Abstract:SNNs are an active research domain towards energy efficient machine intelligence. Compared to conventional ANNs, SNNs use temporal spike data and bio-plausible neuronal activation functions such as Leaky-Integrate Fire/Integrate Fire (LIF/IF) for data processing. However, SNNs incur significant dot-product operations causing high memory and computation overhead in standard von-Neumann computing platforms. Today, In-Memory Computing (IMC) architectures have been proposed to alleviate the "memory-wall bottleneck" prevalent in von-Neumann architectures. Although recent works have proposed IMC-based SNN hardware accelerators, the following have been overlooked- 1) the adverse effects of crossbar non-ideality on SNN performance due to repeated analog dot-product operations over multiple time-steps, 2) hardware overheads of essential SNN-specific components such as the LIF/IF and data communication modules. To this end, we propose SpikeSim, a tool that can perform realistic performance, energy, latency and area evaluation of IMC-mapped SNNs. SpikeSim consists of a practical monolithic IMC architecture called SpikeFlow for mapping SNNs. Additionally, the non-ideality computation engine (NICE) and energy-latency-area (ELA) engine performs hardware-realistic evaluation of SpikeFlow-mapped SNNs. Based on 65nm CMOS implementation and experiments on CIFAR10, CIFAR100 and TinyImagenet datasets, we find that the LIF/IF neuronal module has significant area contribution (>11% of the total hardware area). We propose SNN topological modifications leading to 1.24x and 10x reduction in the neuronal module's area and the overall energy-delay-product value, respectively. Furthermore, in this work, we perform a holistic comparison between IMC implemented ANN and SNNs and conclude that lower number of time-steps are the key to achieve higher throughput and energy-efficiency for SNNs compared to 4-bit ANNs.

ACCLAIM: an End-to-End SystemC-AMS Simulation Framework for Analog In-Memory-Computing

An Overview of Computing-in-Memory Interfaces

CSA-CiM: Enhancing Multi-Functional Computing-in-Memory with Configurable Sense Amplifiers

Modeling and Benchmarking Computing-in-Memory for Design Space Exploration.

EasyACIM: An End-to-End Automated Analog CIM with Synthesizable Architecture and Agile Design Space Exploration

ASiM: Improving Transparency of SRAM-based Analog Compute-in-Memory Research with an Open-Source Simulation Framework

IMAC-Sim: A Circuit-level Simulator For In-Memory Analog Computing Architectures

A Python Framework for SPICE Circuit Simulation of In-Memory Analog Computing Circuits

A User-Friendly Fast and Accurate Simulation Framework for Non-Ideal Factors in Computing-in-Memory Architecture

A High-Throughput, High-Accuracy System-Level Simulation Framework for System on Chips

Benchmarking and modeling of analog and digital SRAM in-memory computing architectures

CIMulator: A Comprehensive Simulation Platform for Computing-In-Memory Circuit Macros with Low Bit-Width and Real Memory Materials

CAMASim: A Comprehensive Simulation Framework for Content-Addressable Memory based Accelerators

ALPINE: Analog In-Memory Acceleration with Tight Processor Integration for Deep Learning

A Non-Volatile Computing-In-Memory Framework with Margin Enhancement Based CSA and Offset Reduction Based ADC.

AIsim: Functional Simulator for Analog-to-Information Perceptual Systems

Analog or Digital In-memory Computing? Benchmarking through Quantitative Modeling

An Extensible Memory Simulation Framework for Chip Multi-Processors

A 3-8Bit Reconfigurable Hybrid ADC Architecture with Successive-approximation and Single-slope Stages for Computing in Memory.

SpikeSim: An end-to-end Compute-in-Memory Hardware Evaluation Tool for Benchmarking Spiking Neural Networks

CiMLoop: A Flexible, Accurate, and Fast Compute-In-Memory Modeling Tool