A RRAM-Based CIM Design With in-Situ Transposable Computing and Hybrid-Precision Scheme for Edge Learning
Edge computing devices demand efficient transposable computing architectures to enable on-chip learning, necessitating novel hardware designs that balance performance and flexibility. We present an RRAM-based compute-in-memory macro capable of supporting both in-situ forward and backward propagation operations. The design incorporates an orthogonal-WL array structure with weight-level parallelism adjustment and a precision-driven input mechanism to enable flexible transposable computing. Additionally, optimized ADCs provide high throughput while maintaining area efficiency. The work exhibits a SOTA normalized area efficiency of 126.7 TOPS/mm2/bit, an energy efficiency of 2348.96 TOPS/W/bit, and a storage density of 4.84 Mb/mm2.
Authors
- Jianshi Tang (ORCID: https://orcid.org/0000-0001-8369-0067)
- Bin Gao (ORCID: https://orcid.org/0000-0002-2417-983X)
- Lu Jie (ORCID: https://orcid.org/0000-0001-5046-3917)
- Qiumeng Wei (ORCID: https://orcid.org/0009-0005-0550-4835)
- Huaqiang Wu (ORCID: https://orcid.org/0000-0001-8359-7997)
- Sining Pan (ORCID: https://orcid.org/0000-0002-5409-3367)
- Qingtian Zhang (ORCID: https://orcid.org/0000-0003-2732-3419)
- Peng Yao (ORCID: https://orcid.org/0000-0003-0172-4310)
- He Qian (ORCID: https://orcid.org/0000-0003-3377-5366)
- Qi Qin (ORCID: https://orcid.org/0000-0001-9932-9057)
- Dong Wu
Institutions
- Tsinghua University (CN)
Publication Details
- Journal
- IEEE Solid-State Circuits Letters
- Published
- 2025-01-01
- DOI
- https://doi.org/10.1109/lssc.2025.3589580
- Citations
- 2
- Primary Topic
- Advanced Memory and Neural Computing
- Type
- article
- Field-Weighted Citation Impact
- 1.46
Funders
- National Natural Science Foundation of China
- National Major Science and Technology Projects of China