Salvato in:
| Autori principali: | , , , , , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | https://arxiv.org/abs/2512.11342 |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866909957662703616 |
|---|---|
| author | Ge, Jinming Du, Linfeng Anaparty, Likith Li, Shangkun Liang, Tingyuan Ahmad, Afzal Chaturvedi, Vivek Sinha, Sharad Xie, Zhiyao Xu, Jiang Zhang, Wei |
| author_facet | Ge, Jinming Du, Linfeng Anaparty, Likith Li, Shangkun Liang, Tingyuan Ahmad, Afzal Chaturvedi, Vivek Sinha, Sharad Xie, Zhiyao Xu, Jiang Zhang, Wei |
| contents | High-Level Synthesis (HLS) tools are widely adopted in FPGA-based domain-specific accelerator design. However, existing tools rely on fixed optimization strategies inherited from software compilations, limiting their effectiveness. Tailoring optimization strategies to specific designs requires deep semantic understanding, accurate hardware metric estimation, and advanced search algorithms -- capabilities that current approaches lack.
We propose DAPO, a design structure-aware pass ordering framework that extracts program semantics from control and data flow graphs, employs contrastive learning to generate rich embeddings, and leverages an analytical model for accurate hardware metric estimation. These components jointly guide a reinforcement learning agent to discover design-specific optimization strategies. Evaluations on classic HLS designs demonstrate that our end-to-end flow delivers a 2.36 speedup over Vitis HLS on average. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2512_11342 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | DAPO: Design Structure-Aware Pass Ordering in High-Level Synthesis with Graph Contrastive and Reinforcement Learning Ge, Jinming Du, Linfeng Anaparty, Likith Li, Shangkun Liang, Tingyuan Ahmad, Afzal Chaturvedi, Vivek Sinha, Sharad Xie, Zhiyao Xu, Jiang Zhang, Wei Machine Learning High-Level Synthesis (HLS) tools are widely adopted in FPGA-based domain-specific accelerator design. However, existing tools rely on fixed optimization strategies inherited from software compilations, limiting their effectiveness. Tailoring optimization strategies to specific designs requires deep semantic understanding, accurate hardware metric estimation, and advanced search algorithms -- capabilities that current approaches lack. We propose DAPO, a design structure-aware pass ordering framework that extracts program semantics from control and data flow graphs, employs contrastive learning to generate rich embeddings, and leverages an analytical model for accurate hardware metric estimation. These components jointly guide a reinforcement learning agent to discover design-specific optimization strategies. Evaluations on classic HLS designs demonstrate that our end-to-end flow delivers a 2.36 speedup over Vitis HLS on average. |
| title | DAPO: Design Structure-Aware Pass Ordering in High-Level Synthesis with Graph Contrastive and Reinforcement Learning |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2512.11342 |