RushHour: A Dynamically Reconfigurable Lattice-Surgery Architecture

• URL: http://arxiv.org/abs/2608.18985v2
• Date: Mon, 24 Aug 2026 07:51:42 GMT
• Title: RushHour: A Dynamically Reconfigurable Lattice-Surgery Architecture
• Abstract summary: We present dynamic LS, which enables efficient reconfiguration of the ancilla space, just-in-time allocation of resource states, and dynamic rotations of logical qubits.<n>We realize dynamic LS with RushHour through a hardware-compiler co-design.<n>On the smallest chips, 86% of benchmarks run only with RushHour, while existing approaches require 1.2-3.5$times$ larger chips.
• Score: 40.10865338207471
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Practical fault-tolerant quantum computing (FTQC) requires efficient lattice surgery (LS), so that large algorithms fit on resource-constrained quantum chips. Existing approaches, however, are rigid: qubits, routing space, and resource states are allocated ahead of execution, which prevents running on small chips, leaves statically scheduled executions with large time overheads, and fixes each design at a single area of the space-time trade-off. We present dynamic LS, which enables efficient reconfiguration of the ancilla space, just-in-time allocation of resource states, and dynamic rotations of logical qubits, thereby spanning the entire space-time trade-off with a single, unified approach. We realize dynamic LS with RushHour through a hardware-compiler co-design: the RushHour ISA formalizes and programs our dynamic lattice model, the Lattice Management Unit abstracts dynamic lattice management and performs efficient lattice reconfiguration, and the RushHour Compiler compiles logical circuits for physical chips into optimized ISA programs while pipelining instructions. We evaluate RushHour against six state-of-the-art compilers and two resource models. On the smallest chips, 86% of benchmarks run only with RushHour, while existing approaches require 1.2-3.5$\times$ larger chips. On space-constrained early-FTQC chips, RushHour runs a median 2.3-7.2$\times$ faster than the best feasible alternative, while achieving results comparable to the state of the art on very large chips. RushHour's constructive results run 4.8$\times$ from an idealized-machine resource limit.

Related papers

• Architecting Early Fault Tolerant Neutral Atoms Systems with Quantum Advantage [3.470681946214809]
  We introduce a teleportation-based scheme that leverages the reconfigurable connectivity of neutral atoms to parallelize logical operations.<n>Our approach achieves up to textbf$mathbfsim 3 times$ speedup over extractor architectures at no extra space cost.<n>We find that our speedups still apply and report exact space-time cost along with success probabilities.
  arXiv  Detail & Related papers  (2026-04-21T17:57:43Z)
• General circuit compilation protocol into partially fault-tolerant quantum computing architecture [0.0]
  We propose a circuit execution protocol for the space-time efficient analog rotation (STAR) architecture.<n>Our protocol aims to reduce such time overhead by parallel trials of resource state creations and more frequent trials of joint measurements.
  arXiv  Detail & Related papers  (2026-03-18T07:08:56Z)
• AdaFuse: Accelerating Dynamic Adapter Inference via Token-Level Pre-Gating and Fused Kernel Optimization [84.25316984309725]
  Integration of dynamic, sparse structures with parameter-efficient adapters (e.g., LoRA) is a powerful technique for enhancing Large Language Models (LLMs)<n>Despite minimal increases in computational load, the computation latency often skyrockets, leading to decoding speeds slowing by over 2.5 times.<n>AdaFuse is a framework built on a tight co-design between the algorithm and the underlying hardware system to enable efficient dynamic adapter execution.
  arXiv  Detail & Related papers  (2026-03-12T12:46:42Z)
• Batched high-rate logical operations for quantum LDPC codes [2.722479714583866]
  High-rate quantum LDPC codes reduce memory overhead by densely packing many logical qubits into a single block of physical qubits.<n>We extend this concept to high-rate computation by constructing emphbatched fault-tolerant operations that apply the same logical gate across many code blocks in parallel.
  arXiv  Detail & Related papers  (2025-10-07T17:26:10Z)
• Flexiffusion: Training-Free Segment-Wise Neural Architecture Search for Efficient Diffusion Models [50.260693393896716]
  Diffusion models (DMs) are powerful generative models capable of producing high-fidelity images but constrained by high computational costs.<n>We propose Flexiffusion, a training-free NAS framework that jointly optimize generation schedules and model architectures without modifying pre-trained parameters.<n>Our work pioneers a resource-efficient paradigm for searching high-speed DMs without sacrificing quality.
  arXiv  Detail & Related papers  (2025-06-03T06:02:50Z)
• InTAR: Inter-Task Auto-Reconfigurable Accelerator Design for High Data Volume Variation in DNNs [5.762543012823378]
  InTAR is a novel accelerator design methodology for HDV applications on FPGAs.<n>It switches execution patterns automatically with a static schedule determined before circuit design.<n>InTAR achieves a high clock frequency with fewer resources and low reconfiguration time.
  arXiv  Detail & Related papers  (2025-02-12T21:43:51Z)
• FusionLLM: A Decentralized LLM Training System on Geo-distributed GPUs with Adaptive Compression [55.992528247880685]
  Decentralized training faces significant challenges regarding system design and efficiency. We present FusionLLM, a decentralized training system designed and implemented for training large deep neural networks (DNNs) We show that our system and method can achieve 1.45 - 9.39x speedup compared to baseline methods while ensuring convergence.
  arXiv  Detail & Related papers  (2024-10-16T16:13:19Z)
• Compilation of Trotter-Based Time Evolution for Partially Fault-Tolerant Quantum Computing Architecture [0.6449786007855248]
  We present an efficient method for simulating the time evolution of the 2D Hubbard model Hamiltonian. Our analysis reveals an acceleration of over 10 times compared to naive serial compilation. For devices with a physical error rate of $p_rm phys = 10-4$, we estimate that approximately $6.5 times 104$ physical qubits are required to achieve faster ground state energy estimation.
  arXiv  Detail & Related papers  (2024-08-27T10:07:34Z)
• Latency-aware Unified Dynamic Networks for Efficient Image Recognition [72.8951331472913]
  LAUDNet is a framework to bridge the theoretical and practical efficiency gap in dynamic networks. It integrates three primary dynamic paradigms-spatially adaptive computation, dynamic layer skipping, and dynamic channel skipping. It can notably reduce the latency of models like ResNet by over 50% on platforms such as V100,3090, and TX2 GPUs.
  arXiv  Detail & Related papers  (2023-08-30T10:57:41Z)
• RESPECT: Reinforcement Learning based Edge Scheduling on Pipelined Coral Edge TPUs [12.952987240366781]
  This work presents a reinforcement learning (RL) based scheduling framework, which learns the behaviors of optimal optimization algorithms. RL generates near-optimal scheduling results with short solving runtime overhead. Our framework has demonstrated up to $sim2.5times$ real-world on-chip runtime inference speedups over the commercial compiler.
  arXiv  Detail & Related papers  (2023-04-10T17:22:12Z)
• Towards Optimal VPU Compiler Cost Modeling by using Neural Networks to Infer Hardware Performances [58.720142291102135]
  'VPUNN' is a neural network-based cost model trained on low-level task profiling. It consistently outperforms the state-of-the-art cost modeling in Intel's line of VPU processors.
  arXiv  Detail & Related papers  (2022-05-09T22:48:39Z)
• CoSA: Scheduling by Constrained Optimization for Spatial Accelerators [1.9149970150912705]
  We present CoSA, a constrained-optimization-based approach for scheduling Deep Neural Networks (DNNs) accelerators. As opposed to existing approaches that either rely on designers's or iterative methods to navigate the search space, CoSA expresses scheduling decisions as a constrained-optimization problem. We demonstrate that CoSA-generated schedules significantly outperform state-of-the-art approaches by a geometric mean of up to 2.5x.
  arXiv  Detail & Related papers  (2021-05-05T07:17:25Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.