Standards make life easier, and we depend on them for more than we might realize — from knowing exactly how to drive any car, to knowing how to get hot or cold water from a faucet. When they fail us, the outcome can be comical or disastrous: non-standard plumbing, for instance, can result in an unexpected cold shower or a nasty scald. We need standards, and the entire computing world is built on them.
NNEF and ONNX are two similar open formats to represent and interchange neural networks among deep learning frameworks and inference engines. At the core, both formats are based on a collection of often used operations from which networks can be built. Because of the similar goals of ONNX and NNEF, we often get asked for insights into what the differences are between the two. Although Khronos has not been involved in the detailed design principles of ONNX, in this post we explain how we see the differences according to our understanding of the two projects. We welcome constructive discussion as the industry explores the need for neural network exchange and hope this post may be a constructive start to that conversation.
Authoring content for a new file format can be exciting, liberating, and at the same time scary. To be the most efficient and avoid frustration, it helps to understand the format's requirements. To help achieve that, I am going to walk through several paths for authoring content in the glTF format as well as outline specific settings to maximize your success. I will touch on both free and commercial software packages to ensure everyone has a path into glTF, but first let's outline a few important concepts.
Every year in December, millions of people get in the holiday spirit with NORAD Tracks Santa, the website that lets you track Santa’s magical midnight voyage through the sky on Christmas Eve. Part of what makes the NORAD Tracks Santa website possible are Khronos standards WebGL and glTF. Today, over 22 million people follow Santa’s journey on a 3D map built with Cesium. Before gITF and WebGL, Mr. Claus’s delivery route was much harder to trace.
Previous blog posts have stressed that the deployment process of neural networks to inference engines is becoming fragmented. An accepted standard can facilitate the industrial use of artificial intelligence by creating mutual compatibility between deep-learning frameworks and inference engines. The Neural Network Exchange Format (NNEF) is the Khronos Group’s solution to this problem.
There is a wide range of open-source deep learning training networks available today offering researchers and designers plenty of choice when they are setting up their project. Caffe, Tensorflow, Chainer, Theano, Caffe2, the list goes on and is getting longer all the time. This diversity is great for encouraging innovation, as the different approaches taken by the various frameworks make it possible to access a very wide range of capabilities, and, of course, to add functionality that’s then given back to the community. This helps to drive the virtuous cycle of innovation.
As part of the ongoing work to ensure glTF meets the needs of the developer community the Khronos™ 3D Formats working group is working on a new glTF compression extension to greatly improve transmission efficiency of texture assets while providing efficient, cross-platform transcoding into a wide range of GPU hardware-accelerated texture formats.
The Khronos™ Group is about to release a new standard method of moving trained neural networks among frameworks, and between frameworks and inference engines. The new standard is the Neural Network Exchange Format (NNEF™); it has been in design for over a year and will be available to the public by the end of 2017.
Supercomputing is underway in Denver, Colorado! The 30th annual conference is this week from November 13 through 16, and explores high-performance computing, networking, storage, and analysis. Khronos will be at the show to demonstrate how Khronos standards, especially SYCL, are playing their parts in HPC today.
Bringing 25 graphics standards to various industries is a collaborative effort, run worldwide across Khronos’ over 100 member companies. We come together three times each year for face-to-face meetings, which present the rare opportunity for all of our active members, Working Groups, and personnel to discuss the goals for the upcoming year to drive our standards forward. These events also present the opportunity to give Khronie Awards to those whose contributions made significant impact to Khronos, our standards, and our mission.
After announcing the OpenXR working group this year at GDC, Khronos members are working hard to bring the standard to life, while also educating the industry on the importance of this upcoming technology. This week, VRDC brings together creators and key influencers in the virtual and augmented reality industries for the latest technology demos and discussions on best practices. During the show in San Francisco, some of our members will join a panel to discuss OpenXR and fragmentation in the AR/VR/MR industry.
Khronos is responsible for bringing more than Vulkan, OpenGL, and WebGL to the world. The Khronos Group and its members have created over 25 standards to date, a list that continually grows as needs for new standards arise from growing industries.
Recently I asked the community for beginner-friendly resources on Vulkan, and I compiled a list of them that you can find below. For the beginners reading this, Vulkan is a new graphics API-- in other words, a way to communicate with your GPU and make it do things. It's managed by the Khronos Group, which means it's under multi-company governance - being managed by the industry for the industry. Anyone who wants to do work on GPUs (not restricted to graphics programmers!) should at least have a high level knowledge of what it is.
In early August the team was at SIGGRAPH in Los Angeles, where we celebrated OpenGL’s 25th anniversary at the BOF Blitz Party. We also announced a new website, as well as OpenGL 4.6, a growing glTF ecosystem, and the Vulkan Portability Initiative.
Following the successful release of glTF 2.0, Khronos’ 3D asset transmission format continues to gain strong industry momentum, including support from Microsoft and Google. Today, Khronos has revealed that Google has released a new draft extension to use Draco geometry compression to make glTF files significantly more compact, that the Blender Exporter for glTF 2.0 is now complete and in beta, as well as Microsoft continuing to use glTF 2.0 to bring 3D capabilities to Paint 3D and Microsoft office. So – what is glTF? And why is it gaining so much adoption throughout the industry?