arXiv 2004.10746
Chip Placement with Deep Reinforcement Learning
By Azalia Mirhoseini, Anna Goldie, et al.
Published 2020-04-22
Citation lineage
Review the prior work and downstream research connected to this paper.
In this work, we present a learning-based approach to chip placement, one of the most complex and time-consuming stages of the chip design process. Unlike prior methods, our approach has the ability to learn from past experience and improve over time. In particular, as we train over a greater number of chip blocks, our method becomes better at rapidly generating optimized placements for previously unseen chip blocks…