Quadric, a prominent processor technology intellectual property (IP) licensor, and MegaChips, a leading ASIC and SoC services company based in Japan, have announced a strategic partnership. This collaboration is set to deliver cutting-edge ASIC and SoC solutions built on Quadric’s innovative edge AI processor architecture. MegaChips, a significant investor in Quadric, aims to support the development of the next version of Quadric’s processor architecture, enhance the Quadric software development kit (SDK), and introduce IP products that will be integrated into MegaChips’ ASICs and SoCs.
Key Developments and Features
- Innovative AI Vision Applications: The partnership brings advanced AI vision applications to the forefront, leveraging Quadric’s expertise in neural backbones and dynamic data-parallel algorithms.
- General-Purpose Neural Processing Unit (GPNPU): Quadric’s unique GPNPU architecture combines the functionality of neural network accelerators and DSPs, offering high machine learning (ML) inference performance and general-purpose control and signal processing capabilities.
- Flexible Edge Processing Solutions: Unlike traditional edge processing solutions that rely on high-power CPU clusters or specialized DSPs, Quadric’s GPNPU architecture provides a versatile approach to accelerating the entire application pipeline.
- Commitment to Edge AI Innovation: Douglas Fairbairn of MegaChips expressed enthusiasm about the partnership, citing Quadric’s innovative edge AI solution and market traction as key reasons for MegaChips’ investment and collaboration.
This partnership is set against the backdrop of an industry trend where cloud-based AI, ML, and inference capabilities are increasingly shifting to the network edge. This shift necessitates robust on-device AI processing to manage the vast data from IoT devices and edge-based applications. The collaboration aims to target applications such as autonomous driving, industrial robotics, factory automation, and advanced medical equipment.
Veerbhan Kheterpal, co-founder and CEO of Quadric, highlights the partnership as a pivotal moment in bringing accelerated computing to edge devices. He emphasizes the uniqueness of Quadric’s processing platform, which is fully programmable and supports a wide range of AI and ML algorithms, offering an open ecosystem to developers.