POSTER: TENSORMD: Molecular Dynamics Simulation with Ab Initio Accuracy of 50 Billion Atoms
AI has been coupled into HPC in many scientific fields and demonstrated significantly enhanced performances. Among them, molecular dynamics simulation emerges as an important area that HPC+AI helps to investigate the physical properties at the atomic scale, with machine-learning interatomic potentials (MLIPs) being used. Until now, general-purpose machine-learning (ML) tools (e.g., TensorFlow) have been leveraged in MLIPs, but they are not perfectly matched with each other, since many optimization opportunities in MLIPs have been missed by ML tools. This inefficiency arises from the fact that HPC+AI applications work with far more computational complexity and range compared with pure AI scenarios, which general-purpose ML tools target. This paper has developed an MLIP, named TensorMD, independently from any ML tool, to enable a set of optimizations to be flexibly performed, which are otherwise impossible to be applied in ML tools. TensorMD has been evaluated on two supercomputers and scaled to $51.8$ billion atoms, i.e., $\sim 3\times$ compared with state-of-the-art. We further examine the reasons why optimizations successfully applied in TensorMD are beyond the capabilities of ML tools, providing insights into compiler optimization design for HPC+AI applications.
Sun 2 MarDisplayed time zone: Pacific Time (US & Canada) change
18:00 - 20:00 | |||
18:00 2hPoster | POSTER: A General and Scalable GCN Training Framework on CPU Supercomputers Main Conference Chen Zhuang Tokyo Institute of Technology, Riken Center for Computational Science, Peng Chen National Institute of Advanced Industrial Science and Technology, Xin Liu National Institute of Advanced Industrial Science & Technology, Rio Yokota Tokyo Institute of Technology, Nikoli Dryden Lawrence Livermore National Laboratory, Toshio Endo Tokyo Institute of Technology, Satoshi Matsuoka RIKEN, Mohamed Wahib RIKEN Center for Computational Science | ||
18:00 2hPoster | POSTER: Triangle Counting on Tensor Cores Main Conference | ||
18:00 2hPoster | POSTER: Minimizing speculation overhead in a parallel recognizer for regular texts Main Conference Angelo Borsotti Politecnico di Milano, Luca Breveglieri Politecnico di Milano, Stefano Crespi Reghizzi Politecnico di Milano and CNR-EIIT, Angelo Morzenti Politecnico di Milano | ||
18:00 2hPoster | POSTER: Boost Lock-free Queue and Stack with Batching Main Conference Ao Li Wuhan University, Wenhai Li Wuhan University, Yuan Chen Wuhan University, Lingfeng Deng Wuhan University | ||
18:00 2hPoster | POSTER: Frontier-guided Graph Reordering Main Conference Xinmiao Zhang SKLP, Institute of Computing Technology, CAS, Cheng Liu ICT CAS, Shengwen Liang SKLP, Institute of Computing Technology, CAS, Chenwei Xiong SKLP, Institute of Computing Technology, CAS, Yu Zhang School of Computer Science and Technology, Huazhong University of Science and Technology, Lei Zhang ICT CAS, Huawei Li SKLP, Institute of Computing Technology, CAS, Xiaowei Li SKLP, Institute of Computing Technology, CAS | ||
18:00 2hPoster | POSTER: Big Atomics and Fast Concurrent Hash Tables Main Conference Daniel Anderson Carnegie Mellon University, Guy E. Blelloch Carnegie Mellon University, USA, Siddhartha Jayanti Google Research | ||
18:00 2hPoster | POSTER: FastBWA: Practical and Cost-Efficient Genome Sequence Alignment Pipeline Main Conference Zhonghai Zhang Institute of Computing Technology, Chinese Academy of Sciences / University of Chinese Academy of Sciences, Yewen Li Institute of Computing Technology, Chinese Academy of Sciences / University of Chinese Academy of Sciences, Ke Meng Chinese Academy of Sciences, Chunming Zhang Institute of Computing Technology, Chinese Academy of Sciences, Guangming Tan Chinese Academy of Sciences(CAS) | ||
18:00 2hPoster | POSTER: Transactional Data Structures with Orthogonal Metadata Main Conference | ||
18:00 2hPoster | POSTER: High-performance Visual Semantics Compression for AI-Driven Science Main Conference Boyuan Zhang Indiana University, Luanzheng Guo Pacific Northwest National Laboratory, Jiannan Tian Indiana University, Jinyang Liu University of California, Riverside, Daoce Wang Indiana University, Fanjiang Ye Indiana University, Chengming Zhang University of Alabama, Jan Strube Pacific Northwest National Laboratory, Nathan R. Tallent Pacific Northwest National Laboratory, Dingwen Tao Institute of Computing Technology, Chinese Academy of Sciences | ||
18:00 2hPoster | POSTER: Magneto: Accelerating Parallel Structures in DNNs via Co-Optimization of Operators Main Conference Zhanyuan Di State Key Lab of Processors, Institute of Computing Technology, CAS, Leping Wang State Key Lab of Processors, Institute of Computing Technology, CAS, Beijing, Ziyi Ren State Key Lab of Processors, Institute of Computing Technology, CAS, En Shao State Key Lab of Processors, Institute of Computing Technology, CAS, Beijing, Jie Zhao Hunan University, Siyuan Feng Shanghai Jiao Tong University, Dingwen Tao Institute of Computing Technology, Chinese Academy of Sciences, Guangming Tan Chinese Academy of Sciences(CAS), Ninghui Sun State Key Laboratory of Computer Architecture, Institute of Computing Technology, Chinese Academy of Sciences, University of Chinese Academy of Sciences | ||
18:00 2hPoster | POSTER: TENSORMD: Molecular Dynamics Simulation with Ab Initio Accuracy of 50 Billion Atoms Main Conference Yucheng Ouyang Institute of Computing Technology, Chinese Academy of Sciences, Xin Chen Institute of Applied Physics and Computational Mathematics, Ying Liu Institute of Computing Technology, Chinese Academy of Sciences, Xin Chen , Honghui Shang Institute of Computing Technology, Chinese Academy of Sciences, Zhenchuan Chen Institute of Computing Technology, Chinese Academy of Sciences, Rongfen Lin National Research Center of Parallel Computer Engineering and Technology, Xingyu Gao Institute of Applied Physics and Computational Mathematics, Lifang Wang Institute of Applied Physics and Computational Mathematics, Fang Li National Research Center of Parallel Computer Engineering and Technology, Jiahao Shan Institute of Computing Technology, Chinese Academy of Sciences, Haifeng Song Institute of Applied Physics and Computational Mathematics, Huimin Cui Institute of Computing Technology, Chinese Academy of Sciences, Xiaobing Feng ICT CAS |