Efficient yet Accurate End-to-End SC Accelerator Design
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866909084671803392 |
|---|---|
| author | Li, Meng Hu, Yixuan Zhang, Tengyu Wei, Renjie Zhang, Yawen Huang, Ru Wang, Runsheng |
| author_facet | Li, Meng Hu, Yixuan Zhang, Tengyu Wei, Renjie Zhang, Yawen Huang, Ru Wang, Runsheng |
| contents | Providing end-to-end stochastic computing (SC) neural network acceleration for state-of-the-art (SOTA) models has become an increasingly challenging task, requiring the pursuit of accuracy while maintaining efficiency. It also necessitates flexible support for different types and sizes of operations in models by end-to-end SC circuits. In this paper, we summarize our recent research on end-to-end SC neural network acceleration. We introduce an accurate end-to-end SC accelerator based on a deterministic coding and sorting network. In addition, we propose an SC-friendly model that combines low-precision data paths with high-precision residuals. We introduce approximate computing techniques to optimize SC nonlinear adders and provide some new SC designs for arithmetic operations required by SOTA models. Overall, our approach allows for further significant improvements in circuit efficiency, flexibility, and compatibility through circuit design and model co-optimization. The results demonstrate that the proposed end-to-end SC architecture achieves accurate and efficient neural network acceleration while flexibly accommodating model requirements, showcasing the potential of SC in neural network acceleration. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2401_15332 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Efficient yet Accurate End-to-End SC Accelerator Design Li, Meng Hu, Yixuan Zhang, Tengyu Wei, Renjie Zhang, Yawen Huang, Ru Wang, Runsheng Hardware Architecture Providing end-to-end stochastic computing (SC) neural network acceleration for state-of-the-art (SOTA) models has become an increasingly challenging task, requiring the pursuit of accuracy while maintaining efficiency. It also necessitates flexible support for different types and sizes of operations in models by end-to-end SC circuits. In this paper, we summarize our recent research on end-to-end SC neural network acceleration. We introduce an accurate end-to-end SC accelerator based on a deterministic coding and sorting network. In addition, we propose an SC-friendly model that combines low-precision data paths with high-precision residuals. We introduce approximate computing techniques to optimize SC nonlinear adders and provide some new SC designs for arithmetic operations required by SOTA models. Overall, our approach allows for further significant improvements in circuit efficiency, flexibility, and compatibility through circuit design and model co-optimization. The results demonstrate that the proposed end-to-end SC architecture achieves accurate and efficient neural network acceleration while flexibly accommodating model requirements, showcasing the potential of SC in neural network acceleration. |
| title | Efficient yet Accurate End-to-End SC Accelerator Design |
| topic | Hardware Architecture |
| url | https://arxiv.org/abs/2401.15332 |