There are a few projects on the web that are looking at how to run WebAssemby in Kernel space, removing the need for ‘protection rings’ and context switching, allowing applications to run faster. In this bog post the wasmjit project announces that they have managed to get nginx (the web server) compiled to wasm, using emscripten, running on their platform. As nginx has significant IO operations, this is a massive step forwards.
Even more cool, wasmer, another kernel space runtime, also announced that they can run nginx just a day beforehand. The race to make WebAssembly a universal runtime is on!
At this years Google I/O conference they announced TensorFlow.js which allows you to run machine learning models, and perform training, from within the browser. The newly release ONNX.js framework, from Microsoft, is another framework for running neural networks in the browser. With ONNX.js you can run the models on the GPU, using WebGL, or the CPU, using WebAssembly and Web Workers.
You know a technology has become mainstream when it has an ‘In Action’ book! This book will be available in Spring 2019, however, you can access it early via the Manning Early Access Programme. There is a 50% discount if you use the code mlgallant.
And Finally …