摘要
This work presents the first dedicated processor that supports the entire hyperspectral image (HSI) processing work-flow. Folding and data interleaving are applied to the processing elements (PEs), reducing the area by 57%. The reconfigurable PE datapath enables 92.7% utilization with 38.1% less area. Data sparsity is leveraged for clock gating, saving 32.2% power. The chip dissipates 44.3mW at 175MHz from a 0.68V supply. It delivers a 62.4fps throughput for 256×256×64 HSIs. Compared to a high-end CPU, this work achieves a 544× higher throughput and 82,352× higher energy efficiency.
| 原文 | English |
|---|---|
| 主出版物標題 | 2022 IEEE Symposium on VLSI Technology and Circuits, VLSI Technology and Circuits 2022 |
| 發行者 | Institute of Electrical and Electronics Engineers Inc. |
| 頁面 | 74-75 |
| 頁數 | 2 |
| ISBN(電子) | 9781665497725 |
| DOIs | |
| 出版狀態 | Published - 2022 |
| 事件 | 2022 IEEE Symposium on VLSI Technology and Circuits, VLSI Technology and Circuits 2022 - Honolulu, 美國 持續時間: 12 6月 2022 → 17 6月 2022 |
出版系列
| 名字 | Digest of Technical Papers - Symposium on VLSI Technology |
|---|---|
| 卷 | 2022-June |
| ISSN(列印) | 0743-1562 |
Conference
| Conference | 2022 IEEE Symposium on VLSI Technology and Circuits, VLSI Technology and Circuits 2022 |
|---|---|
| 國家/地區 | 美國 |
| 城市 | Honolulu |
| 期間 | 12/06/22 → 17/06/22 |
UN SDG
此研究成果有助於以下永續發展目標
-
SDG 7 經濟實惠的清潔能源
指紋
深入研究「A 44.3mW 62.4fps Hyperspectral Image Processor for MAV Remote Sensing」主題。共同形成了獨特的指紋。引用此
- APA
- Author
- BIBTEX
- Harvard
- Standard
- RIS
- Vancouver