Abstract
Modular architectures are a promising approach to scaling quantum computers beyond the limits of monolithic designs. However, non-local communications between different quantum processors might significantly impact overall system performance. In this work, we investigate the role of the network infrastructure in modular quantum computing architectures, focusing on coherence loss due to communication constraints. We analyze the impact of classical network latency on quantum teleportation and identify conditions under which it becomes a bottleneck. Additionally, we study different network topologies and assess how communication resources affect the number and parallelization of inter-core communications. Finally, we conduct a full-stack evaluation of the architecture under varying communication parameters, demonstrating how these factors influence the overall system performance. The results show that classical communication does not become a bottleneck for systems exceeding one million qubits, given current technology assumptions, even with modest clock frequencies and parallel wired interconnects. Additionally, increasing quantum communication resources generally shortens execution time, although it may introduce additional communication overhead. The optimal number of quantum links between QCores depends on both the algorithm being executed and the chosen inter-core topology. Our findings offer valuable guidance for designing modular architectures, enabling scalable quantum computing.
Introduction
As quantum computing hardware advances, monolithic chip designs—where all qubits reside on a single substrate—are increasingly constrained by physical limitations such as crosstalk, wiring complexity, and cryogenic cooling capacities. To overcome these barriers, the field is shifting toward modular architectures that interconnect multiple smaller quantum processors (QCores) into a distributed network. While modularity offers a path to scaling, it introduces significant communication overhead. Inter-core interactions require a coordinated exchange of quantum entanglement and classical data, potentially impacting the overall fidelity and performance of the system.
The Challenge of Modular Scaling
In a modular system, executing a gate between qubits located in different cores necessitates non-local operations, typically implemented via quantum teleportation. This process requires two primary resources: a quantum link to distribute entangled Einstein-Podolsky-Rosen (EPR) pairs between cores, and a classical link to transmit measurement results (typically 2 bits) for state correction.
A critical design challenge is determining which network component—classical or quantum—will first become a performance bottleneck. If classical latency is too high, the quantum state may decohere while waiting for correction signals. Conversely, allocating too much physical space to communication hardware (QLinks) can reduce the number of qubits available for actual computation.
Methodology and Benchmarking
To evaluate these trade-offs, a full-stack simulation framework was developed to analyze modular architectures across various inter-core topologies, including Line, Ring, Star, Grid, and All-to-all. The framework utilizes an adapted Hungarian Qubit Assignment (HQA) algorithm to map virtual qubits from an algorithm to physical QCores, optimizing for minimal inter-core communication.
Performance was assessed using four standard benchmarks: Quantum Fourier Transform (QFT), Quantum Volume (QVol), GHZ State Preparation, and the Cuccaro Adder. Fidelity was estimated using an analytical model that accounts for gate-induced depolarizing noise and $T_1/T_2$ relaxation times. To account for future hardware improvements, the study introduced scaling factors ($\Delta_{time}$ and $\Delta_{improv}$) for gate speed and reliability.
Analysis of Communication Bottlenecks
Contrary to some expectations, the study found that classical networking is remarkably resilient. With current technology assumptions (e.g., a 100 MHz clock), classical latency does not become a dominant bottleneck for systems up to one million qubits. The primary constraint is instead the "qubit real estate crisis": every quantum link (QLink) requires dedicated communication qubits to serve as buffers for entangled pairs. Adding excessive parallel links can paradoxically increase total latency by forcing the mapping algorithm to spread computation across more cores, thereby increasing the communication burden.
Results and Topology Performance
The evaluation revealed a "coherence plateau" where network latency imposes a hard limit on fidelity, regardless of how fast or reliable individual gates become. In terms of topology, All-to-all connectivity generally yielded the highest fidelity by minimizing inter-core hops. However, for structured algorithms like QFT, optimized Grid topologies performed competitively. For a 256-qubit system distributed across 64-qubit cores, the optimal configuration often involved only one or two parallel QLinks; beyond this, the overhead of communication qubits outweighed the benefits of parallelization.
Conclusion
This research provides a comprehensive roadmap for the design of modular quantum computers. It demonstrates that while classical communication is currently sufficient for scaling, the optimization of quantum communication resources and inter-core topologies is vital. Future work will extend this model to include Quantum Error Correction (QEC) and entanglement distillation, addressing the next generation of challenges in building truly scalable, high-fidelity quantum systems.
Summary updated