Language Selection

English French German Italian Portuguese Spanish

Graphics: Gallium3D, Mesa, RADV, ACO

Filed under
Graphics/Benchmarks
  • Intel's Gallium3D OpenGL Driver Taps Another Optimization - ~32% For GFXBench

    Intel's new OpenGL Linux driver, their Gallium3D-based "Iris" implementation that is aiming to be the default before year's end, continues making striking progress.

    Just this past week when testing the very latest Mesa code for this Intel Gallium3D driver I was quite impressed with it near universally being faster than their existing "i965" Mesa driver. For some OpenGL workloads, this Gallium3D driver is significantly faster than the driver it's set to replace for Broadwell "Gen 8" graphics and newer.

  • Mesa's Disk Cache Code Now Better Caters To 4+ Core Systems

    Most Linux gamers these days should be running at least quad-core systems so Mesa 19.3 has been updated to reflect that reality with the number of CPU threads used by their disk cache.

  • Performance-Boosting DFSM Support Flipped On & Off For RADV Vulkan Driver

    Back in July of last year the RADV Vulkan driver enabled primitive binning and DFSM for this open-source Radeon Vulkan driver. Well, it thought it enabled DFSM support and paired with the binning did yield a minor performance benefit at the time for Raven Ridge APUs. But now it turns out the DFSM support wasn't properly wired up and is now addressed but is currently introducing a performance regression.

    RADV developer Bas Nieuwenhuizen added the actual DFSM (Deterministic Finite State Machine) support and mirrors the behavior of the RadeonSI OpenGL driver. With the DFSM support he found that it doubles the fill-rate of one of his test samples from around 16 to 32 pixels/cycles for Raven Ridge.

  • The Valve-funded shader compiler 'ACO' is being queued up for inclusion in Mesa directly (updated: merged)

    Back in early July, Valve announced their work on a new AMD GPU shader compiler for Mesa named ACO and now they're trying to get it pulled into Mesa directly.

    Their main aims with ACO were to get the "best-possible code generation for game shaders, and fastest-possible compilation speed" and to replace the currently used shader compiler from the massive LLVM project. It has certainly seemed promising, improving both shader compile time resulting in less stuttering and so helping to improve overall FPS and smoothness in Linux games when played on supported AMD GPUs.

  • Valve's ACO Shader Compiler For The Mesa Radeon Vulkan Driver Just Landed

    It was just two days ago that Valve's performance-focused "ACO" shader compiler was submitted for review to be included in Mesa for the "RADV" Radeon Vulkan driver. Just minutes ago that new shader compiler back-end was merged for Mesa 19.3.

    ACO, short for the AMD COmpiler, is the effort led by Valve at creating a more performant and optimized shader compiler for the Radeon Linux graphics driver. Besides trying to generate the fastest shaders, ACO also aims to provide speedy shader compilation too, as an alternative to the AMDGPU LLVM shader compiler back-end. Initially ACO is for the RADV Vulkan driver but it may be brought to the RadeonSI OpenGL driver in the future. At the moment ACO is in good shape for Volcanic Islands through Vega while the Navi shader support is in primitive form.

More in Tux Machines

Graphics and Standards

  • SHADERed 1.2.3 Released With Support For 3D Textures & Audio Shaders

    SHADERed is the open-source, cross-platform project for creating and testing HLSL/GLSL shaders. While a version number of 1.2.3 may not seem like a big update, some notable additions can be found within this new SHADERed release.

  • Vulkan 1.1.125 Released With SPIR-V 1.4 Support

    Succeeding Vulkan 1.1.124 one week later is now Vulkan 1.1.125 with a lone new extension. Vulkan 1.1.125 has its usual clarifications and corrections to this graphics API specification. Meanwhile the new extension introduced in the overnight v1.1.125 release is VK_KHR_spirv_1_4.

  • Making Movies Accessible for Everyone

    For the first time, people who are deaf or hard of hearing will be able to enjoy the Nairobi leg of the Human Rights Watch Film Festival, opening on October 15.

New Videos: A Look at Xubuntu 19.10 and Crunchbang++ 10

Debian and Ubuntu Leftovers

  • Ritesh Raj Sarraf: Bpfcc New Release

    bpfcc version 0.11.0 has been uploaded to Debian Unstable and should be accessible in the repositories by now. After the 0.8.0 release, this has been the next one uploaded to Debian.

  • Utkarsh Gupta: Joining Debian LTS!

    Back during the good days of DebConf19, I finally got a chance to meet Holger! As amazing and inspiring a person he is, it was an absolute pleasure meeting him and also, I got a chance to talk about Debian LTS in more detail. [...] I had almost no idea what to do next, so the next month I stayed silent, observing the workflow as people kept committing and announcing updates. And finally in September, I started triaging and fixing the CVEs for Jessie and Stretch (mostly the former). Thanks to Abhijith who explained the basics of what DLA is and how do we go about fixing bugs and then announcing them. With that, I could fix a couple of CVEs and thanks to Holger (again) for reviewing and sponsoring the uploads! :D

  • Ubucon Europe 2019 in local media

    News from the new Ubuntu distribution, the exploration of the several platforms and many “how to”, rule the 4-days agenda where the open source and open technologies are in the air. The Olga Cadaval Cultural centre in Sintra, is the main stage of a busy agenda filled with several talks and more technical sessions, but at Ubucon Europe there’s also room for networking and cultural visits, a curious fusion between spaces full of history, like the Pena Palace or the Quinta da Regaleira, and one of the youngest “players” in the world of software. For 4 days, the international Ubuntu Community gathers in Sintra for an event open to everyone, where the open source principles and open technology are dominating. The Ubucon Europe Conference begun Thursday, October 10th, and extends until Sunday, October 13th, keeping an open doors policy to everyone who wants to Afterall, what is the importance of Ubucon? The number of participants, which should be around 150, doesn’t tell the whole story of what you can learn during these days, as the SAPO TEK had the opportunity to check this morning. Organised by the Ubuntu Portugal Community, with the National Association for Open Software, the Ubuntu Europe Federation and the Sintra Municipality, the conference brings to Portugal some of the biggest open source specialists and shows that Ubuntu is indeed alive, even if not yet known by most people, and still far from the “world domain” aspired by some.

Devices/Embedded: Win Enterprises and Raspberry Pi 4

  • Win Enterprises unveils Atom-based LAN gateway and compact SBC

    Win Enterprises unveiled a fanless “PL-82000” networking gateway with 6x GbE and 2x SFP ports based on an Atom C3000. It also launched a Raspberry Pi sized “MB-5000” SBC that runs Ubuntu or Win 10 on Intel Apollo Lake. We tend to forget Win Enterprises because as its name suggests, the company typically sticks to Windows-supported products. Yet, they have increasingly produced barebones products without listed OS support, such as the new PL-82000 networking appliance, as well as Linux supported systems such as the MB-5000 SBC announced back in June. (In 2017, we covered an Intel Bay Trail based MB-80580 SBC and Win IoT-380 Gateway with Linux support.)

  • Raspberry Pi 4 PCI Express: It actually works! USB3, SATA… GPUs?

    Recently, Tomasz Mloduchowski posted a popular article on his blog detailing the steps he undertook to get access to the hidden PCIe interface of Raspberry Pi 4: the first Raspberry Pi to include PCIe in its design. After seeing his post, and realizing I was meaning to go buy a Raspberry Pi 4, it just seemed natural to try and replicate his results in the hope of taking it a bit further. I am known for Raspberry Pi Butchery, after all.

  • Raspberry Pi 4 B+ - PCI Express

    Why did I do it? Because I wanted to see if it can be done. Because Raspberry Pi 4 might be the cheapest device that is PCIe capable after a relatively minor modification (if I didn't lift the capacitors when desoldering the VL805, this is literally 12 soldering points). That, in turn, can be quite handy for developing own PCIe cores for various FPGA based experiments.

    I'm sharing it to allow people to learn from this - and to dispel the myth that PCIe is somehow out of reach of hobbyists due to some concerns over signal integrity or complexities. Stay tuned for more Pi4/PCIe experimentation!