pith. sign in

arxiv: 2604.25976 · v1 · submitted 2026-04-28 · 🪐 quant-ph · cs.AR

No Tile Left Behind: Multiprogramming for Surface-Code Architectures

Archisman Ghosh , Avimita Chatterjee , Swaroop Ghosh This is my paper

Pith reviewed 2026-05-07 16:39 UTC · model grok-4.3

classification 🪐 quant-ph cs.AR
keywords fault-tolerant quantum computingmultiprogrammingsurface codequantum schedulingmagic statesClifford+Tquantum error correction
0
0 comments X

The pith

A scheduler for surface-code quantum architectures achieves 3.1 times speedup in multiprogramming by accounting for tiles, ancilla, and magic-state resources.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper develops a formal framework for running multiple programs at once on fault-tolerant quantum hardware built from surface codes. Unlike simpler NISQ systems, these machines require careful management of data tiles, ancilla tiles for error correction, and shared magic-state resources to prevent fragmentation and contention during concurrent execution. The authors formulate static allocation, extend it with hierarchy-aware policies for online and limited-resource settings, and generalize further to architectures with dynamic magic-state cultivation. Simulations on synthetic Clifford+T workloads show the scheduler delivers a normalized system speedup of 3.1x while improving over prior FTQC multiprogramming baselines by roughly 29 percent and keeping mean slowdown low. A sympathetic reader cares because this approach could let expensive large-scale quantum hardware handle more work efficiently as systems scale.

Core claim

The paper proposes a hierarchy-aware scheduling framework for FTQC multiprogramming that models the constraints of data tiles, ancilla tiles, and magic-state resources in surface-code architectures. It formulates the static allocation problem and extends it to online and cultivation-enabled settings, demonstrating through simulation on Clifford+T workloads a normalized system speedup of 3.1x over baselines while keeping mean slowdown low.

What carries the argument

Hierarchy-aware scheduling policies that handle limited resources, online admission decisions, and dynamic magic-state cultivation while respecting tile placement, connectivity, routing headroom, and shared support infrastructure.

If this is right

  • Multiprogramming becomes feasible on structured FTQC floorplans without rapid fragmentation of remaining space.
  • Overall system throughput rises as more programs execute concurrently on the same hardware.
  • Individual programs experience only low mean slowdown, preserving acceptable quality of service.
  • Dynamic magic-state generation integrates into the scheduler without major degradation in performance.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same modeling of tile and ancilla constraints could extend to other error-correcting codes that also impose structured layouts.
  • Hardware featuring extra ancilla pools or more flexible routing might show even larger gains than the simulated results.
  • Coupling the scheduler with circuit compilers that optimize for tile adjacency could increase the observed speedups.
  • Real application traces might expose contention patterns not present in the synthetic Clifford+T test cases.

Load-bearing premise

The synthetic Clifford+T workloads and the modeled tile, ancilla, and magic-state constraints accurately represent the behavior and resource demands of future real-world FTQC applications.

What would settle it

Running the proposed scheduler on physical surface-code hardware or with workloads drawn from practical algorithms such as Shor's factoring or variational quantum eigensolvers and measuring whether the 3.1x speedup and 29 percent improvement over baselines still appear.

Figures

Figures reproduced from arXiv: 2604.25976 by Archisman Ghosh, Avimita Chatterjee, Swaroop Ghosh.

Figure 1
Figure 1. Figure 1: Overview of the Proposed Multiprogramming Scheduler. Diagrammatic summary of our end-to-end framework for multiprogramming in surface￾code FTQC. Online workloads are processed by the proposed scheduler, which combines hardware abstraction, workload placement, resource arbitration, and magic-state cultivation scheduling. We present our proposed framework in Section III. The output is a mapped workload graph… view at source ↗
Figure 2
Figure 2. Figure 2: Hardware Abstraction. An example layout comprising four data tiles and four magic state ports. The FTQC abstraction for surface code architecture can be represented as a 2D nearest-neighbor graph, as shown. under which a fault-tolerant workload must execute. Prior work on surface-code architecture design has largely focused on efficiently executing a single fault-tolerant work￾load [15], [18]. These studie… view at source ↗
Figure 3
Figure 3. Figure 3: Workload Placement & Abstraction. A diagrammatic representation of placing two workloads W1 and W2 on the floorplan. In the workload decomposition, we show the core ancilla region of a workload, the primary scratchpad, and the secondary scratchpad allocated to a workload and shared by the other. the Steiner tree are Ri = Di ∪ {mi}, and each available ancilla vertex has a unit cost. By applying the standard… view at source ↗
Figure 4
Figure 4. Figure 4: Resource Arbitration. In this diagram, we represent an example of the three different resource-limited scenarios that can appear during concurrent execution of independent workloads. (1) depicts the scenario where the data tiles are limited. We see that W1 and W2 are scheduled, whereas W3 cannot be scheduled due to a lack of available data tiles. (2) represents the case of having limited magic state ports.… view at source ↗
Figure 5
Figure 5. Figure 5: The state machine governing the resource arbitration for limited resources under online job scheduling. The start state is the QUEUE where the job enters, and the end state is the COMPLETE. A job is held at the PARKED state when the ancilla resources are unavailable, and enters the READY state to get executed. the required ancilla is unavailable or conflicts with a higher￾priority grant, or park the worklo… view at source ↗
Figure 7
Figure 7. Figure 7: The state machine demonstrating the magic state cultivation abstraction in the floorplan. An ancilla tile, when idle, can either be claimed for ROUTING or be sent for CULTIVATING. If the cultivating magic state gets post-selected during cultivation, it gets back to the idle state; otherwise, it escapes when it reaches the desired fault distance and is ready to measure out the T gate. However, a claim for r… view at source ↗
Figure 6
Figure 6. Figure 6: Scheduling Cultivation. A time evolution diagram of the magic state cultivation-enabled architecture. W1 arrives after system initialization, occupies the two data tiles, and the rest of the ancilla (barring the core and primary scratchpad) is cultivating. When W2 arrives, the ancilla demand for measurement and routing increases, and the required ancilla tiles get removed from cultivation. On completion of… view at source ↗
Figure 8
Figure 8. Figure 8: Plots describing the RQ1. In (a) we observe the normalized throughput against random and naive baselines. In (b), we show the system trace of running a Balanced mix of workloads, depicting the total usable free space in the floorplan. Time is abstracted as 1 time step = 1 surface code cycle. framework provides a ∼ 29% increase in speedup compared to their Corner Greedy implementation and ∼ 34% increase com… view at source ↗
Figure 9
Figure 9. Figure 9: Plots describing RQ2. In (a) we show the ablation study on the four core components of our framework. In (b), we demonstrate the scalability of our framework with an increase in floorplan size. TABLE II COMPARISON OF THE PROPOSED IDEA WITH PRIOR RESEARCH Methodology: ILP-C [25] CG [25] Proposed Normalized Throughput: 2.32 2.4 3.1 the connected components of Gfree(t). We define the largest usable free-space… view at source ↗
Figure 10
Figure 10. Figure 10: Plots describing RQ3 and RQ4. In (a) we show the mean slowdown in our proposed method with the increase in the number of workloads. In (b) we show the mean wait time as a percentage of the total execution time of the workload with the increase in the number of workloads. In (c) and (d) we demonstrate the reduction in slowdown and increase in normalized throughput, respectively, when using cultivation-enab… view at source ↗
read the original abstract

Fault-tolerant quantum computing (FTQC) is emerging as the architectural regime in which practical large-scale quantum workloads will execute. In this setting, however, multiprogramming is no longer a matter of partitioning a flat pool of qubits. Quantum error correction exposes a structured floorplan of data tiles, ancilla tiles, and magic-state service resources, so concurrent execution must account for compact placement, connectivity, routing headroom, and shared support infrastructure. This makes FTQC multiprogramming fundamentally harder than its NISQ counterpart: admission decisions can fragment the remaining floorplan, conservative reservations can waste ancilla, and dynamic contention across data, ancilla, and magic-state resources can degrade both throughput and quality of service. In this work, we develop a formal framework for FTQC multiprogramming that captures these structural constraints and their runtime implications. We formulate the baseline static allocation problem, extend it to limited-resource and online settings through hierarchy-aware scheduling policies, and further generalize it to cultivation-enabled architectures with dynamic magic-state generation. Through simulation on synthetic Clifford+T workloads, the proposed scheduler achieves a normalized system speedup of 3.1x, improving over prior FTQC multiprogramming baselines by ~29% while maintaining low mean slowdown.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 1 minor

Summary. The paper develops a formal framework for multiprogramming in surface-code FTQC architectures that accounts for structured floorplans of data tiles, ancilla tiles, and magic-state resources. It formulates the static allocation problem, extends it to limited-resource and online settings via hierarchy-aware scheduling policies, and generalizes further to cultivation-enabled architectures with dynamic magic-state generation. Simulations on synthetic Clifford+T workloads are used to claim a 3.1x normalized system speedup with a ~29% improvement over prior FTQC multiprogramming baselines while maintaining low mean slowdown.

Significance. If the performance claims prove robust, the work would be significant for FTQC systems by providing the first structured treatment of tile fragmentation, routing headroom, and cross-resource contention that distinguish FTQC multiprogramming from NISQ approaches. The hierarchy-aware and cultivation-enabled policies represent a concrete advance over flat allocation models.

major comments (1)
  1. [Evaluation / Simulation Results] The central performance claims (3.1x normalized speedup and ~29% gain over baselines) rest entirely on simulation results for synthetic Clifford+T workloads under a modeled tile/ancilla/magic-state floorplan. No details are provided on workload generation, circuit-shape distributions, statistical error bars, or independent validation against mapped real algorithms (e.g., Shor or chemistry circuits). Because synthetic generation can understate irregular fragmentation or overstate magic-state reuse, the external validity of the reported advantage is not yet established and directly affects the load-bearing speedup claim.
minor comments (1)
  1. [Abstract] The abstract refers to 'prior FTQC multiprogramming baselines' without naming the specific prior schedulers or policies used for the 29% comparison; adding one sentence of identification would improve reproducibility.

Simulated Author's Rebuttal

1 responses · 1 unresolved

We thank the referee for the constructive feedback and for recognizing the potential significance of our framework for FTQC multiprogramming. We address the evaluation concerns point by point below and commit to targeted revisions that strengthen the presentation without altering the core claims.

read point-by-point responses

  1. Referee: The central performance claims (3.1x normalized speedup and ~29% gain over baselines) rest entirely on simulation results for synthetic Clifford+T workloads under a modeled tile/ancilla/magic-state floorplan. No details are provided on workload generation, circuit-shape distributions, statistical error bars, or independent validation against mapped real algorithms (e.g., Shor or chemistry circuits). Because synthetic generation can understate irregular fragmentation or overstate magic-state reuse, the external validity of the reported advantage is not yet established and directly affects the load-bearing speedup claim.

    Authors: We agree that greater transparency on the synthetic workload methodology is needed. In the revised manuscript we will add a dedicated subsection describing the workload generator: the parameterization of circuit shapes (gate-count and depth distributions drawn from Clifford+T ensembles), the sampling procedure used to produce varied fragmentation and contention scenarios, and the inclusion of 95% confidence intervals on all speedup and slowdown figures. These additions directly address the missing details on generation and error bars. On independent validation, our study deliberately employs synthetic workloads to enable systematic sweeps over parameters such as tile fragmentation and cross-resource contention that are hard to isolate when mapping specific real circuits. While we acknowledge that evaluating mapped instances of Shor’s algorithm or quantum-chemistry circuits would provide complementary evidence, performing those mappings and full simulations lies outside the scope of the present framework paper. We will therefore add an explicit limitations paragraph and a future-work statement rather than claim such validation is already present. revision: partial

standing simulated objections not resolved
  • Independent validation on mapped real algorithms (e.g., Shor or chemistry circuits) cannot be supplied in the current revision without substantial new simulation work.

Circularity Check

0 steps flagged

No circularity: performance claims rest on external simulation runs

full rationale

The paper defines a scheduling framework for surface-code multiprogramming and evaluates it via simulation on explicitly described synthetic Clifford+T workloads. The reported 3.1x normalized speedup and 29% improvement are measured outcomes of running the proposed policies against those workloads under modeled tile/ancilla/magic-state constraints; they are not obtained by fitting parameters to the target metric or by renaming inputs as predictions. No self-definitional equations, load-bearing self-citations, or ansatz smuggling appear in the derivation of the scheduler or its reported results. The evaluation chain remains independent of the claims it supports.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Only the abstract is available, so the precise free parameters, axioms, and invented entities cannot be enumerated; the framework appears to rely on standard surface-code layout assumptions and synthetic workload models whose details are not visible.

pith-pipeline@v0.9.0 · 5520 in / 1216 out tokens · 36080 ms · 2026-05-07T16:39:22.984378+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

32 extracted references · 32 canonical work pages

  1. [1]

    High-threshold and low-overhead fault-tolerant quantum memory,

    S. Bravyi, A. W. Cross, J. M. Gambetta, D. Maslov, P. Rall, and T. J. Yoder, “High-threshold and low-overhead fault-tolerant quantum memory,”Nature, vol. 627, p. 778–782, Mar. 2024

    work page 2024

  2. [2]

    Magic state cultivation: growing t states as cheap as cnot gates,

    C. Gidney, N. Shutty, and C. Jones, “Magic state cultivation: growing t states as cheap as cnot gates,” 2024

    work page 2024

  3. [3]

    Atom Computing product page for the AC1000 system; accessed April 7, 2026

    Atom Computing, “Ac1000,” 2026. Atom Computing product page for the AC1000 system; accessed April 7, 2026

    work page 2026

  4. [4]

    2026 in quantum: A strategic preview from atom computing and partners,

    Atom Computing, “2026 in quantum: A strategic preview from atom computing and partners,” Jan. 2026. Atom Computing Tech Perspectives, published January 7, 2026; accessed April 7, 2026

    work page 2026

  5. [5]

    How ibm will build the world’s first large-scale, fault- tolerant quantum computer,

    R. Mandelbaum, J. Gambetta, J. Chow, T. Mittal, T. J. Yoder, A. Cross, and M. Steffen, “How ibm will build the world’s first large-scale, fault- tolerant quantum computer,” June 2025. IBM Quantum Blog, published June 10, 2025; accessed April 7, 2026

    work page 2025

  6. [6]

    Surface codes: Towards practical large-scale quantum computation,

    A. G. Fowler, M. Mariantoni, J. M. Martinis, and A. N. Cleland, “Surface codes: Towards practical large-scale quantum computation,”Physical Review A, vol. 86, Sept. 2012

    work page 2012

  7. [7]

    Surface code quantum computing by lattice surgery,

    D. Horsman, A. G. Fowler, S. Devitt, and R. V . Meter, “Surface code quantum computing by lattice surgery,”New Journal of Physics, vol. 14, p. 123011, Dec. 2012

    work page 2012

  8. [8]

    Lattice surgery with a twist: Simplifying clifford gates of surface codes,

    D. Litinski and F. v. Oppen, “Lattice surgery with a twist: Simplifying clifford gates of surface codes,”Quantum, vol. 2, p. 62, May 2018

    work page 2018

  9. [9]

    A game of surface codes: Large-scale quantum computing with lattice surgery,

    D. Litinski, “A game of surface codes: Large-scale quantum computing with lattice surgery,”Quantum, vol. 3, p. 128, Mar. 2019

    work page 2019

  10. [10]

    Design automation in quantum error correction,

    A. Ghosh, A. Chatterjee, and S. Ghosh, “Design automation in quantum error correction,” 2025

    work page 2025

  11. [11]

    Lsqca: Resource-efficient load/store architecture for limited-scale fault- tolerant quantum computing,

    T. Kobori, Y . Suzuki, Y . Ueno, T. Tanimoto, S. Todo, and Y . Tokunaga, “Lsqca: Resource-efficient load/store architecture for limited-scale fault- tolerant quantum computing,” in2025 IEEE International Symposium on High Performance Computer Architecture (HPCA), p. 304–320, IEEE, Mar. 2025

    work page 2025

  12. [12]

    Surface code compila- tion via edge-disjoint paths,

    M. Beverland, V . Kliuchnikov, and E. Schoute, “Surface code compila- tion via edge-disjoint paths,”PRX Quantum, vol. 3, May 2022

    work page 2022

  13. [13]

    Toward designing workload- aware surface code architectures,

    A. Ghosh, A. Chatterjee, and S. Ghosh, “Toward designing workload- aware surface code architectures,” 2026

    work page 2026

  14. [14]

    Lattice surgery for dummies,

    A. Chatterjee, S. Das, and S. Ghosh, “Lattice surgery for dummies,” Sensors, vol. 25, p. 1854, Mar. 2025

    work page 2025

  15. [15]

    Dependency- aware compilation for surface code quantum architectures,

    A. Molavi, A. Xu, S. Tannu, and A. Albarghouthi, “Dependency- aware compilation for surface code quantum architectures,”Proc. ACM Program. Lang., vol. 9, Apr. 2025

    work page 2025

  16. [16]

    Magic state distillation: Not as costly as you think,

    D. Litinski, “Magic state distillation: Not as costly as you think,” Quantum, vol. 3, p. 205, Dec. 2019

    work page 2019

  17. [17]

    Magic-state distillation with low overhead,

    S. Bravyi and J. Haah, “Magic-state distillation with low overhead,” Physical Review A, vol. 86, Nov. 2012

    work page 2012

  18. [18]

    Multi-qubit lattice surgery scheduling,

    A. Silva, X. Zhang, Z. Webb, M. Kramer, C.-W. Yang, X. Liu, J. Lemieux, K.-W. Chen, A. Scherer, and P. Ronagh, “Multi-qubit lattice surgery scheduling,” vol. 310, pp. 1:1–1:22, Schloss Dagstuhl – Leibniz- Zentrum f ¨ur Informatik, 2024

    work page 2024

  19. [19]

    Synchronization for fault-tolerant quantum computers,

    S. Maurya and S. Tannu, “Synchronization for fault-tolerant quantum computers,” inProceedings of the 52nd Annual International Sympo- sium on Computer Architecture, ISCA ’25, (New York, NY , USA), p. 1370–1385, Association for Computing Machinery, 2025

    work page 2025

  20. [20]

    Swiper: Minimizing fault-tolerant quantum program latency via speculative window decoding,

    J. Viszlai, J. D. Chadwick, S. Joshi, G. S. Ravi, Y . Li, and F. T. Chong, “Swiper: Minimizing fault-tolerant quantum program latency via speculative window decoding,” inProceedings of the 52nd Annual International Symposium on Computer Architecture, ISCA ’25, (New York, NY , USA), p. 1386–1401, Association for Computing Machinery, 2025

    work page 2025

  21. [21]

    Better than worst-case decoding for quantum error correction,

    G. S. Ravi, J. M. Baker, A. Fayyazi, S. F. Lin, A. Javadi-Abhari, M. Pedram, and F. T. Chong, “Better than worst-case decoding for quantum error correction,” 2022

    work page 2022

  22. [22]

    Scheduling lattice surgery with magic state cultivation,

    S. Hofmeyr, M. Weiden, J. Kalloor, J. Kubiatowicz, and C. Iancu, “Scheduling lattice surgery with magic state cultivation,” 2026

    work page 2026

  23. [23]

    An approximate solution for steiner problem in graphs,

    H. Takahashi, “An approximate solution for steiner problem in graphs,” Math. Japonica, vol. 24, no. 6, pp. 573–577, 1980

    work page 1980

  24. [24]

    Magic state cultivation on a superconducting quantum processor,

    E. Rosenfeld, C. Gidney, G. Roberts, A. Morvan, N. Lacroix, D. Kafri, J. Marshall, M. Li, V . Sivak, D. Abanin, A. Abbas, R. Acharya, L. A. Beni, G. Aigeldinger, R. Alcaraz, S. Alcaraz, T. I. Andersen, M. Ansmann, F. Arute, K. Arya, W. Askew, N. Astrakhantsev, J. Atalaya, R. Babbush, B. Ballard, J. C. Bardin, H. Bates, A. Bengtsson, M. B. Karimi, A. Bilme...

    work page 2025

  25. [25]

    Online job scheduler for fault-tolerant quantum multiprogramming,

    R. Wakizaka, S. Nishio, D. Sakuma, Y . Ueno, and Y . Suzuki, “Online job scheduler for fault-tolerant quantum multiprogramming,” in2025 IEEE International Conference on Quantum Computing and Engineering (QCE), p. 779–790, IEEE, Aug. 2025

    work page 2025

  26. [26]

    High-performance and scalable fault-tolerant quantum computation with lattice surgery on a 2.5d architecture,

    Y . Ueno, T. Saito, T. Tanimoto, Y . Suzuki, Y . Tabuchi, S. Tamate, and H. Nakamura, “High-performance and scalable fault-tolerant quantum computation with lattice surgery on a 2.5d architecture,” 2024

    work page 2024

  27. [27]

    A case for multi- programming quantum computers,

    P. Das, S. S. Tannu, P. J. Nair, and M. Qureshi, “A case for multi- programming quantum computers,” inProceedings of the 52nd Annual IEEE/ACM International Symposium on Microarchitecture, MICRO- 52, (New York, NY , USA), p. 291–303, Association for Computing Machinery, 2019

    work page 2019

  28. [28]

    A new qubits mapping mechanism for multi- programming quantum computing,

    X. Dou and L. Liu, “A new qubits mapping mechanism for multi- programming quantum computing,” inProceedings of the ACM Inter- national Conference on Parallel Architectures and Compilation Tech- niques, PACT ’20, (New York, NY , USA), p. 349–350, Association for Computing Machinery, 2020

    work page 2020

  29. [29]

    Simultaneous execution of quantum circuits on current and near-future nisq systems,

    Y . Ohkura, T. Satoh, and R. Van Meter, “Simultaneous execution of quantum circuits on current and near-future nisq systems,”IEEE Transactions on Quantum Engineering, vol. 3, p. 1–10, 2022

    work page 2022

  30. [30]

    Adaptive job and resource management for the growing quantum cloud,

    G. S. Ravi, K. N. Smith, P. Murali, and F. T. Chong, “Adaptive job and resource management for the growing quantum cloud,” 2022

    work page 2022

  31. [31]

    Enabling multi-programming mechanism for quantum computing in the nisq era,

    S. Niu and A. Todri-Sanial, “Enabling multi-programming mechanism for quantum computing in the nisq era,”Quantum, vol. 7, p. 925, Feb. 2023

    work page 2023

  32. [32]

    Qos: A quantum operating system,

    E. Giortamis, F. Rom ˜ao, N. Tornow, and P. Bhatotia, “Qos: A quantum operating system,” 2025

    work page 2025