Authors: Kien Nguyen Cong and Jonathan T. Willman (University of South Florida); Stan G. Moore (Sandia National Laboratories); Anatoly B. Belonoshko (KTH Royal Institute of Technology, Sweden); Rahulkumar Gayatri (Lawrence Berkeley National Laboratory (LBNL)); Evan Weinberg (NVIDIA Corporation); Mitchell A. Wood and Aidan P. Thompson (Sandia National Laboratories); and Ivan I. Oleynik (University of South Florida)
Abstract: Billion atom molecular dynamics (MD) using quantum-accurate machine-learned Spectral Neighbor Analysis Potential (SNAP) observed long-sought high pressure BC8 phase of carbon at extreme pressure (12 Mbar) and temperature (5,000 K). 24-hour, 4650-node production simulation on OLCF Summit demonstrated unprecedented scaling and unmatched real-world performance of SNAP MD while sampling one nanosecond physical time. Efficient implementation of SNAP force kernel in LAMMPS using the Kokkos CUDA backend on NVIDIA GPUs combined with excellent strong scaling (better than 97% parallel efficiency) enabled peak computing rate of 50.0 PFLOPS (24.9% of theoretical peak) for a 20 billion atom MD simulation on the full Summit machine (27,900 GPUs). The peak MD performance of 6.21 M atom steps/node-s is 22.9 times greater than a previous record for quantum-accurate MD. Near perfect weak scaling of SNAP MD highlights its excellent potential to advance the frontier of quantum-accurate MD to trillion atom simulations on upcoming exascale platforms.
Back to Technical Papers Archive Listing