Language Selection

English French German Italian Portuguese Spanish

Graphics/Benchmarks

Graphics: Iris Xe, NVIDIA, VKD3D-Proton, Gallium/Zink

Filed under
Graphics/Benchmarks

  • Intel Announces Iris Xe Desktop Graphics For OEMs

    Intel today announced Iris Xe (DG1) discrete graphics cards are coming to OEMs with ASUS and Colorful being among the initial partners.

    The initial Iris Xe desktop graphics cards feature 80 execution units and a 30 Watt TDP. This is not the high-end, high performance desktop graphics but seems to largely be the Xe MAX discrete laptop graphics (but with 16 less EUs) now fitted for PCI Express cards for the desktop. The OEM cards are expected to feature 4GB of LPDDR4X memory. Other details are still light.

  • Nvidia Gets Certifiable About Systems

    If the emergence of Nvidia in datacenter compute shows anything, it is the value of controlling the software stack as you come to dominate the compute – and the revenue and profits – in the hardware stack.

    When it comes to AI, the combination of open source frameworks from the wider AI community, which Nvidia contributes to, and closed source libraries and tools that make up the Nvidia GPU Compute software stack that is underpinned by the CUDA environment, gives Nvidia the kind of control over a complete software/hardware stack that we have not seen in the datacenter since the RISC/Unix server days of the dot-com boom and earlier with proprietary systems from IBM, DEC, and HP, as well as IBM mainframes since the dawn of the data processing age.

    There are some differences this time around, and they are significant. The operating system is consequential, of course, but with all AI workloads being deployed on Linux, it really doesn’t matter which one you pick. Linux is about as interchangeable as DRAM memory modules in the server and it really comes down to preferences and a few technical differentiations. And to a certain extent, the X86 server that houses the Nvidia GPUs is fairly interchangeable, too. But fi you want to make GPU compute fluid and easy, then you have to realize that not every can – or wants to – buy an Nvidia DGX-A100 or DGX-2 system. Hyperscalers and cloud builders have their own ODM suppliers, enterprises have their own OEM suppliers, and they want to be able to run the Nvidia AI stack on platforms from their suppliers rather than having to add a new vendor into the mix.

  • NVIDIA 460.39 Linux Driver Brings RTX 30 Laptop Enablement, Improved 5.10+ Kernel Support

    NVIDIA has released 460.39 as their latest stable Linux proprietary graphics driver build.

    With this latest NVIDIA 460 series driver is support now for the RTX 3060 / RTX 3070 / RTX 3080 laptop GPUs as well as for the low-end GeForce GT 1010.

  • VKD3D-Proton begins work to support DirectX Raytracing on Linux | GamingOnLinux

    There's a few mountains that Steam Play Proton still needs to climb over the next few years, to enable more Windows games and more features in those games to work under Linux. One big one is at least in progress.

    Ray Tracing being one of the big things in gaming tech now, thanks to both AMD and NVIDIA having Ray Tracing cards out in the wild. With that, we can expect more games to begin using it.

    Thankfully, VKD3D-Proton, which is the Valve-funded fork of vkd3d to work with Direct3D 12 has a Pull Request open with the start of the work towards supporting Ray Tracing. Keep in mind though, while exciting for Steam Play Proton users, this is far from complete and not enabled directly for games as of yet as stated in the PR "Don't expose any features to app yet, but allow overriding FL to 12.2 for local testing while bringing up DXR.".

  • Mike Blumenkrantz: Samplin

    The goal in this post is to migrate a truckload block of code I wrote to handle sampler updating out of zink and into Gallium, thereby creating several days worth of rebase work for myself but also removing a costly codepath from the driver thread.

    The first step in getting sampler creation to work right in zink is getting Gallium to create samplers with the correct filters in accordance with Chapter 42 of the Vulkan Spec:

    VK_FORMAT_FEATURE_SAMPLED_IMAGE_FILTER_LINEAR_BIT specifies that if VK_FORMAT_FEATURE_SAMPLED_IMAGE_BIT is also set, an image view can be used with a sampler that has either of magFilter or minFilter set to VK_FILTER_LINEAR, or mipmapMode set to VK_SAMPLER_MIPMAP_MODE_LINEAR. If VK_FORMAT_FEATURE_BLIT_SRC_BIT is also set, an image can be used as the srcImage to vkCmdBlitImage2KHR and vkCmdBlitImage with a filter of VK_FILTER_LINEAR. This bit must only be exposed for formats that also support the VK_FORMAT_FEATURE_SAMPLED_IMAGE_BIT or VK_FORMAT_FEATURE_BLIT_SRC_BIT.

    If the format being queried is a depth/stencil format, this bit only specifies that the depth aspect (not the stencil aspect) of an image of this format supports linear filtering, and that linear filtering of the depth aspect is supported whether depth compare is enabled in the sampler or not. If this bit is not present, linear filtering with depth compare disabled is unsupported and linear filtering with depth compare enabled is supported, but may compute the filtered value in an implementation-dependent manner which differs from the normal rules of linear filtering. The resulting value must be in the range [0,1] and should be proportional to, or a weighted average of, the number of comparison passes or failures.

Graphics: GPUOpen, Vulkan, and NVIDIA

Filed under
Graphics/Benchmarks
  • AMD Celebrates Five Years Of GPUOpen

    Today marks five years since AMD began the GPUOpen initiative for providing more open-source Radeon GPU code projects, code samples, and more for better engaging GPU/game developers in the open.

    As any longtime Phoronix reader will know, AMD's open-source Linux driver initiative is going on for more than a decade while the celebration today is just over their GPUOpen initiative turning five years old. The three principles that continue to guide GPUOpen are providing code and documentation to PC developers to exert more control on the GPU, a commitment to open-source software, and a collaborative engagement with the developer community.

  • Vulkan 1.2.168 Released With Two New Extensions

    Today's Vulkan 1.2.168 specification update brings the usual specification corrections/clarifications while also introducing two new KHR extensions.

  • VKD3D-Proton Begins Working On DirectX 12 Ray-Tracing Atop Vulkan

    Those working on VKD3D-Proton as the Direct3D 12 implementation atop the Vulkan API are beginning to work on DirectX Ray-Tracing support but it isn't yet ready for gamers.

    Hans-Kristian Arntzen has opened the initial pull request for enabling ray-tracing extensions with VKD3D-Proton.

  • NVIDIA release the Vulkan Beta Driver 455.50.03, new extensions supported

    Need to be on the bleeding edge of what NVIDIA have to offer? They just released driver version 455.50.03, as part of their Vulkan Beta Driver series. This is actually the second driver released this month, with 455.50.02 appearing on January 19. Here's a look over all that's new in them together.

Patched Linux 5.11 Continues Looking Great For AMD Ryzen/EPYC Performance

Filed under
Graphics/Benchmarks

While the initial AMD Linux 5.11 performance regression written about at the end of last year was of much concern given the performance hits to AMD Zen 2 / Zen 3 processors with the out-of-the-box "Schedutil" governor, with a pending patch the regression is not only addressed but in various workloads we continue seeing better performance than even compared to Linux 5.10. Here is the latest from several more days of extensive performance testing.

Read more

Nouveau X.Org Driver Sees First Release In Two Years

Filed under
Graphics/Benchmarks

Two years and nine patches later, xf86-video-nouveau 1.0.17 is out as the latest X.Org driver update for this open-source NVIDIA driver component.

Like the other DDX drivers with the exception of the generic xf86-video-modesetting driver that is quite common now to those still running on X.Org with the open-source stack, xf86-video-nouveau seldom sees new activity. Since the prior v1.0.16 release two years ago there has been less than a dozen patches for it. The interesting activity happens in DRM/KMS kernel space and an increasing number of users are just relying upon xf86-video-modesetting over these hardware-specific X.Org user-space drivers.

Read more

Graphics: OpenGL, Intel and Zink

Filed under
Graphics/Benchmarks
  • More OpenGL Threading Improvements Land For Mesa 21.1 - Phoronix

    Even in 2021 longtime open-source AMD Mesa driver developer Marek Olšák isn't done optimizing OpenGL for delivering the best possible performance with the Radeon graphics driver. Marek's latest work includes more OpenGL threading enhancements and other work seemingly targeted at SPECViewPerf workloads.

    Marek has spent the past several weeks working to remove the last OpenGL threading synchronization stalls that happen with SPECViewPerf 13. As part of this latest pull request he added support to glthread for executing display lists asynchronously. Plus there are some other OpenGL code improvements too.

  • More Intel Graphics Work In Linux 5.12: Gen7 Improvements, Faster Suspend/Resume

    New feature material for Linux 5.12 continues getting ready ahead of the merge window opening in February to formally kick off the cycle.

    On top of the prior Intel graphics driver improvements queued up in recent weeks to DRM-Next, another batch of Intel updates were sent out this week.

  • Zink OpenGL On Vulkan Now Supports OpenGL 4.2 With Mesa 21.1

    Going back to last summer there have been patches experimentally taking Zink as far as OpenGL 4.6 albeit it's been a lengthy process getting all of the relevant patches upstreamed. Additionally, some patches have required reworking or proper adjustments after going through the conformance test suite to ensure they are up to scratch for merging. Thanks to that ongoing effort by Mike Blumenkrantz working under contract for Valve and the work by Collabora developers, it was a quick jump this month from seeing OpenGL 4.1 to OpenGL 4.2 in mainline.

  • Mike Blumenkrantz: Itshappening.gif

    I meant to blog a couple times this week, but I kept getting sidetracked by various matters. Here’s a very brief recap on what’s happened in zinkland over the past week.

  • Zink OpenGL-On-Vulkan vs. RadeonSI OpenGL Performance As Of January 2021 - Phoronix

    With the Zink OpenGL-on-Vulkan implementation within Mesa on a nice upward trajectory with most recently now having the backing of a Valve contract developer and a focus on getting the backlog of patches to this Gallium3D code upstreamed, here are some fresh benchmarks looking at where the performance currently stands when using Zink atop the RADV Vulkan driver compared to using the native RadeonSI driver with this round of testing from a Radeon RX 5700 XT graphics card.

Zink OpenGL-On-Vulkan vs. RadeonSI OpenGL Performance As Of January 2021

Filed under
Graphics/Benchmarks

With the Zink OpenGL-on-Vulkan implementation within Mesa on a nice upward trajectory with most recently now having the backing of a Valve contract developer and a focus on getting the backlog of patches to this Gallium3D code upstreamed, here are some fresh benchmarks looking at where the performance currently stands when using Zink atop the RADV Vulkan driver compared to using the native RadeonSI driver with this round of testing from a Radeon RX 5700 XT graphics card.

Read more

Linux Kernel Space and Graphics: Linux Memory Management, Dbus-Broker 26, Vulkan Wayland Compositors

Filed under
Graphics/Benchmarks
Linux
  • The Maple Tree, A Modern Data Structure for a Complex Problem

    The Linux Memory Management layer supports the very common technique of virtual memory. Linux splits blocks of virtual memory into areas specified by the c structure vm_area_struct. Each vm_area_struct contain information associated with mapped memory and are used to find the associated pages of memory which contain the actual information. Virtual memory areas (VMAs) could be the contents of a file on disk, the memory that contains the program, or even the memory the program uses during execution. Literally everything that is run on Linux uses vm_area_struct for memory mapping. This vital area of the kernel needs to be quick and avoid contention whenever possible.

  • Dbus-Broker 26 Released For High Performance D-Bus

    With the BUS1 in-kernel IPC not panning out and not seeing any major code work in nearly two years, the user-space based, D-Bus compatible DBus-Broker remains the performant and current option for those looking at something faster and more reliable than D-Bus itself.

  • Vulkan Wayland Compositors Are Nearing Reality - Phoronix

    One of the last pieces of the puzzle for supporting an entirely Vulkan-based Wayland compositor is coming together with a new extension that looks like it will be merged soon and there already being work pending against Sway/WLROOTS to make use of the Vulkan path.

    The VK_EXT_physical_device_drm extension to Vulkan has been in the works for a number of months and is for allowing the mapping of Vulkan physical devices and DRM nodes. VK_EXT_physical_device_drm allows for querying DRM properties for physical devices and in turn matching the with DRM nodes on Linux systems.

  • Mesa's R600 Driver Nears Feature Complete NIR Support For Radeon HD 5000/6000 Series - Phoronix

    For those still making use of pre-GCN AMD graphics cards supported by the R600 Gallium3D driver (namely the Radeon HD 5000/6000 series), the open-source "R600g" Gallium3D driver now has nearly feature complete NIR support.

    Gert Wollny has been near single handedly working on NIR support for the R600g driver to make use of this modern graphics driver intermediate representation as an alternative to the long-standing Gallium3D TGSI IR.

XanMod's Linux 5.10 Kernel Helping Tap Extra Performance With The AMD Ryzen 9 5900X

Filed under
Graphics/Benchmarks

For those wondering how the likes of the XanMod and Liquorix kernel spins are competing these days with the mainline Linux kernel, here are some fresh benchmarks looking at these popular derivatives of the Linux kernel. XanMod in particular atop Ubuntu can easily help squeeze extra performance out of the system as shown by these benchmarks on an AMD Ryzen 9 5900X desktop.

The Liquorix kernel patches the Linux kernel with the MuQSS scheduler, high resolution scheduling, hard kernel preemption, BFQ I/O scheduler, and other tuning to help with the system responsiveness and throughput. More details on the Liquorix changes can be found at Liquorix.net.

XanMod is another popular kernel flavor with enthusiasts. XanMod carries some of the same patches as Liquorix along with various networking patches, the CacULE scheduler, and many other additional patches. More details on it can be found at XanMod.org.

Read more

Kernel and Graphics: exFAT, Linux 5.12, Mesa's Panfrost Gallium3D and Mesa 21.0

Filed under
Graphics/Benchmarks
Linux

  • Linux's exFAT Driver Will Soon Be Able To Delete Big Files Much Faster - Phoronix

    For those making use of Linux's modern exFAT file-system, a significant optimization is on the way for when deleting files with the "dirsync" mount option set. 

    The exFAT file-system driver for Linux 5.12 should be able to delete (large) files much faster when running with the "dirsync" mount option set where the directory updates are done synchronously. 

  •  

  • AMD + Older Intel CPUs To See Much Faster AES-NI XTS Crypto Performance On Linux 5.12 - Phoronix

    AMD processors along with older Intel processors will enjoy much faster AES-NI XTS crypto performance with the Linux 5.12 kernel this spring. 

    Processors supporting AES-NI instructions and that also are subject to Retpolines as part of their Spectre V2 mitigations will enjoy a big speed-up with AES-NI XTS for that next version of the Linux kernel. This includes past and current AMD processors as well as older Intel CPUs, but recent Intel processors do not need the return trampolines and thus not subject to this speed-up as they are not currently handicapped. 

  • Mesa's Lima Driver Finally Implements OpenGL Shader Cache Support - Phoronix

    While Mesa's Panfrost Gallium3D driver has been working out well for modern ARM Mali open-source graphics support, for the old Mali 400/450 series hardware there still is the "Lima" driver within Mesa that doesn't receive too much attention these days (just around 70 commits over the past year) but as its first work of 2021 saw an initial shader cache implementation.

  • Radeon Vulkan "RADV" Driver Saw Many Optimizations This Week For Mesa 21.0 - Phoronix

    Prior to Mesa 21.0 being branched this week in preparations for the quarterly stable Mesa3D release, a number of open-source Radeon Vulkan "RADV" driver optimizations were merged.

Graphics: Latest From Intel and AMD

Filed under
Graphics/Benchmarks
  • Intel "Compute Walker" Support Lands For Xe HP In Linux Drivers - Phoronix

    A new compute code path has been merged into Intel's open-source "ANV" Vulkan and "iris" Gallium3D drivers for the forthcoming Xe HP graphics hardware.

    The initial "COMPUTE_WALKER" code has been in development at Intel for the past 1~2 years while the merge request was just submitted a few days ago to replace the "legacy" compute path from the upcoming Gen12.5 Xe HP hardware. The check is for newer than Gen12 or GEN12HP, so the new compute path will trickle down to lower-tier parts outside of the Xe HP scope moving forward after the existing Gen12 parts like Tiger Lake or the upcoming Rocket Lake still flagged as Gen12 (not 12.5).

  • AMD Sends In More "New Stuff" For Radeon Graphics With Linux 5.12 - Phoronix

    Sent in last week were many AMD graphics driver updates slated for Linux 5.12 including the likes of Radeon RX 6000 series OverDrive support. This week marks another batch of AMDGPU kernel driver changes being submitted to DRM-Next ahead of the Linux 5.12 cycle.

    Among the "new stuff" for AMDGPU in Linux 5.12 include more code warning fixes, enabling of GPU reset and recovery for Navy Flounder yet-to-be-released RDNA 2 SKUs, SMU workload profile fixes for APUs, various display code updates, SR-IOV fixes, various VanGogh APU updates, another Renoir device ID being added, the new bits around Secure Display TA, another Green Sardine device ID being added, and a range of other code improvements.

  • Mesa's RADV ACO Adds Support For Rapid Packed Math - Phoronix

    Hitting the Mesa tree when Mesa 21.0 was being branched (but looks like it will still make it now part of "staging/21.0") is support for AMD's "rapid packed math" with the RADV driver's ACO compiler back-end.

    Rapid Packed Math is the AMD terminology for allowing two FP16 operations within a single FP32 operation. ACO, which is the default shader compiler back-end since last year in Mesa's RADV driver, now supports this functionality for Vec2 16-bit operations.

Syndicate content

More in Tux Machines