NeuroSim V1.5: Improved Software Backbone for Benchmarking Compute-in-Memory Accelerators With Device- and Circuit-Level Nonidealities
IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems · 2026-11 · 45 (11) · 5451-5464
DOI: 10.1109/tcad.2026.3674448 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems, 2026-11. Verfasst von James Read und 5 weiteren Autor:innen. Die Arbeit wurde bisher 7-mal zitiert (Crossref) und nennt 70 Literaturangaben. Gefördert von: Semiconductor Research Corporation.
Abstract (Originalsprache)
The exponential growth of artificial intelligence (AI) applications has strained conventional von Neumann architectures, where frequent data transfers between compute units and memory create significant energy and latency bottlenecks. Compute-in-memory (CIM) addresses this challenge by performing multiply–accumulate (MAC) operations directly in memory arrays, substantially reducing data movement. As Transformers increasingly dominate AI workloads, extending CIM frameworks to support these architectures becomes critical. In this work, we present NeuroSim V1.5, an integrated framework for co-optimizing accuracy and hardware efficiency in CIM accelerator design. Key contributions include: a hybrid ACIM/DCIM architecture enabling Transformer acceleration, with a case study comparing Vision Transformer (ViT) and CNN performance; an integrated co-optimization framework combining TensorRT-based quantization, flexible noise modeling, and circuit-level PPA estimation; expanded device support including nonvolatile capacitive memories; and up to $6.5\times $ faster runtime through GPU-accelerated simulation. Our ViT case study reveals that ViTs exhibit lower noise tolerance and area efficiency than CNNs of comparable size, suggesting CNNs remain well-suited for vision applications on CIM hardware. The hybrid architecture provides the foundation for future Transformer workloads, including large language models. All versions are available open-source at https://github.com/neurosim/NeuroSim
Förderung
- Semiconductor Research Corporation
Zitieren
APA
Read, J., Lee, M., Huang, W., Luo, Y., Lu, A., & Yu, S. (2026). NeuroSim V1.5: Improved Software Backbone for Benchmarking Compute-in-Memory Accelerators With Device- and Circuit-Level Nonidealities. IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems, 45(11), 5451-5464. https://doi.org/10.1109/tcad.2026.3674448
Vancouver
Read J, Lee M, Huang W, Luo Y, Lu A, Yu S. NeuroSim V1.5: Improved Software Backbone for Benchmarking Compute-in-Memory Accelerators With Device- and Circuit-Level Nonidealities. IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems. 2026;45(11):5451-5464. doi:10.1109/tcad.2026.3674448
BibTeX
@article{Read2026,
author = {Read, James and Lee, Ming-Yen and Huang, Wei-Hsing and Luo, Yuan-Chun and Lu, Anni and Yu, Shimeng},
title = {NeuroSim V1.5: Improved Software Backbone for Benchmarking Compute-in-Memory Accelerators With Device- and Circuit-Level Nonidealities},
journal = {IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems},
year = {2026},
volume = {45},
number = {11},
pages = {5451--5464},
doi = {10.1109/tcad.2026.3674448}
}
EUVIMED