Technical Papers Archive

Towards Exascale Fully Relativistic Pseudopotential Density Functional Theory Calculations with Mixed-Precision Computation and Compressed Communication Using Residual-Based Subspace Iteration


Authors: Nikhil Kodali, Gourab Panigrahi, Nishant Gupta, Sundaresan GouriShanker, Kartick Ramakrishnan, and Rudra Panch (Indian Institute of Science); Sambit Das (University of Michigan); Vishwas Rao (Argonne National Laboratory); and Phani Motamarri (Indian Institute of Science)

Abstract: Materials exhibiting noncollinear magnetism or strong spin-orbit-coupling underpin many spintronic and topological applications, but their simulations require complex two-component spinors and costs substantially more than scalar density functional theory (DFT). We present a GPU-centric exascale finite-element DFT framework with noncollinear magnetism and spin-orbit-coupling, combining (1) adaptive higher-order finite-element discretization; (2) a matrix-free Poisson solver; (3) residual-based Chebyshev filtered subspace iteration (R-ChFSI) for sparse generalized eigenproblems; (4) R-ChFSI-enabled mixed-precision computation with block floating-point compressed MPI communication; and (5) communication-efficient band partitioning. R-ChFSI permits inexact matrix-multivector products and 4x–6.4x compression of communicated data relative to FP64 while preserving double-precision robustness and reducing computation and data movement. Results demonstrate strong scaling on Aurora and Frontier, with up to 3.4x faster Chebyshev filtering. On Aurora, total self-consistent-field (SCF) solve wall time reduces by 2x. An 800-node (9,600-GPU) calculation demonstrates fully relativistic pseudopotential DFT for ~80,000 electrons in under seven minutes per SCF iteration.


Presentation: file

Back to Technical Papers Archive Listing