Optimizing Cost Functions for Logic Synthesis with AnySyn

In the ever-evolving world of technology, resynthesis algorithms play a crucial role in optimizing the cost of concern for various applications. Traditional algorithms, which are designed for established technologies, may not be effective when applied to emerging ones. To address this challenge, we propose an efficient and adaptable resynthesis algorithm that can quickly test the effectiveness of different cost functions for new technologies.
Our approach leverages a combination of context propagation and node contribution functions to accurately evaluate the cost of concern. We define the context of a node as the information involved in cost evaluation that cannot be determined with the node alone. By restricting the context propagation function to a recursive function of node fanins, we ensure that the algorithm’s efficiency is not compromised by complex global calculations.
We evaluate our algorithm on a selection of benchmarks from the IWLS 2015 tion suite and show that it achieves comparable optimization quality to a specialized algorithm optimizing for a different cost function. Our algorithm takes less than 6 minutes to run on the largest benchmark, highlighting its efficiency in addressing emerging technologies.
In summary, our resynthesis algorithm is designed to keep pace with the rapidly evolving world of technology by providing a flexible and efficient means of optimizing the cost of concern for various applications. By leveraging context propagation and node contribution functions, we can accurately evaluate the cost of concern without sacrificing computational efficiency. Our algorithm’s ability to quickly test the effectiveness of different cost functions makes it an invaluable tool for prototyping and reducing development time in emerging technologies.