Affiliation:
1. School of Physics and Information Engineering, Fuzhou University, Fuzhou 350116, China
Abstract
Conventional sense amplifiers limit the performance of current RRAM computing-in-memory (CIM) macro circuits, resulting in high latency and energy consumption. This paper introduces a multi-bit quantization technology low-latency voltage sense amplifier (MQL-VSA). Firstly, the multi-bit quantization technology enhances circuit quantization efficiency, reducing the number of operational states in conventional VSA. Secondly, by simplifying the sequential logic circuits in conventional VSA, the complexity of sequential control signals is reduced, further diminishing readout latency. Experimental results demonstrate that the MQL-VSA achieves a 1.40-times decrease in readout latency and a 1.28-times reduction in power consumption compared to conventional VSA. Additionally, an 8-bit input, 8-bit weight, 14-bit output macro circuit utilizing MQL-VSA exhibited a 1.11times latency reduction and 1.04-times energy savings.
Funder
National Natural Science Foundation of China
Natural Science Foundation of Fujian Province of China
Science and Technology Plan project of Fujian Province of China
Reference22 articles.
1. 8-b Precision 8-Mb ReRAM Compute-in-Memory Macro Using Direct-Current-Free Time-Domain Readout Scheme for AI Edge Devices;Hung;IEEE J. Solid-State Circuits,2023
2. A 40-nm, 64-Kb, 56.67 TOPS/W Voltage-Sensing Computing-In-Memory/Digital RRAM Macro Supporting Iterative Write with Verification and Online Read-Disturb Detection;Yoon;IEEE J. Solid-State Circuits,2022
3. A 40-nm 118.44-TOPS/W Voltage-Sensing Compute-in-Memory RRAM Macro with Write Verification and Multi-Bit Encoding;Yoon;IEEE J. Solid-State Circuits,2022
4. Zhang, F., He, W., Yeo, I., Liehr, M., Cady, N., Cao, Y., Seo, J.S., and Fan, D. (2023, January 11–14). A 65nm RRAM Compute-in-Memory Macro for Genome Sequencing Alignment. Proceedings of the ESSCIRC 2023 IEEE 49th European Solid State Circuits Conference (ESSCIRC), Lisbon, Portugal.
5. An RRAM-Based Digital Computing-in-Memory Macro With Dynamic Voltage Sense Amplifier and Sparse-Aware Approximate Adder Tree;He;IEEE Trans. Circuits Syst. II Express Briefs,2023