We report atomistic simulations of subcritical crack growth rates in silica glass that quantitatively align with experimental data. This result was achieved by a careful selection of simulation cell geometry and boundary conditions, a novel crack tip tracking algorithm, a robust interatomic potential that balances accuracy with computational efficiency, and high-performance parallel computing algorithms and resources. The simulations reveal a succession of burst-arrest events that produce subcritical crack growth across a broad range of rates, extending up to kilometers per second.