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 this, Patrick Ryan from Microsoft has created a walk through following several paths for authoring content in the glTF format as well as outlining specific settings to maximize your success. Patrick touches on both free and commercial software packages to ensure everyone has a path into glTF. Let's get going... check out this glTF how-to.
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.
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. Learn more about NNEF and the upcoming released in the Khronos blog post.
PerfDoc is a Vulkan layer which aims to validate applications against the Mali Application Developer Best Practices Guide. Just like the LunarG validation layers, this layer tracks your application and attempts to find API usage which is discouraged. PerfDoc focuses on checks which can be done up-front, and checks which can portably run on all platforms which support Vulkan. The intended use of PerfDoc is to be used during development to catch potential performance issues early. The layer will run on any Vulkan implementation, so Mali-related optimizations can be found even when doing bringup on desktop platforms. Just like Vulkan validation layers, errors are reported either through VK_EXT_debug_report to the application as callbacks, or via console/logcat if enabled. Dynamic checking (i.e. profiling) of how an application is behaving in run-time is not currently in the scope of PerfDoc. Some heuristics in PerfDoc are based on "arbitrary limits" in case where there is no obvious limit to use. These values can be tweaked later via config files if needed. Some checks which are CPU intensive (index scanning for example), can also be disabled by the config file. Please visit the GitHub repository for PerfDoc.