Saved in:
Bibliographic Details
Main Authors: Liu, Xianglin, Yang, Kai, Zhou, Fanli, Liu, Yongxiang, Chen, Hao, Zhang, Yijia, Fan, Dengdong, Li, Wenbo, Wang, Bingqiang, Zhang, Shixun, Xu, Pengxiang, Tian, Yonghong
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2604.08250
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908949484142592
author Liu, Xianglin
Yang, Kai
Zhou, Fanli
Liu, Yongxiang
Chen, Hao
Zhang, Yijia
Fan, Dengdong
Li, Wenbo
Wang, Bingqiang
Zhang, Shixun
Xu, Pengxiang
Tian, Yonghong
author_facet Liu, Xianglin
Yang, Kai
Zhou, Fanli
Liu, Yongxiang
Chen, Hao
Zhang, Yijia
Fan, Dengdong
Li, Wenbo
Wang, Bingqiang
Zhang, Shixun
Xu, Pengxiang
Tian, Yonghong
contents The rapid advancement of deep learning is reshaping the hardware design landscape toward AI tasks, posing fundamental challenges for HPC workloads such as atomistic simulation. Here we present SMC-AI, a general algorithmic framework that extends the SMC-X method for efficient canonical Monte Carlo simulation on AI accelerators, including GPUs and NPUs, while maintaining extreme scalability. The implementation of SMC-AI on an NPU cluster reaches unprecedented performance, achieving MC simulation of 4 trillion atoms on 4096 NPU dies. This represents the largest ML-accelerated atomistic simulation reported, delivering 32X system size and 1.3X throughput than previous records, with a relatively small computational budget. Excellent strong and weak scaling efficiency are reached for both the NPU and GPU implementation. By decoupling ML models from simulation, SMC-AI creates an abstraction that facilitates integration and porting of diverse ML models, laying a foundation for the future development of scalable scientific software.
format Preprint
id arxiv_https___arxiv_org_abs_2604_08250
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle SMC-AI: Scaling Monte Carlo Simulation to Four Trillion Atoms with AI Accelerators
Liu, Xianglin
Yang, Kai
Zhou, Fanli
Liu, Yongxiang
Chen, Hao
Zhang, Yijia
Fan, Dengdong
Li, Wenbo
Wang, Bingqiang
Zhang, Shixun
Xu, Pengxiang
Tian, Yonghong
Computational Physics
The rapid advancement of deep learning is reshaping the hardware design landscape toward AI tasks, posing fundamental challenges for HPC workloads such as atomistic simulation. Here we present SMC-AI, a general algorithmic framework that extends the SMC-X method for efficient canonical Monte Carlo simulation on AI accelerators, including GPUs and NPUs, while maintaining extreme scalability. The implementation of SMC-AI on an NPU cluster reaches unprecedented performance, achieving MC simulation of 4 trillion atoms on 4096 NPU dies. This represents the largest ML-accelerated atomistic simulation reported, delivering 32X system size and 1.3X throughput than previous records, with a relatively small computational budget. Excellent strong and weak scaling efficiency are reached for both the NPU and GPU implementation. By decoupling ML models from simulation, SMC-AI creates an abstraction that facilitates integration and porting of diverse ML models, laying a foundation for the future development of scalable scientific software.
title SMC-AI: Scaling Monte Carlo Simulation to Four Trillion Atoms with AI Accelerators
topic Computational Physics
url https://arxiv.org/abs/2604.08250