Research map / Reinforcement learning / Visual and representation RL
On Joint Learning for Solving Placement and Routing in Chip Design
Ruoyu Cheng, Junchi Yan
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| end-to-end joint training of macro and standard-cell placement vs sequential training | DeepPlace · adaptec1 | wirelength: 86675688 -> 80117232 | Improved |
| end-to-end joint training vs sequential on bigblue1 | DeepPlace · bigblue1 | wirelength: 168456768 -> 140435296 | Improved |
| RL macro placement vs DREAMPlace heuristic macro placement | DeepPlace · adaptec2..bigblue4 | congestion check | Improved |
| joint placement-routing with routed wirelength reward vs separate pipeline | DeepPR · adaptec4 | routed wirelength: 79573 -> 63560 | Improved |
| joint placement-routing vs separate pipeline | DeepPR · bigblue4 | routed wirelength: 17829 -> 17585 | Improved |
| pretrain policy on macro-only netlist for 300 iterations vs from scratch (ablation) | DeepPlace · adaptec1 | mean episodic return | Improved |
| pretrain for 500 iterations instead of 300 (ablation) | DeepPlace · adaptec1 | mean episodic return | Hurt |
| add random network distillation intrinsic reward (ablation) | DeepPlace agent · adaptec1 | mean episodic return | Improved |
| combine CNN global and GNN node embeddings vs CNN alone (ablation) | DeepPlace agent · adaptec1 | mean episodic return | Improved |
| use CNN alone vs GNN alone for embedding (ablation) | DeepPlace agent · adaptec1 | mean episodic return | No clear effect |
Closest papers by meaning
- Chip Placement with Diffusion Models (ICML 2025 · similarity 0.90)
- Reinforcement Learning within Tree Search for Fast Macro Placement (ICML 2024 · similarity 0.88)
- Reinforcement Learning Policy as Macro Regulator Rather than Macro Placer (NeurIPS 2024 · similarity 0.88)
- EGPlace: An Efficient Macro Placement Method via Evolutionary Search with Greedy Repositioning Guided Mutation (ICML 2025 · similarity 0.86)
- See it to Place it: Evolving Macro Placements with Vision Language Models (NeurIPS 2026 · similarity 0.86)
- DeepLayout: Learning Neural Representations of Circuit Placement Layout (ICML 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
