Qualcomm Launches Unified AI Stack for Cloud to Edge Intelligence

Advertisement

The new Qualcomm AI Stack lays the groundwork for domain-specific SDKs

While advances in AI accelerators have improved silicon performance by more than a thousand times, it is software that turns bits of silicon into useful capabilities for consumers and businesses. After all, we don’t even think about the underlying chip when we take a photo in low light on our phone, play a VR game, or ask our digital assistant to set a reminder. The underlying AI that makes these functions possible is hidden by software developed by thousands of engineers around the world. These application ninjas need software tools that are closely tied to the silicon to be productive and produce high-performance code. The ultimate goal is that AI engineers can develop, optimize, deploy and accelerate their models on specific hardware. The developers also want multiple access layers within a software stack working with their own design workflow, and write and run this code once on any hardware, regardless of the presence or absence of specific acceleration functions.

Qualcomm Technologies, Inc. (QTI) recently redesigned and enhanced its comprehensive suite of AI software to help these developers gain multiple entry points across software layers and move their code more freely across the company’s products and levels. The company’s new AI software stack also powers domain-specific software development kits (SDKs) that extend capabilities across a wide variety of QTI hardware platforms. It can also be used as a standalone development kit if you just want to develop on the Snapdragon mobile platform, for example.

What is the QTI AI stack?

QTI’s goal in creating this stack was to unify and simplify their suite of software tools for OEMs and developers to create, optimize, and deploy AI applications on Qualcomm Technologies’ products, while improving the performance and efficiency of the Qualcomm® AI Engine to be fully utilized.

Qualcomm AI Stack supports a range of popular different AI frameworks, including TensorFlow, PyTorch, and runtimes including TFLite and ONNXRT. The AI ​​stack consists of underlying developer libraries and services, system software, tools, and compilers, so that any AI feature developed for one device can be easily implemented on other devices. The existing Qualcomm Neural Processing SDK, the popular Qualcomm® AI Model Efficiency Toolkit (AIMET) Pro, the AIMET Model Zoo, model analyzers and Neural Architecture Search (NAS) are all included. Additionally, Qualcomm recently ported the Qualcomm Neural Processing SDK to Microsoft Windows, while extending the Qualcomm AI Engine directly, an advanced AI runtime, to all Qualcomm Technologies products, including the Qualcomm® Cloud AI 100 inference processor. This latest enhancement will help developers achieve the goal QTI once wrote to be able to write anywhere, allowing developers to implement existing models directly into the AI ​​accelerators on Qualcomm Technologies’ platforms.

On top of the Qualcomm AI Software Stack are three domain-specific SDKs for autonomous vehicles (Snapdragon Ride™ SDK), Intelligent Multimedia SDK for robotics and IoT, and virtual reality (Snapdragon Spaces™ SDK). Again, building these SDKs on a common foundation helps developers support the full portfolio of QTI hardware implementations across various segments, including the Qualcomm Cloud AI 100.

Qualcomm AI Stack unifies several AI frameworks and popular runtimes, including TensorFlow, PyTorch, and ONNX. These frameworks build on the underlying developer libraries and services, system software, tools, and compilers, so that any AI feature developed for one device can be easily implemented on other devices. The existing Qualcomm Neural Processing SDK, the popular AI Model Efficiency Toolkit (AIMET), the AIMET Model Zoo, model analyzers, and Neural Architecture Search (NAS) are all included. Additionally, Qualcomm recently ported the Neural Processing SDK to Microsoft Windows, while extending the AI ​​Engine Direct to all Qualcomm products, including the Cloud AI 100 inference processor. This latest enhancement helps developers achieve the goal that QTI can write anywhere, allowing developers to implement existing models directly into the AI ​​accelerators on Qualcomm Technologies platforms.

On top of the Software Stack are three domain-specific SDKs for autonomous vehicles (Snapdragon Ride), Intelligent Multimedia, and virtual reality (Snapdragon Spaces). Again, building these SDKs on a common foundation helps developers support the full portfolio of QTI hardware implementations across various segments, including the Cloud AI 100.

conclusions

By unifying Qualcomm Technologies’ comprehensive suite of AI software, developers can more easily build apps using the same set of tools, regardless of the target deployment platform, including mobile, automotive, XR, compute, IoT, and cloud platforms. As QTI expands its reach into the connected intelligent edge, the importance of a unified software stack has only grown, and the company has now taken that challenge head-on with Qualcomm AI Stack and will have the ability to bring more AI development tools and more domain-specific SDKs in the future.

Visit Source link!

Leave a Comment