Synopsys, Inc. announced that its AI-based design system has been used by Samsung to successfully complete a state-of-the-art, high-performance design at an advanced process technology, the most recent of several products designed using Synopsys artificial intelligence (AI).
Also Read: Snyk Security Intelligence Integrates into the new, enhanced Amazon Inspector
“For decades, autonomous chip design existed only in science fiction,” said Aart de Geus, chairman and co-CEO, Synopsys. “This pivotal moment in semiconductor history will breathe new life into Moore’s law. We congratulate Samsung on this remarkable achievement, and we look forward to catalyzing its next 1000x.”
The AI-designed product will be manufactured on Samsung’s advanced manufacturing process. To achieve the high-performance and low-power market requirements in a timely manner, Samsung used Synopsys’ award-winning autonomous AI system, DSO.ai™ (Design Space Optimization AI), driving the Synopsys Fusion Compiler™ RTL-to-GDSII solution. DSO.ai uses reinforcement learning, an AI technology similar to that used in self-driving vehicles, to achieve better performance, power and area (PPA).
Applied at every stage of design implementation, DSO.ai pushed operating frequency over 100 MHz beyond target and considerably reduced overall power consumption – all while saving Samsung weeks of manual design effort. An early development partner of Synopsys’ autonomous design technology, Samsung began deploying DSO.ai to multiple projects in the fall of 2020.
“This is a remarkable milestone for our program to successfully introduce AI into the chip design process in collaboration with Synopsys,” said Thomas Cho, EVP of Infrastructure & Design Technology Center, System LSI Business, Samsung Electronics. “Not only have we demonstrated that AI can help us achieve PPA targets for even the most demanding process technologies, but through our partnership we have established an ultra-high-productivity design system that is consistently delivering impressive results.
DSO.ai introduces a novel approach to searching vast problem spaces of chip design for optimal solutions, enabled by the latest advancements in AI and machine-learning. Traditional design space exploration has been a very labor-intensive effort, typically requiring months of experimentation, guided by past experiences and institutional knowledge.