Khronos royalty-free open standards for 3D graphics, Virtual and Augmented Reality, Parallel Computing, Machine Learning, and Vision Processing

Khronos has released OpenVX 1.3.2, an update to the open, royalty-free standard that uses a high-level graph-based framework to deliver portable, efficient vision processing. This focused update strengthens error handling, improves conformance, and makes OpenVX easier for vendors to implement and for developers to debug — laying the groundwork for the next major release, OpenVX 2.0.

SIGGRAPH 2026 in Los Angeles delivered a clear snapshot of the future of graphics and spatial computing—and Khronos was at the center of it. The organization presented more than 20 sessions spanning hands-on courses and in-depth Birds of a Feather (BOF) discussions on neural graphics, XR, 3D assets, rendering, and real-time shading. Wednesday, July 22 served as the dedicated Khronos Developer Day, with status updates from the Vulkan, glTF, WebGL/WebGPU, OpenXR, ANARI, and Slang Working Groups, plus the Machine Learning Council, followed by a networking reception for members and guests. Click through for a full recap of activities. 

3D Formats member, Adam Morris (Bentley/Cesium) sits down with host Muthu Kumar to break down how 3D Gaussian Splatting works, why it outperforms traditional 3D models in complex visual inspection tasks, and how ‪Khronos is standardizing the technology via a new glTF extension.

The release candidate of PoCL 7.2 is now available for testing of this Portable Computing Language implementation of OpenCL that allows for CPU-based execution or via alternative LLVM back-ends allows for OpenCL on Level Zero, NVIDIA GPUs, and even remote devices/hardware.

Highlights of SDK 1.4.357:

• Major KosmicKrisp performance gains (up to ~2.35× faster) + full Vulkan 1.4 exposure on Apple platforms
• 13 new extensions
• Scoped GPU-AV + new GPU Dump tool in the Validation Layers
• Available now for Linux, Windows & macOS

ROCm has compiled GPU code ahead of time (AOT), once per target architecture, since its first release. CPU software shifted to runtime just-in-time (JIT) compilation in the 1990s — Java HotSpot, V8, LuaJIT, .NET RyuJIT, PyPy — once the matrix of target × workload × deployment became too large to enumerate at build time. With its adoption of SPIR-V, ROCm now brings the compile-once, specialize-on-device model to AMD GPUs. 

This post covers what SPIR-V is, why AMD adopted it, the compilation model and its trade-offs, and a reproducible single-kernel HIP example with measurements on MI350X (CDNA4), spot-checked on RX 9070 XT (RDNA4). A FAQ and status section at the end cover common questions and what is shipping, in flight, and not yet solved.