Windows 10 Spring Creators Update To Bolster Machine-learning Capabilities

The next major iteration of Windows 10 is due soon, either this month or early next, and based on all that we know so far, it’s going to be worth the upgrade. Called the Spring Creators Update (AKA: Redstone 4), and versioned 1803, one of the biggest (but probably most subtle) things of interest is the deployment of some groundwork to bolster Windows’ machine-learning capabilities.

Machine-learning, deep-learning, and artificial intelligence have made headlines a countless number of times these past few years, and while they’re often referencing the kind of work supercomputers get done, the benefits of the technologies are for everyone. That includes anyone with a phone. Maybe not right now, but definitely in the future, thanks in part to efforts like those from Arm, which announced dedicated ML and object detection processors last month.

A cat inspecting some higher-end deep-learning devices for Windows

Given the prevalence of machine-learning and AI in all of our devices, Microsoft wanted to make sure its new framework would complement the wealth of devices out there. That includes not just desktops and notebooks, but more unique products like HoloLens and IoT devices.

On the desktop, the ML abilities will be able to latch onto certain instruction sets, including AVX512, which is so far found on only the highest-end chips on the market (eg: Intel’s Core i9-7980XE). Hardware dedicated to training or inference will also be supported, but even if you don’t have a piece of hardware you think would be suitable for machine-learning, you may be surprised. According to Microsoft, 80% of Windows 10 users have a GPU capable of delivering suitable performance.

The models this framework will utilize are based on the ONNX format, which is jointly developed as open source by Microsoft, Amazon, and Facebook, and is supported by NVIDIA, AMD, Intel, and Qualcomm. So, all things considered, this sounds like a great advancement overall, and one that’s not too surprising given the explosion of machine-learning and artificial intelligence in recent years.