| Residential College | false |
| Status | 已發表Published |
| A Reconfigurable Floating-Point Compute-In-Memory With Analog Exponent Pre-Processes | |
He, Pengyu3; Zhao, Yuanzhe3; Xie, Heng3; Wang, Yang1; Yin, Shouyi1; Li, Li2 ; Zhu, Yan3; Martins, Rui P.3,4; Chan, Chi Hang3; Zhang, Minglei3
| |
| 2024-09 | |
| Source Publication | IEEE Solid-State Circuits Letters
 |
| ISSN | 2573-9603 |
| Volume | 7Pages:271-274 |
| Abstract | This letter presents a reconfigurable floating-point computein-memory (FP-CIM) macro that preprocesses the exponent in the analog domain, enhancing the energy efficiency of edge devices for the floating-point (FP) inference. The presented FP-CIM macro supports FP8 inference, while can be configured to BP16 precision in a segmented computation manner. Furthermore, a time-domain analog-to-digital converter facilitates the analog compute-in-memory (CIM) macro while improving energy efficiency by sharing the counter and quantizing in a coarsefine structure. Fabricated in a 28-nm CMOS process, the presented FP-CIM macro achieves 314.6-TFLOPS/W energy efficiency and 12.13- TFLOPS/mm2 area efficiency at the FP8 mode. |
| Keyword | Compute-in-memory Macro(Cim) Exponent Pre-process Floating-point(Fp) Reconfigurable Segmented Computation |
| DOI | 10.1109/LSSC.2024.3463208 |
| URL | View the original |
| Indexed By | ESCI |
| Language | 英語English |
| WOS Research Area | Computer Science ; Engineering |
| WOS Subject | Computer Science, Hardware & Architecture ; Engineering, Electrical & Electronic |
| WOS ID | WOS:001327403600002 |
| Publisher | IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC, 445 HOES LANE, PISCATAWAY, NJ 08855-4141 |
| Scopus ID | 2-s2.0-85204419850 |
| Fulltext Access | |
| Citation statistics | |
| Document Type | Journal article |
| Collection | Faculty of Science and Technology THE STATE KEY LABORATORY OF ANALOG AND MIXED-SIGNAL VLSI (UNIVERSITY OF MACAU) THE STATE KEY LABORATORY OF INTERNET OF THINGS FOR SMART CITY (UNIVERSITY OF MACAU) INSTITUTE OF MICROELECTRONICS DEPARTMENT OF ELECTRICAL AND COMPUTER ENGINEERING |
| Corresponding Author | Zhang, Minglei |
| Affiliation | 1.Tsinghua University, School of Integrated Circuits, Beijing, 100084, China 2.University of Macau, State Key Laboratory of Internet of Things for Smart City, 999078, Macao 3.University of Macau, State Key Laboratory of Analog and Mixed Signal VLSI, Institute of Microelectronics, Department of Electrical and Computer Engineering, Faculty of Science and Technology, 999078, Macao 4.Universidade de Lisboa, Instituto Superior Tècnico, Lisbon, 1049-001, Portugal |
| First Author Affilication | Faculty of Science and Technology |
| Corresponding Author Affilication | Faculty of Science and Technology |
| Recommended Citation GB/T 7714 | He, Pengyu,Zhao, Yuanzhe,Xie, Heng,et al. A Reconfigurable Floating-Point Compute-In-Memory With Analog Exponent Pre-Processes[J]. IEEE Solid-State Circuits Letters, 2024, 7, 271-274. |
| APA | He, Pengyu., Zhao, Yuanzhe., Xie, Heng., Wang, Yang., Yin, Shouyi., Li, Li., Zhu, Yan., Martins, Rui P.., Chan, Chi Hang., & Zhang, Minglei (2024). A Reconfigurable Floating-Point Compute-In-Memory With Analog Exponent Pre-Processes. IEEE Solid-State Circuits Letters, 7, 271-274. |
| MLA | He, Pengyu,et al."A Reconfigurable Floating-Point Compute-In-Memory With Analog Exponent Pre-Processes".IEEE Solid-State Circuits Letters 7(2024):271-274. |
| Files in This Item: | There are no files associated with this item. | |||||
Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.
Edit Comment