Quadric

Quadric stands out in the realm of semiconductor IP providers with a pioneering approach to on-device artificial intelligence processing. Their flagship product, the Chimera General Purpose Neural Processing Unit (GPNPU), is designed to scale from 1 to a remarkable 864 TOPs. This scalability meets the needs of various market segments, including automotive applications that demand high performance and adherence to safety standards. The Chimera GPNPU distinguishes itself by supporting all machine learning networks, including large language models and vision transformers, without requiring a rewrite of existing code for unique processor types. One of Quadric's significant innovations is enabling developers to execute complex C++ code concurrently with machine learning inference on a single processor pipeline. This hybrid capability streamlines processor efficiency and eliminates the need to split application code across multiple processing units, a common limitation seen in conventional SoC designs. As a licensable product, the Chimera GPNPU offers exceptional matrix-computation performance, proven across its various implementations. Designed with a futuristic vision, the Chimera GPNPU caters to modern system-on-chip architectures that blend traditional code with machine learning inference complexities. By offering a fully programmable approach, Quadric ensures that their processors can adapt to new developments in AI technology far into the future. This future-proof flexibility allows for ongoing updates and enhancements to deployed SoCs, providing both longevity and competitive advantage in fast-evolving tech landscapes. Quadric’s strategy also extends into automotive technology, where their processors are tailored to deliver both performance and reliability required for systems like Advanced Driver Assistance Systems (ADAS), In-Vehicle Infotainment (IVI), and Electronics Control Units (ECU). By providing scalable, multicore solutions that comply with ASIL B/D standards, their technology supports the varied needs of automotive safety applications. Read more