MemryX and Edge Impulse Integrate Tools to Accelerate AI Hardware Development

Edge Impulse, a leading AI developer platform, and MemryX, a premier AI hardware-based accelerator, have joined forces to revolutionize the AI workflow by simplifying the development, testing, and deployment of AI models.

Traditionally, creating and deploying AI models for edge-based hardware posed significant challenges for enterprises and developers. It involved the arduous task of assembling high-quality datasets, extracting data features, and efficiently deploying models to devices, often filled with complexities. The recent integration between MemryX and Edge Impulse promises to reshape the AI workflow, advancing the deployment of dedicated AI accelerators on the edge.

Developer Hub Integration

At the heart of this integration are the MemryX Developer Hub and Edge Impulse Studio, offering complementary tools for AI model development and deployment. The MemryX Developer Hub is designed to provide a straightforward, robust, and flexible solution for edge AI, empowering developers to harness the full potential of MemryX AI accelerator chips. MemryX AI products support major neural frameworks and thousands of neural models, using popular host hardware ISAs and major OS environments. This platform allows users to swiftly simulate and deploy a wide range of pre-trained AI models while providing flexibility to optimize performance, power efficiency, latency, and resource utilization for any application.

Edge Impulse offers a low-to-no-code tool for quickly creating robust AI models deployable directly to edge-based devices, including those powered by MemryX accelerators. From data collection to algorithm generation to model refinement, Edge Impulse enables enterprises and developers to create, test, and launch AI solutions for real-world applications such as object tracking or anomaly detection.

Through this partnership, Edge Impulse users gain multiple seamless ways to work with MemryX files and devices. Edge Impulse has seamlessly incorporated the MemryX Developer Hub SDK into its platform, enabling users to complete the entire AI model creation and compilation process within the Edge Impulse platform. MemryX developers can leverage the powerful features offered by Edge Impulse, simulate, and deploy to hardware. Moreover, developers can deploy their Edge Impulse Studio projects as MemryX executable files and utilize the MemryX Developer Hub. The Edge Impulse integration also allows developers to download generated files for custom program integration.

MemryX MX3 AI Accelerator Chip

MemryX has introduced its MX3 Edge AI Accelerator samples, featuring an innovative pure dataflow architecture equipped with programmable hardware that seamlessly aligns with the data-driven design of AI algorithms. This advanced architecture is specifically tailored to handle streaming inputs from real-time sources like cameras and sensors, ensuring seamless integration into applications. The MX3 supports multi-chip configurations, working together as one logical chip, offering scalability to achieve desired levels of AI performance and model size. Furthermore, MX3 provides full interoperability with major AI frameworks, application processors, and host operating systems. The MemryX MX3 AI Accelerator is now natively supported in Edge Impulse.

A standout feature of the MemryX MX3 is its proprietary 1-click compilation and mapping software. This software optimizes AI model performance without the need for retraining or software fine-tuning, enabling efficient execution on target hardware. This significantly accelerates development cycles and facilitates hassle-free edge AI upgrades. By offloading AI inference processing from application processors to the dedicated MX3 AI accelerator(s), users can maximize system efficiencies while minimizing costs.

The MemryX Developer Hub integration is now available for Edge Impulse users. Get started by signing up at edgeimpulse.com.

Source:businesswire.com