IP Licensing for Matrix Based AI Accelerator That Optimizes Performance with Low Energy for Self-Driving and Automotive AI Functions
Future car designs incorporating AI must leverage system architectures that bring not only high performance but also power efficiency. A leading car manufacturer recently announced replacing the use of GPUs with their own matrix-based, application specific chip designs for supporting AI and self-driving in vehicles.
This latest development of GPUs being replaced has automakers seeking an AI chip architecture specific to self-driving and automotive AI. An automotive industry executive recently claimed that it would take as long as three years from start to finish to develop a high performing, energy efficient chip that supports self-driving and other AI functions.
Gyrfalcon Technology Inc. (GTI), has been promoting matrix-based application specific chips for all forms of AI since offering their production versions of AI accelerator chips in September 2017. Through the licensing of its proprietary technology, the company is confident it can help automakers bring highly competitive AI chips to production for use in vehicles within 18 months, along with significant gains in AI performance, improvements in power dissipation and cost advantages.
“The need for higher performance, machine-learning devices to advance autonomous driving technology means suppliers must develop AI-based intellectual property or license it from companies like Gyrfalcon for silicon solutions to achieve faster processing and lower power consumption,” said Luca De Ambroggi, Senior AI Research and Automotive Research Director for IHS Markit, based in Munich, Germany.
GTI offers IP licensing based on its production level silicon which has been proven in the hands of customers. The technology is delivering the highest ratio of energy efficiency combined with very high performance. Customers are designing the chips into commercial products, such as smart home & office, consumer electronics, mobile phones & computers, baby & pet monitors, robot vacuums, defect detection equipment, edge servers, automobiles and AI data center solutions.
One matrix based chip design that replaced a GPU for self-driving AI provided equipment cost reductions and a performance-to-efficiency ratio close to 5 TOPS/W. GTI’s matrix based AI chips provide 9.3 TOPS/W (Lightspeeur® 2801) or 24 TOPS/W (Lightspeeur® 2803), which already bring twice to five times the improvement on that ratio.