Abstract:Resistive random access memory (RRAM) is a promising technology for energy-efficient in-memory computing. However, due to technology limits, RRAM device faces a series of reliability issues. Deep neural network (DNN) computing based on RRAM suffers from accuracy degradation. On the one hand, offline DNN training solutions are difficult to fully consider and simulate all nonidealities. Worse still, new error or nonideality may come up with the usage of RRAM, which further deteriorates the effectiveness of offline training. On the other hand, online training poses great challenges on programming overhead and device lifetime. The iterative writeverify technique to program multi-bit RRAM cells prolongs write latency more than 10x longer than read latency. To overcome these issues, we propose a compensation architecture and a software and hardware co-training design to mitigate the realistic network accuracy loss in RRAM-based computing-inmemory chips. Firstly, we add trainable compensation channels in crossbars utilizing the residual resource after original weight mapping. Secondly, an offline training procedure with computing output from hardware is triggered to settle down appropriate weight value in compensation channels. Experimental results demonstrate that the proposed design can guarantee <= 0.8% loss of accuracy in DNN on MNIST and CIFAR10 dataset even when nonidealities reduce the original accuracy down to =73%.

An RRAM Retention Prediction Framework Using a Convolutional Neural Network Based on Relaxation Behavior

An 8-Bit in Resistive Memory Computing Core with Regulated Passive Neuron and Bitline Weight Mapping

Lifetime Improvement of 28 Nm Resistive Random Access Memory Chip by Machine Learning‐Assisted Prediction Model Collaborated with Resurrection Algorithm

Bit-Aware Fault-Tolerant Hybrid Retraining and Remapping Schemes for RRAM-Based Computing-in-Memory Systems

Crossbar-Level Retention Characterization in Analog RRAM Array-Based Computation-in-Memory System

An On-chip Layer-wise Training Method for RRAM Based Computing-in-memory Chips.

A Compact Model for Relaxation Effect in Analog RRAM for Computation-in-Memory System Design and Benchmark

Compensation Architecture to Alleviate Noise Effects in RRAM-based Computing-in-memory Chips with Residual Resource

Relaxation Signal Analysis and Optimization of Analog Resistive Random Access Memory for Neuromorphic Computing

Rescuing RRAM-Based Computing from Static and Dynamic Faults

Fast and Accurate Weight Updating Strategy for Resistive Random-Access Memory (Rram)-Based Neural Networks

Improvement of State Stability in Multi-Level Resistive Random-Access Memory (RRAM) Array for Neuromorphic Computing

Impacts of State Instability and Retention Failure of Filamentary Analog RRAM on the Performance of Deep Neural Network

Compensation Architecture Design Utilizing Residual Resource to Mitigate Impacts of Nonidealities in RRAM-based Computing-in-memory Chips

CLEAR: a Full-Stack Chip-in-loop Emulator for Analog RRAM Based Computing-in-memory System

Thermal Induced Retention Degradation of RRAM-based Neuromorphic Computing Chips

Investigation of Statistical Retention of Filamentary Analog RRAM for Neuromophic Computing

Enhance the Robustness to Time Dependent Variability of ReRAM-Based Neuromorphic Computing Systems with Regularization and 2R Synapse

Impact and Quantization of Short-Term Relaxation Effect in Analog RRAM

Accurate Program/Verify Schemes of Resistive Switching Memory (RRAM) for In-Memory Neural Network Circuits

Early-Stage Fluctuation in Low-Power Analog Resistive Memory: Impacts on Neural Network and Mitigation Approach