Guwahati: Machine learning framework developed for semiconductor industry

Guwahati: Researchers at the Indian Institute of Technology, Guwahati (IIT-G) have developed an innovative machine learning framework called ‘LEAP’, which marks a significant advancement in the field of electronic design automation (EDA) used in the semiconductor industry. The development of this cutting-edge solution improves the design process of integrated circuits (ICs), a critical component of the $600 billion semiconductor industry that powers modern electronic devices, an official release said on Thursday. Designing ICs involves solving complex problems that can be challenging to solve and often yield less-than-ideal results.

A team of researchers, comprising Professor Chandan Karfa and Dr Sukanta Bhattacharjee from the Department of Computer Science and Engineering, along with BTech students Chandrabhushan Reddy Chigarpalli and Harshavardhan Nitin Bhakkar, have leveraged machine learning to improve efficiency in IC design. The project also involved another collaborator, Dr Animesh Basak Chowdhury from New York University, USA. Karaffa said the LEAP framework streamlines the technology mapping process within EDA. “Instead of evaluating thousands of possible configurations, LEAP intelligently identifies and prioritizes the most promising options, reducing the number of configurations considered by the mapping tool by more than 50 percent,” he said.

The framework not only speeds up the mapping process but also improves circuit performance, he said. Karaffa said LEAP estimates delays for different configurations and selects only the top ten options for each node in the design, while the traditional method typically evaluates about 250 configurations. This targeted approach streamlines the workflow and increases overall efficiency. This research holds real-world implications for the semiconductor industry, which is essential for the development of electronic devices such as smartphones and computers. It will lead to faster, more efficient electronic devices with lower energy consumption, ultimately benefiting consumers and driving innovation across various technology sectors. The results of this work have been published in the ACM/IEEE International Conference on Computer-Aided Design (ICCAD 2024), the release said.