Microsoft Build 2024 – Unleashing the potential of AI with Windows on Snapdragon

-

 



Devang Aggarwal, Senior Product Manager, Qualcomm Technologies, Inc., discussed the Qualcomm AI Hub – a tool to help developers optimize, validate, and deploy their AI models to Qualcomm Technologies platforms using popular runtimes like ONNX RT and DirectML. With Qualcomm AI Hub, you can deploy in minutes while seeing significant performance gains.

During this session, we announced support for Snapdragon X Series platforms. Additionally, the team announced the new bring-your-own-model capability of the Qualcomm AI Hub. Now, developers can upload any AI model to the Qualcomm AI Hub, in addition to using the available prebuilt models. This means developers can take advantage of the tool for their existing development models, or other models they come across that they’d like to experiment with.

In this session, Devang was joined by George Wu, Principal Software Engineer Manager at Microsoft for ONNX RT who discussed Microsoft’s integration of ONNX RT as a runtime option in the Qualcomm AI Hub, and gave a live demo.

To round things out with a real-world example, Devang and George were joined by Jon Campbell, Director of Engineering of Cephable, an assistive technology company. Jon showed how their technology enables users to control PowerPoint slides via head movements captured through a live camera feed which uses computer vision algorithms running on the Snapdragon X Plus NPU. John highlighted how processing on the NPU frees up the CPU and GPU, ensuring privacy and assistive capabilities anywhere users want to work.

On the development side, the Qualcomm AI Hub enabled Jon’s team to convert an existing TFLite model to ONNX RT. The Qualcomm AI Hub also gave them confidence that their model would work correctly and perform well on the target platform — aspects traditionally determined later in a development lifecycle. The Qualcomm AI Hub allowed their two developers to build their AI capabilities in about a week ­– much shorter than an estimated 30-day outlook.



Ankit Kapoor, Product Manager, Qualcomm Technologies, Inc., dived deeper into AI development at the edge and how to take advantage of the Snapdragon X series 40+ TOPS of NPU performance.

Ankit began with the architecture of the NPU and how it efficiently handles scalar, vector, and matrix math for ML models. He then looked at how the Qualcomm AI Stack supports different frameworks and runtimes, including ONNX RT and DirectML, along with the different levels of SDKs available for development. Developer resources include Jupyter Notebooks, sample code, whitepapers, add-ons, and documentation.

Ankit also gave his vision for hybrid AI – the distribution of AI between the cloud and the edge. He also discussed the collaboration of Microsoft and Qualcomm Technologies on Windows Studio Effects to support next-gen AI features such as Live Captions, Paint live Co-creator, and automatic super resolution.



In this interview, Dileep Karpur, Staff Product Manager, Qualcomm Technologies, Inc., discussed how over 300 of the top ISVs are now building or porting to native Windows on Snapdragon. These projects are realizing the benefits that Qualcomm Technologies’ advances in mobile have brought to the PC, namely battery efficiency and mobile performance. This was backed by real-world testimony from Carl Woodward, Principal Engineer from security juggernaut McAfee, whose team uses Qualcomm Technologies solutions to detect deep fakes at the edge, while preserving what users care about most: low-latency performance, battery life, and privacy.


Dileep highlighted the Qualcomm Developer Portal, qualcomm.com/developer as the starting point where developers can learn more. He then showed off the Snapdragon Developer Kit for Windows, which Qualcomm Technologies announced at the event. The dev kit is a cost-effective platform that lets developers build, test, and optimize their Windows on Snapdragon apps for Snapdragon X Series platforms.

Additional demos:

Windows on Snapdragon and Qualcomm AI Hub Demos at Build

Ready to Build your Next Windows Project on Co-Pilot+ PCs?

To get started with your Windows and AI development journey, check out the following resources to learn more:

  • Windows on Snapdragon Developer Portal: Documentation, tools, and other learning resources to build or port existing Windows apps to Windows on Snapdragon.

  • Qualcomm AI Hub: Sign up for free to convert, optimize, and deploy ML models for your next Windows app running on Snapdragon X Series platforms.

  • Qualcomm AI Stack: Documentation, tools, and other learning resources to integrate Qualcomm Technologies' AI solutions into your apps.

Comments

Post a Comment

Popular Posts

Iphone 16e Launch Event: A Sign of Things To Come

-
Image
  Apple’s toned-down announcement of the iPhone 16e marks a strategic shift in how the company approaches product differentiation within its lineup. Unlike the flagship iPhone 16 and 16 Pro models, which will undoubtedly receive a major marketing push later in the year, the 16e appears to be a more pragmatic addition aimed at a specific audience rather than a mass-market spectacle. Pragmatism-Infused Marketing Strategy Apple has positioned the iPhone 16e as a lower-cost alternative that retains core iPhone capabilities without the premium features of the higher-end models. This approach aligns with Apple’s ongoing segmenting user base strategy: high-end users gravitate toward Pro models. How the iPhone 16e Online Launch Flowed The prerecorded iPhone 16e unveiling film follows Apple’s traditional style, making a mid-cycle model feel noteworthy. Still, it was shorter than a typical major Apple smartphone launch, less than 13 minutes. Of course, Apple launched the iPhone 16e via a tra...
Read More

Nvidia Blackwell is a high-performance processor that generates significant heat, requiring advanced cooling solutions.

-
Image
  Nvidia Blackwell is a high-performance processor that generates significant heat, requiring advanced cooling solutions. (Image Credit: Nvidia) Nvidia has faced scrutiny this month because some servers with a whopping 72 Blackwell processors were overheating. The issue arose because some initial OEM deployments were not properly water-cooled, which Lenovo aggressively identified and mitigated with its Neptune warm water-cooling solutions. As AI advances, we’ll need more highly dense, incredibly powerful AI processors, which suggests that air cooling in server rooms may become obsolete. Let’s talk about Blackwell, water cooling, and why Lenovo’s Neptune solution stands out at the moment. We’ll close with my Product of the Week: Microsoft’s Windows 365 Link, which could be the missing link between PCs and terminals that could forever change desktop computing. Blackwell Blackwell is Nvidia’s premier, AI-focused GPU. When it was announced, it was so far over what most would have thoug...
Read More