App developers can now access Qualcomm’s artificial intelligence models, which are optimized for the Qualcomm 45 TOPS Hexagon neural processing unit (NPU) and available through the Qualcomm AI Hub. These models are aimed at enhancing the Snapdragon X Elite platforms.
Enhancing AI Functionalities
Qualcomm’s AI Hub serves as a platform for developers, offering a range of advanced machine learning models optimized for Qualcomm devices.
The AI Hub seamlessly manages the translation of models from a source framework to the device runtime, implements hardware-aware optimizations, and conducts both performance and numerical validation. Additionally, the system provisions devices in the cloud automatically for on-device profiling and inference tasks.
Good news, developers: Qualcomm has made compute models on Qualcomm AI Hub available to everyone. Sign up today to immediately run, download, and deploy optimized #AI models on @Snapdragon X Elite devices: https://t.co/yxBVHjJHI4
— Qualcomm (@Qualcomm) June 21, 2024
In the AI Hub, developers can find pre-trained models for tasks like image classification, object detection, semantic segmentation, and generative AI. These models leverage the Hexagon NPU’s high-performance capabilities.
With Qualcomm AI Hub, you can:
-
Convert trained models from frameworks like PyTorch and ONNX for optimized on-device performance on Qualcomm devices.
-
Profile models on-device to obtain detailed metrics including runtime, load time, and compute unit utilization.
-
Verify numerical correctness by performing on-device inference.
-
Easily deploy models using Qualcomm AI Engine Direct, TensorFlow Lite, or ONNX Runtime.
Supporting Tools and Documentation
To aid developers, Qualcomm provides comprehensive tools and documentation via the AI Hub. Models can be filtered by attributes such as ‘backbone,’ ‘foundation,’ ‘quantized,’ and ‘real-time.’ The hub also offers detailed examples and code snippets, facilitating the efficient deployment and operation of AI models on Snapdragon X Elite devices.
Qualcomm’s focus on AI aims to attract software developers to the Snapdragon X Elite platforms. The availability of optimized AI models is expected to stimulate the development of AI-powered applications, increasing the demand for PCs equipped with Qualcomm’s system-on-chips (SoCs).
This approach places Qualcomm in a competitive position against AMD and Intel, both of which are integrating advanced NPUs into their new CPUs for AI applications. The Snapdragon X Elite, unveiled at Computex 2024, is designed with a focus on AI rather than traditional CPU and GPU features.
Last Updated on November 7, 2024 3:49 pm CET