SSM-CIM: An Efficient CIM Macro Featuring Single-Step Multi-bit MAC Computation for CNN Edge Inference
- 10 August 2023
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems I: Regular Papers
- Vol. 70 (11) , 4357-4368
Abstract
Compute-in-memory (CIM) is a promising approach to solving the memory-wall problem existing in traditional computing architectures. In this paper, we introduce SSM-CIM, a charge-domain, static random-access memory (SRAM)-based CIM macro designed for area-energy-efficient convolutional neural network (CNN) inference. SSM-CIM utilizes an original sign-magnitude data encoding method for both inputs and weights. By codesigning four adjacent SRAM computing cells and employing a 3-bit digital-to-analog converter (DAC), SSM-CIM performs accurate 4-bit multiply-and-accumulate (MAC) computation in a single step, eliminating the peripheral digital shift-and-add circuits. To digitize the MAC computing results, a dedicated multi-reference assisted SAR ADC is designed by reusing the reference voltages from the DAC, which offers significant power and area savings. In addition, analog computing errors and quantization errors are analyzed to ensure the multi-bit computing accuracy of SSM-CIM. SSM-CIM is implemented and evaluated using 28-nm global foundry process. The post-layout simulation results validate the excellent computing linearity and accuracy of SSM-CIM. Benefitting from the compact layout design and fully parallel computing flow, the $144\times 256$ macro achieves a peak throughput of 2.3 TOPS, an area efficiency of 10.2 TOPS/mm2, and an energy efficiency of 205.4 TOPS/W with 4-bit weights and 4-bit inputs.Keywords
Funding Information
- National Key Research and Development Program of China (2022YFB4400900)
- National NSF of China (62141411, 62004096, 62004097)
This publication has 40 references indexed in Scilit:
- A Mixed-Signal Binarized Convolutional-Neural-Network Accelerator Integrating Dense Weight Storage and Multiplication for Reduced Data Movement2018
- In-Memory IntelligenceIEEE Micro, 2017
- A Reconfigurable Streaming Deep Convolutional Neural Network Accelerator for Internet of ThingsIEEE Transactions on Circuits and Systems I: Regular Papers, 2017
- In-Memory Computation of a Machine-Learning Classifier in a Standard 6T SRAM ArrayIEEE Journal of Solid-State Circuits, 2017
- Eyeriss: An Energy-Efficient Reconfigurable Accelerator for Deep Convolutional Neural NetworksIEEE Journal of Solid-State Circuits, 2016
- Deep Residual Learning for Image Recognition2016
- Matching Properties of Femtofarad and Sub-Femtofarad MOM CapacitorsIEEE Transactions on Circuits and Systems I: Regular Papers, 2016
- Going deeper with convolutions2015
- Mismatch Characterization of Small Metal Fringe CapacitorsIEEE Transactions on Circuits and Systems I: Regular Papers, 2014
- A 10-bit 50-MS/s SAR ADC With a Monotonic Capacitor Switching ProcedureIEEE Journal of Solid-State Circuits, 2010