GBT Technologies Inc. received a notice of publication for its microchip's reliability verification and auto-correction EDA patent application, which has an internal project's code name of Epsilon. As integrated circuits (ICs) manufacturing node is constantly scaling down, design firms are facing broad spectrum, complex challenges in the electrical reliability domain, which is dictated by the laws of physics. High performance computing systems require optimal electrical and thermal characteristics to ensure reliability consistency, and accuracy of the processed data.

Advanced integrated circuits are the core of these systems and need to function with a high level of dependability. Particularly in the fields of medicine, aviation, transportation, data storage and communication, microchip's reliability factor has become a crucial factor. The company's Epsilon patent application introduces innovative methods and systems seek to address advanced semiconductor node's physics with the goal of ensuring a high level of reliability, optimal thermal design, lower power consumption and high performance.

The technology is designed to analyze an IC's layout using machine learning algorithms and to identify electrical reliability flaws. The patent application describes the detection and automatic correction of these reliability issues early during the IC's design phase. Neural networks will be performing data analysis, identification, categorization, and reasoning about executing the IC layout automatic correction.

The goal of this technology is to allow IC designers to analyze and fix circuits more efficiently, in real-time, and get visibility into electrical issues early during the design stages. The goal of this technology is to target the microchip's early design stages electrical and power analysis to identify potential failures and provide an on-the-fly solution which theoretically can save weeks to months of engineering re-design time.