Large-scale quantum computers promise unprecedented computational power, but their susceptibility to noise necessitates quantum error correction (QEC) for fault-tolerant operation. Existing decoding methods, however, struggle to scale due to data backlog issues where the decoder can't keep pace with the incoming syndrome data, leading to exponential slowdown. This paper introduces a parallelized window decoding technique that mitigates this issue by processing syndrome data in parallel, achieving nearly arbitrary speed. While this introduces a polynomial slowdown in logical clock speed, it avoids the exponential slowdown of previous methods. Numerical simulations for the surface code demonstrate comparable logical fidelity to existing decoders, along with the predicted speedup.

Luka Skoric, Dan E. Browne, Kenton M. Barnes, Neil I. Gillespie, Earl T. Campbell