Towards exascale fully relativistic pseudopotential density functional theory calculations enabled by mixed-precision computation and compressed-communication using residual based subspace iteration
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 (i) adaptive higher-order finite-element discretization, (ii) a matrix-free Poisson solver, (iii) residual-based Chebyshev filtered subspace iteration (R-ChFSI) for sparse generalized eigenproblems, (iv) R-ChFSI-enabled mixed precision computation with block floating-point compressed MPI communication, and (v) communication-efficient band partitioning. R-ChFSI permits inexact matrix-multivector products and 4×--6.4× 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 3.4× faster Chebyshev filtering. On Aurora, total self-consistent-field (SCF) solve wall time reduces by 2×. An 800-node (9,600-GPU) calculation demonstrates fully relativistic pseudopotential DFT for ∼80,000 electrons in under 7 minutes per SCF iteration.
Related
- High precision, fast computation and high accuracy? Linear scaling double hybrid DFT calculations with spin-polarization
- The Breakdown of Classical Minicrypt Equivalences in the Quantum-Computation Classical-Communication Model
- Concurrent Terahertz Spin Excitations and Phase Shift Control in Fe 4 Nb 2 O 9 : A Material for Synergizing Computation and Communication Technologies
- Origin of the Difference between Kohn-Sham and ∆SCF Band gaps for Delocalized Electron–Hole Excitations in Hybrid-Functional Supercell Calculations
- Accelerating real-time hybrid-functional calculations with learned ACE and symmetry-based FFT reuse
BibTeX
@misc{kodali2026towards,
title = {{Towards exascale fully relativistic pseudopotential density functional theory calculations enabled by mixed-precision computation and compressed-communication using residual based subspace iteration}},
author = {Nikhil Kodali and Gourab Panigrahi and Nishant Gupta and Kartick Ramakrishnan and Sundaresan G and Rudra Panch and Sambit Das and Vishwas Rao and Phani Motamarri},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2605.30128},
eprint = {2605.30128},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2605.30128},
}
RIS
TY - UNPB
TI - Towards exascale fully relativistic pseudopotential density functional theory calculations enabled by mixed-precision computation and compressed-communication using residual based subspace iteration
AU - Nikhil Kodali
AU - Gourab Panigrahi
AU - Nishant Gupta
AU - Kartick Ramakrishnan
AU - Sundaresan G
AU - Rudra Panch
AU - Sambit Das
AU - Vishwas Rao
AU - Phani Motamarri
PY - 2026
DA - 2026/09/10
PB - arXiv
DO - 10.48550/arxiv.2605.30128
UR - https://arxiv.org/abs/2605.30128
AB - 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 (i) adaptive higher-order finite-element discretization, (ii) a matrix-free Poisson solver, (iii) residual-based Chebyshev filtered subspace iteration (R-ChFSI) for sparse generalized eigenproblems, (iv) R-ChFSI-enabled mixed precision computation with block floating-point compressed MPI communication, and (v) communication-efficient band partitioning. R-ChFSI permits inexact matrix-multivector products and 4×--6.4× 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 3.4× faster Chebyshev filtering. On Aurora, total self-consistent-field (SCF) solve wall time reduces by 2×. An 800-node (9,600-GPU) calculation demonstrates fully relativistic pseudopotential DFT for ∼80,000 electrons in under 7 minutes per SCF iteration.
ER -