MegaChips took an equity stake in Quadric in January 2022 and is also a major investor in a $21m Series B funding round announced in March through their MegaChips LSI USA Corporation subsidiary. The round was intended to help Quadric release the next version of its processor architecture, improve the performance and breadth of the Quadric software development kit (SDK), and roll out IP products to be integrated in MegaChips’ ASICs and SoCs.
Quadric has the ability to handle both neural backbones and classical dynamic data-parallel algorithms in a unified architecture, bringing advanced on-device AI capabilities to edge-based applications.
The Quadric architecture is said to be the first general-purpose neural processing unit (GPNPU) and it delivers high machine learning (ML) inference performance, but unlike other neural network accelerators that support a limited number of ML graph operators, the Quadric solution also has general-purpose control and signal processing capabilities, blending the attributes of NPU accelerators with DSPs.
Quadric GPNPUs can run both neural net graphs and C++ code for signal pre-processing and post-processing
While many edge processing solutions combine high-power CPU clusters or exotic DSPs with application-specific network processing units (NPUs), Quadric’s GPNPU architecture has been developed to provide the flexibility to accelerate the entire application pipeline without the need for a companion processor paired with the NPU.
“Our continued investment in edge AI leaders like Quadric reflects our commitment to supporting disruptive AI-based on-device computing solutions,” said Douglas Fairbairn, Director, Business Development at MegaChips. “Quadric’s stellar team, market traction with key customers and innovative edge AI solution contributed to our decision to invest in Quadric. We are proud to work with Quadric to bring its end-to-end IP products to our mutual customers in the form of customised silicon.”
The collaboration between the two companies looks to address a growing industry trend in which cloud-based AI, ML and inference are migrating to the network edge. This migration requires powerful on-device AI processing capabilities to handle the massive amounts of data generated by billions of IoT devices and edge-based applications.
Target applications for MegaChips’ ASICs and SoCs based on Quadric’s IP include autonomous driving (ADAS/LiDAR), industrial robotics, factory automation, automated guided vehicles (AGV) and advanced medical equipment.
Commenting Veerbhan Kheterpal, co-founder and CEO of Quadric, said, “While the AI market is saturated with rigid accelerators that only address a portion of the machine learning compute challenge, our processing platform fills the void with a fully programmable, multi-kernel processing architecture. We've reimagined a general-purpose processor architecture and built it from scratch to give developers what they want: a versatile, high-performance processing platform that can support any type of AI and ML algorithm with an open ecosystem.”
