• LunarG's Vulkan developer survey results out now - Vulkan also turns 4

  LunarG, the software company that Valve sponsors who work on building out the ecosystem for the Vulkan API recently conducted a Vulkan developer survey with the results out now.

  Before going over the results, just a reminder that Vulkan just recently turned four years old! The 1.0 specification went public on February 16, 2016. Since then, we've seen some pretty amazing things thanks to it. We've had Linux ports that perform really nicely, the mighty DXVK translation layer advanced dramatically, to the vkBasalt post-processing layer and so on—there's been a lot going on. However, as a graphics API do remember it's pretty young and has a long life ahead of it.

  As for the LunarG survey: there were 349 replies to it, and while not a huge amount it gives us an interesting insight into what some developers think and feel about how Vulkan is doing as a whole. Overall, it gives quite a positive picture on the health of Vulkan with over 60% feeling the overall quality of the Vulkan ecosystem as "Good" and almost 20% rating it as "Excellent".

• AMDVLK 2020.Q1.2 Released With Vulkan 1.2 Support

  AMDVLK 2020.Q1.2 is out as the first official AMD open-source Vulkan Linux driver code drop in one month.

  AMDVLK has been off its wagon this quarter with their previous weekly/bi-weekly code drops of AMDVLK but that just means the v2020.Q1.2 is quite a big one. First up, AMDVLK 2020.Q1.2 now is supporting Vulkan 1.2 that debuted back in January and with Mesa's RADV Radeon Vulkan driver already having supported it for weeks.

• Radeon Pro Software for Enterprise 20.Q1.1 for Linux Released

  AMD's Radeon Pro Software for Enterprise 20.Q1.1 Linux driver release was made available this week as their newest quarterly driver installment intended for use with Radeon Pro graphics hardware.

• RADV Vulkan Driver Adds Option For Zeroing Out Video Memory

  New to Mesa 20.1-devel is a new option for the Radeon Vulkan "RADV" driver to enable zeroing out video memory allocations.

  This isn't a new concept with other graphics drivers offering similar functionality for zeroing out the vRAM either for security reasons or working around pesky game/app issues. For example, RadeonSI OpenGL zeros out the vRAM for Rocket League to workaround buggy behavior with that game. But zeroing out the video memory normally isn't done by default for all allocations due to performance reasons.

  With the new flag to zero vRAM allocations for the RADV Vulkan driver it was done by Valve's Samuel Pitoiset. In this case he mentions it's in part for "future work."

• Intel Gen12/Xe Graphics To Support 12-Bit HEVC/VP9 Decode

  We are learning more about the media engine capabilities with the forthcoming Intel "Gen12" (Xe) Tiger Lake graphics.

  The documentation for Intel's open-source media-driver that exposes VA-API capabilities on the Linux desktop was recently updated. That updated Intel VA-API Media Driver points to Intel Gen12 dropping VP8 video capabilities but expanding when it comes to 12-bit codec support.

• Intel Sends Out Latest Patches For Mitigating Graphics Flaw On Ivybridge/Haswell

  It has been one month and a few days since Intel first made public the need for graphics driver patching of Gen 7/7.5 graphics for older Ivybridge / Haswell hardware to fix a graphics hardware flaw. That vulnerability also affected the common Intel Gen9 graphics but there the mitigation was uneventful and quickly merged without causing any performance hit. But for Ivybridge/Haswell one month later the graphics driver mitigation for CVE-2019-14615 is still being addressed.

  This vulnerability is also known as iGPU Leak by the researchers that discovered it but for the Gen7/Gen7.5 protection the mitigation has been particularly problematic. With the initial Gen7/Gen7.5 patches posted in mid-January there was a huge hit to the graphics performance while Intel worked towards no performance loss.

• NVIDIA Posts Firmware Needed For Open-Source GeForce 16 Series Acceleration

  As written about last week, in the works for the Linux 5.7 kernel this spring is open-source NVIDIA "Nouveau" acceleration for the GeForce 16 series. That code is currently sitting in the Nouveau development tree until landing in DRM-Next for Linux 5.7, but NVIDIA has now posted the necessary firmware binaries needed for enabling the hardware acceleration on these Turing GPUs.

• mesa 20.0.0

  Hi list,
  
  I'd like to announce mesa 20.0.0 as available for download immediately. I'm very
  pleased that we could get all of the issues blocking the release nailed down
  quickly and make a release on time for once!
  
  This is a .0 release, and you may want to continue to to track 19.3.x until
  20.0.1 comes out in two weeks. 19.3.5 is planned to be the final 19.3 release
  and is planned for next Wednesday.
  
  Dylan
  
  Shortlog
  ========
  
  Alyssa Rosenzweig (3):
        pan/midgard: Fix missing prefixes
        pan/midgard: Don't crash with constants on unknown ops
        pan/midgard: Use fprintf instead of printf for constants
  
  Danylo Piliaiev (1):
        st/nir: Unify inputs_read/outputs_written before serializing NIR
  
  Dylan Baker (6):
        .pick_status.json: Update to 2a98cf3b2ecea43cea148df7f77d2abadfd1c9db
        .pick_status.json: Update to 946eacbafb47c8b94d47e7c9d2a8b02fff5a22fa
        .pick_status.json: Update to bee5c9b0dc13dbae0ccf124124eaccebf7f2a435
        Docs: Add 20.0.0 release notes
        docs: Empty new_features.txt
        VERSION: bump for 20.0.0 release
  
  Erik Faye-Lund (1):
        Revert "nir: Add a couple trivial abs optimizations"
  
  Francisco Jerez (6):
        intel/fs/cse: Make HALT instruction act as CSE barrier.
        intel/fs/gen7: Fix fs_inst::flags_written() for SHADER_OPCODE_FIND_LIVE_CHANNEL.
        intel/fs: Add virtual instruction to load mask of live channels into flag register.
        intel/fs/gen12: Workaround unwanted SEND execution due to broken NoMask control flow.
        intel/fs/gen12: Fixup/simplify SWSB annotations of SIMD32 scratch writes.
        intel/fs/gen12: Workaround data coherency issues due to broken NoMask control flow.
  
  Krzysztof Raszkowski (1):
        gallium/swr: simplify environmental variabled expansion code
  
  Marek Olšák (1):
        radeonsi: don't wait for shader compilation to finish when destroying a context
  
  Mathias Fröhlich (1):
        egl: Implement getImage/putImage on pbuffer swrast.
  
  Peng Huang (1):
        radeonsi: make si_fence_server_signal flush pipe without work
  
  Pierre-Eric Pelloux-Prayer (1):
        radeonsi/ngg: add VGT_FLUSH when enabling fast launch
  
  Tapani Pälli (2):
        glsl: fix a memory leak with resource_set
        iris: fix aux buf map failure in 32bits app on Android
  
  Thong Thai (1):
        Revert "st/va: Convert interlaced NV12 to progressive"
  
  Timothy Arceri (1):
        glsl: fix gl_nir_set_uniform_initializers() for image arrays
  
  luc (1):
        zink: confused compilation macro usage for zink in target helpers.
  
  
  
  git tag: mesa-20.0.0
  

• Mesa 20.0 Released With Big Improvements For Intel, AMD Radeon Vulkan/OpenGL

  The Mesa 20.0 release switches to the new Intel OpenGL driver default, Vulkan 1.2 support for both AMD Radeon and Intel drivers, the RadeonSI OpenGL driver now has GL 4.6 compliance as part of switching to NIR, the Valve-backed ACO code-path for RADV is in much better shape, and many other improvements. See our Mesa 20.0 feature overview to learn about this big update.

• Mesa 20.0 Is Imminent With New Intel OpenGL Default, Intel + RADV Vulkan 1.2, OpenGL 4.6 For RadeonSI

  With the release of Mesa 20.0 being imminent, here is a look at all of the new features for this first quarter update to the Mesa 3D stack for open-source OpenGL/Vulkan drivers.
  Highlights of the soon-to-be-out Mesa 20.0 are outlined below. Mesa 20.0 will be out as soon as today / this week unless delays happen over lingering bugs.
  - This is the first Mesa release where for those with Broadwell (Gen8) Intel graphics or newer the Intel Gallium3D driver is the new default for OpenGL support. This Intel Gallium3D driver is faster and in better shape than the i965 classic driver. That older OpenGL driver will stick around for supporting Haswell graphics and prior generations.

• RADV Driver Adds VK_EXT_line_rasterization In Preparing For Eventual Vulkan CAD Apps

  Added to the Vulkan API last summer was VK_EXT_line_rasterization for line rasterization like employed by CAD applications. The open-source Mesa Radeon Vulkan "RADV" driver is now supporting this extension.

When taking the geometric mean of all these benchmark results, the Windows 10 Professional performance was the same as Windows 10 Enterprise for this Threadripper 3970X testing, unlike the Enterprise advantage we've seen on the larger Threadripper 3990X. The slowest of the eight Linux distributions tested was the Ubuntu 20.04 development snapshot, but that still came out to be 9.5% faster than Windows 10. The fastest Linux distribution was Clear Linux on the Threadripper 3970X with a 19% over Windows in these cross-platform benchmarks. Following Clear Linux with a strong showing was the new rolling-release CentOS Stream.