RenderDoc 1.5 has been released. Changes include: SPIR-V reflection and disassembly has been refactored to be more reliable and is based on the publicly available grammar json; Vulkan has a new replay-time optimisation which takes advantage of the above replay options dialog; OpenGL has a low-memory optimisation to defer copying initial contents of textures and buffers that are rarely modified; Support for twenty-one (21) Vulkan extensions have been added; Support for fifteen (15) OpenGL extensions have been added along with whitelisting of some OpenGL ES extensions that were already supported. A complete list of changes and improvements is available in the Latest release notes.
Basemark GPU 1.2 adds iOS and macOS to our list of already supported operating systems: Android, Windows, and Linux. We include OpenGL ES 3.1, OpenGL 4.5, Vulkan 1.0, DirectX 12 and Metal. With the new features in Basemark GPU 1.2, you will be able to analyze and compare the 3D graphics performance of smartphones, tablets, notebooks, Windows and Linux desktop PCs, etc.
Last week saw the release of Vulkan 1.1.123. The release sees four issues from GitHub addressed and several Khronos internal issues fixed. Two new extensions have also be included: VK_KHR_shader_subgroup_extended_type: enables the Non Uniform Group Operations in SPIR-V to support 8-bit integer, 16-bit integer, 64-bit integer, 16-bit floating-point, and vectors of these types, and VK_GOOGLE_user_type: allows use of the SPV_GOOGLE_user_type extension in SPIR-V shader modules.
NVIDIA Nsight Systems 2019.5 is now available. This release release offers several improvements to refine the user experience with CLI sessions for simultaneous usage of more commands, improved GUI timeline zooming levels of detail, enhanced Vulkan API coloring, and Linux GPU context switch trace.
Andrew Richards, CEO and Co-founder of Codeplay Software, presents the “Can We Have Both Safety and Performance in AI for Autonomous Vehicles?” tutorial at the May 2019 Embedded Vision Summit. Andrews presentation includes discussion on how SYCL, OpenCL and Vulkan can play into Safety and Performance in vehicles.
A new release of MoltenVK is now available that is updated against Vulkan 1.1.121 for allowing much of this graphics/compute API to function under Apple’s macOS and iOS platforms by remapping the Vulkan calls to Metal. More details are available on Phoronix and the MoltenVK repository on GitHub.
Intel’s “ANV” Vulkan Linux driver has now hooked into the DriConf infrastructure for handling driver tunables and workarounds. As of Friday afternoon, the Intel Vulkan driver hooks into DriConf. That was followed up by adding a toggle for vk_x11_override_min_image_count behavior. That was followed by adding a generic override for when running gfxbench. Learn more about DriConf and this change.
Geekbench 5, the latest major upgrade to Primate Labs’ cross-platform benchmark, is now available for download. The Compute Benchmark now supports Vulkan, along with CUDA, Metal, and OpenCL. Vulkan Compute Benchmarks are available on Android, Windows, and Linux.
Vulkan 1.1.122 has been released. This update sees many internal issues resolved plus one new extension from a public issue. VK_IMG_format_pvrtc allows an application to capture performance data to be interpreted by an external application or library.
The FFmpeg library up to this point has supported AMD’s Advanced Media Framework (AMF) library just on Windows for H.264/HEVC encoding on GPUs. The Windows code-path makes use of DirectX while now AMD AMF support for Linux via Vulkan is now exposed by the latest FFmpeg code.
bgfx popular open source cross-platform rendering library adds Vulkan renderer backend. Vulkan renderer backend will be bgfx’ default renderer on Linux, and it can be used on all supported platforms including MacOS via MoltenVK emulation.
Another blog explaining Vulkan synchronization has just been posted by @Themaister. Synchronization in Vulkan is a large hurdle to overcome when learning the API, and rather than mechanically explaining how it works, the goal in this blog is to instil a mental model in the reader. Despite its reputation for maddening complexity, it is actually understandable and quite logical once you get over the initial hurdles.