Research map / Learning theory and algorithms / Combinatorial and search optimization
NeuroSchedule: A Novel Effective GNN-based Scheduling Method for High-level Synthesis
Jun Zeng, Mingyang Kou, Hailong Yao
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 |
|---|---|---|---|
| use GNN-predicted priorities instead of entropy-directed priorities | list scheduling · CHStone+MiBench | average schedule length improvement: 6.10% | Improved |
| use GNN-predicted priorities instead of entropy-directed priorities | list scheduling · CHStone+MiBench | optimal rate: 16.67% -> 88.89% | Improved |
| use GNN-predicted priorities instead of entropy-directed priorities | list scheduling · Synthesized complex benchmark suite | better solutions rate: 94% | Improved |
| use NeuroSchedule instead of ILP-based exact scheduling | scheduling · CHStone+MiBench | runtime: more than 50,000x speedup | Improved |
| use NeuroSchedule instead of ILP-based exact scheduling | scheduling · CHStone+MiBench | schedule quality: 0.30% | Mixed |
| use NeuroSchedule instead of EDS | scheduling · CHStone+MiBench | total runtime: 0.167 -> 0.952 | Hurt |
| fine-tune from a pre-trained model instead of training from scratch on small dataset (ablation) | NeuroSchedule · Synthesized complex benchmark suite | optimal rate: 3% improvement | Improved |
| train with rank loss instead of MSE loss (ablation) | NeuroSchedule · Synthesized complex benchmark suite | optimal rate | Improved |
| ILP-based method on large CDFGs (dfdiv, dfmul, dfsin, jpeg) | ILP-based · CHStone | solvability: >104 | Hurt |
Closest papers by meaning
- Hierarchical Mixture of Experts: Generalizable Learning for High-Level Synthesis (AAAI 2025 · similarity 0.85)
- Towards Next-Generation Logic Synthesis: A Scalable Neural Circuit Generation Framework (NeurIPS 2024 · similarity 0.84)
- Eco Search: A No-delay Best-First Search Algorithm for Program Synthesis (AAAI 2025 · similarity 0.83)
- A Circuit Domain Generalization Framework for Efficient Logic Synthesis in Chip Design (ICML 2024 · similarity 0.83)
- Retrieval-Guided Reinforcement Learning for Boolean Circuit Minimization (ICLR 2024 · similarity 0.83)
- SymRTLO: Enhancing RTL Code Optimization with LLMs and Neuron-Inspired Symbolic Reasoning (NeurIPS 2025 · similarity 0.82)
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.
