Delving into Macro Placement with Reinforcement Learning
| Autor: | Anna Goldie, Ebrahim M. Songhori, Young-Joon Lee, David Z. Pan, Zixuan Jiang, Joe Jiang, Azalia Mirhoseini, Shen Wang |
|---|---|
| Jazyk: | angličtina |
| Rok vydání: | 2021 |
| Předmět: |
FOS: Computer and information sciences
Network architecture Computer Science - Machine Learning Computer science business.industry Computer Science - Artificial Intelligence Markov process Trial and error Machine Learning (cs.LG) symbols.namesake Artificial Intelligence (cs.AI) Value network symbols Reinforcement learning Markov decision process Artificial intelligence Physical design Macro business |
| Zdroj: | MLCAD |
| Popis: | In physical design, human designers typically place macros via trial and error, which is a Markov decision process. Reinforcement learning (RL) methods have demonstrated superhuman performance on the macro placement. In this paper, we propose an extension to this prior work (Mirhoseini et al., 2020). We first describe the details of the policy and value network architecture. We replace the force-directed method with DREAMPlace for placing standard cells in the RL environment. We also compare our improved method with other academic placers on public benchmarks. Accepted at 3rd ACM/IEEE Workshop on Machine Learning for CAD (MLCAD) |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|