Mixed-Precision In-Memory Computing

Abstract

As CMOS scaling reaches its technological limits, a radical departure from traditional von Neumann systems, which involve separate processing and memory units, is needed in order to significantly extend the performance of today's computers. In-memory computing is a promising approach in which nanoscale resistive memory devices, organized in a computational memory unit, are used for both processing and memory. However, to reach the numerical accuracy typically required for data analytics and scientific computing, limitations arising from device variability and non-ideal device characteristics need to be addressed. Here we introduce the concept of mixed-precision in-memory computing, which combines a von Neumann machine with a computational memory unit. In this hybrid system, the computational memory unit performs the bulk of a computational task, while the von Neumann machine implements a backward method to iteratively improve the accuracy of the solution. The system therefore benefits from both the high precision of digital computing and the energy/areal efficiency of in-memory computing. We experimentally demonstrate the efficacy of the approach by accurately solving systems of linear equations, in particular, a system of 5,000 equations using 998,752 phase-change memory devices.

References in corpus (7)

• Stochastic Memristive Devices for Computing and Neuromorphic Applications
• Memcomputing: a computing paradigm to store and process information on the same physical platform
• Temporal correlation detection using computational phase-change memory
• On dynamic network entropy in cancer
• Universal Memcomputing Machines
• Changing Computing Paradigms Towards Power Efficiency
• Mixed-precision training of deep neural networks using computational memory

Cited by in corpus (45)

• Parallel convolution processing using an integrated photonic tensor core
• Neuromorphic computing with multi-memristive synapses
• Accurate deep neural network inference using computational phase-change memory
• Long short-term memory networks in memristor crossbars
• Programmable Phase-change Metasurfaces on Waveguides for Multimode Photonic Convolutional Neural Network
• Photoferroelectric oxides
• Programmable black phosphorus image sensor for broadband optoelectronic edge computing
• Analog content addressable memories with memristors
• Magnetic Skyrmions for Unconventional Computing
• Mixed-precision deep learning based on computational memory
• One-step regression and classification with crosspoint resistive memory arrays
• A TinyML Platform for On-Device Continual Learning with Quantized Latent Replays
• Committee machines -- a universal method to deal with non-idealities in memristor-based neural networks
• Edge Deep Learning for Neural Implants
• Time complexity of in-memory solution of linear systems
• Bulk-Switching Memristor-based Compute-In-Memory Module for Deep Neural Network Training
• An Overview of In-memory Processing with Emerging Non-volatile Memory for Data-intensive Applications
• Analog Neural Computing with Super-resolution Memristor Crossbars
• Progressive amorphization of GeSbTe phase-change material under electron beam irradiation
• In-memory eigenvector computation in time O(1)
• Energy Efficient Learning with Low Resolution Stochastic Domain Wall Synapse Based Deep Neural Networks
• Noisy Machines: Understanding Noisy Neural Networks and Enhancing Robustness to Analog Hardware Errors Using Distillation
• Programmable wave-based analog computing machine: a metastructure that designs metastructures
• Harnessing Intrinsic Noise in Memristor Hopfield Neural Networks for Combinatorial Optimization
• Chiral Spin Textures for Next-Generation Memory and Unconventional Computing
• Uncovering Phase Change Memory Energy Limits by Sub-Nanosecond Probing of Power Dissipation Dynamics
• Energy Efficient In-memory Hyperdimensional Encoding for Spatio-temporal Signal Processing
• Skyrmion-mediated Nonvolatile Ternary Memory
• Benchmarking energy consumption and latency for neuromorphic computing in condensed matter and particle physics
• ALPINE: Analog In-Memory Acceleration with Tight Processor Integration for Deep Learning
• hxtorch: PyTorch for BrainScaleS-2 -- Perceptrons on Analog Neuromorphic Hardware
• TDO-CIM: Transparent Detection and Offloading for Computation In-memory
• Mixed-precision training of deep neural networks using computational memory
• 5 Parallel Prism: A topology for pipelined implementations of convolutional neural networks using computational memory
• Quantized Non-Volatile Nanomagnetic Synapse based Autoencoder for Efficient Unsupervised Network Anomaly Detection
• Non-thermal transport of energy driven by photoexcited carriers in switchable solid states of GeTe
• Synaptic modulation of conductivity and magnetism in a CoPt-based electrochemical transistor
• Network insensitivity to parameter noise via adversarial regularization
• CrossNAS: A Cross-Layer Neural Architecture Search Framework for PIM Systems
• Tree-based machine learning performed in-memory with memristive analog CAM
• Ultra-low power on-chip learning of speech commands with phase-change memories
• An In-Memory Analog Computing Co-Processor for Energy-Efficient CNN Inference on Mobile Devices
• Post-Moore Technologies for Plasma Simulation: A Community Roadmap
• The CAPSARII Approach to Cyber-Secure Wearable, Ultra-Low-Power Networked Sensors for Soldier Health Monitoring
• Efficient transformer adaptation for analog in-memory computing via low-rank adapters