Abstract:The demand for computation driven by machine learning and deep learning applications has experienced exponential growth over the past five years [1], leading to a significant surge in computing hardware products. Meanwhile, this rapid increase has exacerbated the memory wall bottleneck within mainstream Von Neumann architectures [2]. For instance, NVIDIA GPUs have gained nearly a 200x increase in fp32 computing power, transitioning from P100 to H100 in the last five years [3-4], accompanied by a mere 8x scaling in memory bandwidth. Addressing the need to mitigate data movement challenges, process-in-memory designs, especially ReRAM-based solutions, have emerged as compelling candidates [5-6]. However, this shift in hardware design poses distinct challenges at the design phase, given the limitations of existing hardware design tools. Popular design tools today can be used to characterize analog behavior via SPICE tools [7], system and logical behavior using Verilog tools [8], and mixed signal behavior through toolbox like CPPSIM [9]. Nonetheless, the design of in-memory computing systems, especially those involving non-CMOS devices, presents a unique need for characterizing mixed-signal computing behavior across a large number of cells within a memory bank. This requirement falls beyond the scope of conventional design tools. In this paper, we bridge this gap by introducing the ARAGORN framework — a GPU-accelerated mixed-signal transient simulator for analyzing ReRAM crossbar array. This tool facilitates the characterization of analog circuit and device behavior on a large scale, while also providing enhanced simulation performance for complex algorithm analysis, sign-off, and verification.

The Fast Simulation Model of SRAM

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

Latency Insertion Method for Accelerated Simulation of Memristor Crossbar Array in Neuromorphic Chip

Ultrafast Self-Rectifying Memristors for Advanced In-Memory Computing

Highly accurate power modeling method for SRAM structures with simple circuit simulation

A 512 Kb SRAM in 65nm CMOS with Divided Bitline and Novel Two-Stage Sensing Technique

FAST: A Fully-Concurrent Access Technique to All SRAM Rows for Enhanced Speed and Energy Efficiency in Data-Intensive Applications

An Analytical Delay Model for SRAM-based FPGA Interconnections

Fast Setup Time Characterization of Static Random Access Memory Based on Search-Delay

Fast Processor Simulation Model for SoC Function Verification

A High Performance Embedded Sram Compiler

Design Of A Dynamic Memory Access Scheduler

An Architecture-Level Cache Simulation Framework Supporting Advanced PMA STT-MRAM

-Memory Computing Based Reliable and High Speed Schmitt trigger 10T SRAM cell design

Quantitative modeling of racetrack memory, a tradeoff among area, performance, and power

Architecture-level energy model for high-capacity STT-MRAM memory

An Efficient SRAM Yield Analysis and Optimization Method with Adaptive Online Surrogate Modeling

Simulation of the RRAM based Nonvolatile SRAM cell

ReARTSim: an ReRAM array transient simulator with GPU optimized runtime Acceleration

Simulation Study of New 3-Terminal Devices for High Speed STT-RAM

An Improved Memory System Simulator Based on Dramsim2