Get the latest tech news
Scalable self-improvement for compiler optimization
October 23, 2024 Teodor V. Marinov and Alekh Agarwal, Research Scientists, Google Research We introduce Iterative BC-Max, a novel technique that aims to reduce the size of the compiled binary files by improving inlining decisions.
Since it is challenging to optimally manage these resources, there is increasing interest in the use of machine learning (ML) to make this decision-making data driven rather than heuristic. Iterative BC-Max produces a decision-making policy by solving carefully designed supervised learning problems instead of using unstable and computationally demanding RL algorithms. We would like to thank Mircea Trofin for all of his contributions that led to successfully deploying this algorithm, including implementation, help with the compiler side, and fostering tight collaboration with other Google teams.
Or read this on Hacker News
