it was a Chromebook. It was a small thing but still a minor nuisance. That now, appears to no longer be needed.

Graphics Leftovers

Filed under
Graphics/Benchmarks
  • Intel Graphics Driver Patches Revived For Per-Client Engine Activity

    One of the interesting Intel Linux graphics driver patches to be sent out last year were for per-client engine reporting to allow on a per-application/process basis to see how the GPU's render/blitter/video engines were being utilized.

    That work for per-client "engine busyness" reporting went through a few rounds of review but as of Linux 5.9 there still isn't the support within Intel's i915 kernel driver.

  • Vulkan Present Timing Extension With Aim To Avoid Stuttering

    The Vulkan System Integration working group has decided to publish their work-in-progress extension on VK_EXT_present_timing as their effort to expose the presentation engine's display details and better allow scheduling a present to happen at a specific time.

    VK_EXT_present_timing is a big effort about helping to reduce stuttering and use-cases like better handling of variable refresh rate setups and other scenarios in wanting to ensure the presentation of a frame/image happens on schedule to avoid anomalies.

  • Intel Graphics Compiler 1.0.4944 Brings Many Minor Optimizations

    The Intel Graphics Compiler (IGC) that is used by their Linux OpenCL/Level-Zero compute stack as well as now being used by their Windows graphics driver and potentially their Linux OpenGL/Vulkan drivers in the future is out with a new release.

    IGC releases tend to come frequent with a large team working on this open-source graphics compiler code while the 1.0.4944 milestone is a bit of a larger release.

  • X.Org Developer's Conference 2020

    After DebConf, Linux Plumbers and Akademy, the lineup of great virtual conferences continues this week with the 2020 edition of X.Org Developer's Conference (XDC), the leading event for developers working on all things Open graphics, including the Linux kernel, Mesa, DRM, Wayland and X11.

    Taking place entirely online for the first time, XDC 2020 brings a packed schedule of talks, workshops and lightning talks spread out over three days. Collaborans will giving two presentations & a lightning talk during the week, for which you can find full details below. The entire conference will be live-streamed on YouTube (Day 1, Day 2 & Day 3), however if you would like to take part in any of the discussions, there's still time to register (free of charge)!

  • AMD Radeon Navi 2 / VCN 3.0 Supports AV1 Video Decoding

    It turns out the Radeon RX 6000 series will have AV1 hardware video decode capabilities.

    In addition to Intel Xe / Tigerlake and the NVIDIA GeForce RTX 30 series supporting AV1 hardware decoding, it's now firmed up that the next-gen Navi 2 GPUs will also have AV1 decode.

    As previously reported, the AMD next-gen GPUs feature VCN 3.0 for Video Core Next. The previous VCN 3.0 Linux/open-source patches didn't reveal AV1 capabilities but new patches out today confirm AV1 support with VCN3.

Graphics: VGA Console Code, VGA Issues, and AMD

Filed under
Graphics/Benchmarks
  • Linux 5.9 Dropping Soft Scrollback Support From FB + VGA Console Code

    Linus Torvalds has decided to do away with the "soft scrollback" functionality found in the FBCON and VGACON kernel code as a sign of the times.

    VGACON/FBCON for the basic Linux console has supported a software scrollback buffer with the Shift + PageUp keyboard sequence for scrolling up in the output for contents out of view. But with most people not making heavy use of the frame-buffer console these days and the code being unmaintained, it's being stripped out from Linux 5.9.

  • Russell Coker: More About the PowerEdge T710

    I’ve got the T710 (mentioned in my previous post [1]) online. When testing the T710 at home I noticed that sometimes the VGA monitor I was using would start flickering when in some parts of the BIOS setup, it seemed that the horizonal sync wasn’t working properly. It didn’t seem to be a big deal at the time. When I deployed it the KVM display that I had planned to use with it mostly didn’t display anything. When the display was working the KVM keyboard wouldn’t work (and would prevent a regular USB keyboard from working if they were both connected at the same time). The VGA output of the T710 also wouldn’t work with my VGA->HDMI device so I couldn’t get it working with my portable monitor.

    Fortunately the Dell front panel has a display and tiny buttons that allow configuring the IDRAC IP address, so I was able to get IDRAC going. One thing Dell really should do is allow the down button to change 0 to 9 when entering numbers, that would make it easier to enter 8.8.8.8 for the DNS server. Another thing Dell should do is make the default gateway have a default value according to the IP address and netmask of the server.

  • AMD Shows First Glimpse Of Radeon RX 6000 Series Graphics Card

    Ahead of the RDNA 2 / Navi 2 reveal on October 28, AMD has shown off the first official render of a Radeon RX 6000 series graphics card.

    On Twitter AMD commented, "Take a first look at the design of the new Radeon RX 6000 series. Our upcoming @AMD #RDNA2 graphics cards will feature a brand new cooler design." They also noted there is a render of the Radeon RX 6000 series card within Fortnite for those wanting to explore there.

Radeon GPU Profiler 1.8 Released With Redesigned Developer Panel

Filed under
Graphics/Benchmarks

AMD today released a new version of their Radeon GPU Profiler utility for Linux and Windows systems for profiling games/applications on Radeon graphics hardware under both Linux and Windows.

Among the changes with this new Radeon GPU Profiler 1.8 release include:

- Support for Ubuntu 20.04 LTS. However, Radeon GPU Profiler continues to require AMDGPU-PRO Vulkan rather than Mesa's RADV Vulkan.

Read more

Graphics and Hardware: AMD, OpenGL/Vulkan, ARM/Nvidia and More

Filed under
Graphics/Benchmarks
Hardware
  • AMDVLK 2020.Q3.5 Vulkan Driver Released With Several Game Fixes

    AMD has kicked off the new week with the release of AMDVLK 2020.Q3.5 as their official open-source Radeon Vulkan driver for Linux systems derived in part from the same sources as their Windows Vulkan driver.

    With AMDVLK 2020.Q3.5 there is updating against the Vulkan 1.2.152 API as well as a change to eliminate an invisible copy of trace buffers on AMD APU platforms. Most interesting though with this routine update are the fixes, to which there are several game fixes.

  • AOMP 11.9 Released For OpenMP Offloading To Radeon GPUs

    AOMP 11.9 was released on Friday as AMD's LLVM-based compiler with Clang for C/C++ and Flang for Fortran in offloading capable OpenMP code to Radeon GPUs.

    AOMP 11.9 is AMD's latest work on their LLVM 11 derived code-base for OpenMP GPU compute until the necessary patches have worked their way eventually back into the upstream code-base.

  • Mike Blumenkrantz: Draw Parameters

    Let’s talk about ARB_shader_draw_parameters. Specifically, let’s look at gl_BaseVertex.

    In OpenGL, this shader variable’s value depends on the parameters passed to the draw command, and the value is always zero if the command has no base vertex.

    In Vulkan, the value here is only zero if the first vertex is zero.

    The difference here means that for arrayed draws without base vertex parameters, GL always expects zero, and Vulkan expects first vertex.

    Hooray.

  • Nvidia Buys Arm From SoftBank for $40 Billion

    If completed, the transaction would instantly transform Nvidia into one of the most influential players in smartphone technology, a market that had previously eluded it. Arm, which licenses designs that other companies turn into chips, has long defined the computing technology found in most mobile devices. And Arm designs are starting to play a bigger role in cloud data centers.

    But the deal is likely to prompt close scrutiny by antitrust authorities around the world. Influential Arm customers potentially affected by the transaction include Apple, Samsung Electronics, Amazon.com, Qualcomm and Huawei.

  • NVIDIA Announces $40 Billion Deal To Acquire Arm

    The recent rumors panned out and NVIDIA just announced they have reached a definitive deal with SoftBank to acquire Arm.

    NVIDIA is set to acquire Arm in a deal worth $40 billion USD between cash and stock. The deal is expected to take around 18 months to close and NVIDIA has stated their commitment to keeping Arm independent and their brand identity. Additionally, NVIDIA will keep Arm headquartered in the UK and will also expand Arm's presence there with a new AI research center.

  • NVIDIA to Acquire Arm for $40 Billion from Softbank

    A few weeks ago, I read rumors about NVIDIA acquiring Arm, and I thought it was probably just a joke because of the obvious conflicts of interests since NVIDIA would be providing IP to competitors, who may then be wary of starting designs based on Arm NVIDIA cores and GPUs.

  • NVIDIA confirms $40 billion deal to buy Arm

    Huge industry news to mention this morning! NVIDIA has confirmed they're buying Arm for $40 billion. This news comes after speculation over it for some time, which yesterday was finally announced.

    Before getting wild with speculation about what will happen, NVIDIA noted a few keys points about the acquisition. Notably, they will actually keep the headquartered presence in Cambridge, UK and expand the R&D there with "establishing a world-class AI research and education center, and building an Arm/NVIDIA-powered AI supercomputer for groundbreaking research". Additionally, they will be continuing the same open-licensing model that Arm has along with "customer neutrality" and additionally they will be expanding Arm's IP licensing with some of NVIDIA's own tech.

    Nothing is actually complete yet though, as these take time to go through all the proper channels. This includes regulatory approvals across he U.K., China, the European Union and the United States which they're estimating to take 18 months. See the full announcement here.

  • Nvidia will keep ARM licensing “neutral,” wants to license GPU tech, too

    Nvidia has officially announced that it is buying ARM from SoftBank for $40 billion. The deal is one of the biggest tech acquisitions of all time and will see Nvidia control the world's most popular CPU architecture.

    Nvidia's press release oddly paints the deal as primarily about "AI," saying the deal "brings together NVIDIA's leading AI computing platform with ARM's vast ecosystem to create the premier computing company for the age of artificial intelligence." Nvidia apparently sees GPU-accelerated AI as its next big growth sector, and the company currently sells embedded systems for self-driving cars and multi-GPU systems for workstations and servers, offering high-teraflop deep-learning performance. Somehow it thinks ARM will help with this.

  • Phoronix Test Suite 10.0 Milestone 2 Now Available For Testing

    The second development release of the forthcoming Phoronix Test Suite 10.0-Finnsnes is now available for testing.

    Following last month's debut of Phoronix Test Suite 10.0 M1 and the alpha of the new OpenBenchmarking.org, the second development release is now available ahead of the planned Phoronix Test Suite 10.0 release in October.

Mesa 20.3 + Linux 5.9 Is In Great Shape Against AMDVLK, AMDGPU-PRO

Filed under
Graphics/Benchmarks

The Mesa Radeon Vulkan driver "RADV" with its ACO back-end by default is now winning nearly across the board against not only AMD's AMDVLK Vulkan driver with LLVM back-end but also AMDGPU-PRO with the proprietary shader compiler back-end.

Recently I wrapped up tests on a Radeon RX Vega 56, Radeon RX 5600 XT, Radeon RX 5700 XT, and Radeon VII graphics cards in a few different driver configurations...

Read more

The Latest On The Linux 5.9 Kernel Regression Stemming From Page Lock Fairness

Filed under
Graphics/Benchmarks

Last week we reported on a Linux 5.9 kernel regression following benchmarks from Linux 5.0 to 5.9 and there being a sharp drop with the latest development kernel. That kernel regression was bisected to code introduced by Linus Torvalds at the start of the Linux 5.9 kernel cycle. Unfortunately it's not a trivial problem and one still being analyzed in coming up with a proper solution. So the short story is it's a work-in-progress while this article has some additional insight and benchmarks done over the course of the past few days.

For those wondering about that Linux 5.9 kernel regression spotted for the likes of Apache HTTPD / Nginx / Hackbench, it's still being evaluated. Linus Torvalds knows what's going on but he's still trying to sort out the best way to solve it in addressing those sharp performance drops but also trying to address his motives for writing that patch originally.

Read more

Integrating libcamera into PipeWire

Filed under
Development
Graphics/Benchmarks

Cameras are complex devices which require heavy hardware image processing. The complexity in the userspace camera application development originates with the evolution of complex hardware. libcamera addresses this problem and offers ease of camera application development. libcamera can be described as a full camera stack library in the userspace to provide the feature that do not belong in the kernel.

The core of libcamera framework exposes kernel driver APIs to userspace while the libcamera application layer abstracts the developer from complex hardware usage. libcamera supports multiple video streams from a single device. In real time use cases such as video conferencing, we may need to preview streams with different resolutions than what we tend to stream over the network. It is natural that we would like to display the live streaming simaultaneously while we capture the video with different resolutions. libcamera offers a perfect solution to fulfill these requirements.

Having said that, what if we'd like to stream a camera device data using different applications simaultaneously? Well, PipeWire answers this question. Besides acting as an interface between hardware devices and applications, PipeWire eases the sharing of hardware devices between applications. Moreover, integration of libcamera into PipeWire brings all the abilities offered by libcamera into PipeWire. With this, PipeWire can become a first class user of modern cameras.

Read more

Linux Kernel and Mesa News

Filed under
Graphics/Benchmarks
Linux
  • Kernel Bisecting Has Never Been Faster Than With AMD EPYC + AMD Threadripper

    While Zen 3 is just around the corner, current-generation Ryzen Threadripper and EPYC processors continue to impress particularly for build boxes and tasks like bisecting the Linux kernel's massive codebase.

    Traditionally bisecting every performance regression I come across at Phoronix whether it be in compilers, the Linux kernel, or other code-bases has been prohibitively time consuming (and in turn cooling/power expensive) especially when operating on thin margins as is from the depressed state of the web ad industry, etc, but the incredible speed of the latest EPYC and Threadripper processors -- and having many systems with them around -- has been working wonders for massive time savings and tracking down such issues.

  • Why TensorFlow Lite Has Been Running Slower On Recent Linux Kernels

    The Linux 5.0 to 5.9 kernel benchmarking posted this week showed TensorFlow Lite running slower since the Linux 5.5 kernel... On top of looking at the new Linux 5.9 regressions, I also spent some time bisecting and figuring out what happened for TensorFlow Lite last year that has at least for the system under test caused it to run slower for all the kernel releases this year as shown in the aforelinked article.

    So with the speedy AMD EPYC box and knowing the regression happened between Linux 5.4 and 5.5, it was on to bisecting the TensorFlow Lite performance drop. TensorFlow Lite is the open-source deep learning framework for on-device inference. TensorFlow Lite is an optimized version of TensorFlow catering to mobile and edge devices. This testing is focused on TensorFlow Lite with not yet evaluating the full TensorFlow performance or other deep learning workloads across the tested kernel versions with regards to this regression.

  • Mesa 20.3 Picks Up New Capabilities To Help With Display Presentation Jitter, Stuttering

    Being merged today to Mesa 20.3-devel were some improvements aiming to help with display presentation jitter and hopefully avoid stuttering in the frame-rate. 

    Adam Jackson of Red Hat has been working on supporting GLX_EXT_swap_control/GLX_EXT_swap_control_tear for Mesa's GLX implementation and on the Vulkan side support for VK_PRESENT_MODE_FIFO_RELAXED_KHR. 

  • Mesa Refactors Disk Cache - Working Towards Windows Support

    Mesa 20.3 feature development continues progressing nicely. 

    Timothy Arceri who works extensively on Mesa under his employment by Valve has been refactoring Mesa's disk cache. This revised disk cache that is now merged allows experimenting with different cache layouts for performance reasons. But in the process he also prepped support for Mesa's disk cache to support Microsoft Windows. 

Syndicate content

More in Tux Machines

Android Leftovers

Arm Officially Supports Panfrost Open-Source Mali GPU Driver Development

Most GPU drivers found in Arm processors are known to be closed-source making it difficult and time-consuming to fix some of the bugs since everybody needs to rely on the silicon vendor to fix those for them, and they may even decide a particular bug is not important to them, so you’d be out of luck. So the developer community has long tried to reverse-engineer GPU drivers with projects like Freedreno (Qualcomm Adreno), Etnaviv (Vivante), as well as Lima and Panfrost for Arm Mali GPUs. Several years ago, Arm management was not interested at all collaborating with open-source GPU driver development for Mali GPUs, but as noted by Phoronix, Alyssa Rosenzweig, a graphics software engineer employed by Collabora, explained Panfrost development was now done in partnership with Arm during a talk at the annual X.Org Developers’ Conference (XDC 2020). Read more

Open Up: Open Source Hardware — A Chat with Carl

From a broader lens, to produce “open source hardware” means that we have developed and shared the recipe to create a high-end commercial product that can be learned from, adapted, and used by anyone else. In the same way we’ve stood on the shoulders of the Linux and open source software giants who came before us, we now get to be pioneers in developing open source hardware for those who come next. If you want to learn more how a computer is designed or how something is made, our schematics are the instructions for how to do it. It describes every step of the process, from each piece of the machine and its dimensions, to the type of aluminum used and how to bend it. It’s similar to open source software in that you can learn from the product, adapt it to your needs, and distribute it. The difference is that it requires outside equipment to produce your own version. Open hardware has become more accessible with 3-D printing, but as we found when we were making acrylic prototypes of Thelio, you reach a point where it’s time to work with metal, which presents its own challenges. You have to cut it, bend it, and paint it, all of which requires specific equipment. Read more

Today in Techrights