Windows ML through WinML APIs gives developers tools to use AI in their apps. By training models in the cloud, dev’s can export the AI to their apps in ONNX format through Visual Studio. The platform will use existing GPU hardware on a device to deal with AI workloads evaluations. As for the Intel VPU, it has been custom created to accelerate machine learning workloads on devices. Movidius Myriad X is the first system-on-a-chip in the world to provide a dedicated Neural Compute Engine that helps boost deep learning inferences. Integrated on-chip, the Neural Compute Engine can run networks at low power but at high speeds. The VPU can do this without taking accuracy from the AI, allowing devices to react in real-time. Intel and Microsoft collaborated to allow the Movidius Myriad X to support the Windows ML platform. “Intel Movidius VPU technology will deliver increasingly sophisticated AI experiences for the hundreds of millions of Microsoft users worldwide,” said Intel’s Remi El-Ouazzane, Intel vice president and general manager of Intel Movidius. “This is just the latest example of how Intel is accelerating the promise of bringing AI from the data center to edge devices through our high-performance, low-power vision processor technology.”
Device Development
The clear positive of Windows ML and Intel Myriad X integration is devices arriving with dedicated AI. None of the hardware is available yet, but Windows AI will help to change that. OEMs already partnered with Microsoft can leverage the platform to create devices with AI integration. “We’re excited to work closely with Intel to enable developers around the world to build engaging and magical AI-powered experiences using Windows ML and the Intel Movidius VPU,” said Kevin Gallo, corporate vice president, Windows Developer Platform, Microsoft.
