Characterizing the spatio-temporal qubit traffic of a quantum intranet aiming at modular quantum computer architectures

Santiago Rodrigo, Domenico Spanò, Medina Bandić, Sergi Abadal, Hans van Someren, Anabel Ovide, Sebastian Feld, Carmen G. Almudéver, Eduard Alarcón

TL;DR

This paper introduces a specialized software tool for analyzing "qubit traffic"—the movement of quantum information—within modular, multi-core quantum processors. By characterizing how different algorithms utilize inter-core communication over time and space, the authors provide a framework to optimize the interconnects and compilers essential for scaling quantum computers beyond the single-chip limit.

What Problem Does This Paper Address?

Scaling quantum computers is a massive engineering hurdle. While single-chip processors are improving, they struggle with noise, decoherence, and the complexity of per-qubit control as qubit counts rise. A promising alternative is the modular quantum architecture, or "quantum intranet," where multiple small, manageable quantum chips (cores) are interconnected.

However, moving qubits between these cores—typically via a process called quantum teleportation—is significantly slower (5x to 100x) and more resource-intensive than local operations. To design efficient modular systems, researchers need to understand the spatio-temporal qubit traffic: when do qubits move, where are the bottlenecks (hotspots), and how much "idle time" is wasted waiting for communication? This paper fills that gap by providing a tool to visualize and quantify these patterns.

Methods

The authors developed a software toolchain that treats quantum traffic analysis similarly to classical Network-on-Chip (NoC) design. The process involves:

  1. Input Generation: Taking a quantum circuit and a target multi-core architecture.
  2. Compilation: Using the OpenQL compiler and a modified Qmap mapper to schedule gates and handle qubit movement across cores.
  3. Trace Extraction: Parsing the compiled code to create a detailed "life history" for every qubit, tracking its location and the operations it performs.
  4. Metric Extraction: Calculating specific performance metrics, such as:
  5. Temporal Burstiness: How much communication spikes at specific moments.
  6. Spatial Hotspotness: Which cores or links are most heavily utilized.
  7. Execution Distribution: The ratio of time spent on computation vs. communication vs. idling.

Datasets

The tool was evaluated using a suite of five representative quantum benchmarks:

  • Quantum Fourier Transform (QFT): A fundamental building block for many algorithms like Shor’s.
  • Grover’s Search: A standard algorithm for unstructured database searching.
  • Cuccaro Adder: A specific circuit for quantum arithmetic.
  • Quantum Volume: Synthetic circuits used to measure the holistic performance of quantum systems.
  • Random Circuits: High-density circuits used to stress-test the architecture's limits.

Key Insights and Ideas

  • Predictable Traffic: Unlike classical systems, where traffic depends on runtime data and cache misses, quantum traffic is largely deterministic. Because the sequence of gates is known at compile-time, communication patterns can be predicted and optimized before the program even runs.
  • The "Idle" Bottleneck: Because inter-core teleportation is so slow compared to local gates, qubits spend a vast majority of their time idling. In the Cuccaro adder example, only about 10% of the execution time featured simultaneous computation and communication.
  • Virtual vs. Physical Mapping: The tool allows researchers to see the "life" of a virtual qubit. Some qubits are central to an algorithm and stay active throughout, while others are transient. Mapping these "high-traffic" virtual qubits to specific physical locations is critical for reducing overhead.

Key Results

  • Burstiness vs. Hotspotness: The authors found a clear trade-off. Random and Quantum Volume circuits have low burstiness but high, uniform traffic. Conversely, Cuccaro and Grover algorithms exhibit high initial bursts of communication that can overwhelm a network if not properly managed.
  • Core Hotspots: In algorithms like QFT, specific cores (often Cores 0 and 1) become hotspots because they accumulate results. This suggests that future modular architectures might benefit from non-uniform designs or specialized "hub" cores.
  • Scaling Trends: The study showed that while random circuits scale their communication needs linearly with the number of cores, structured algorithms like QFT are more sensitive to how the compiler maps qubits to the hardware.

Limitations and Future Work

The current analysis assumes a deterministic and error-free communication model. In reality, quantum teleportation is stochastic and prone to errors, which would introduce unpredictable latencies.

The authors identify several paths for future research:

  • Integrating decoherence and error models to see how traffic delays impact the final fidelity of the computation.
  • Exploring different interconnect topologies (e.g., rings, meshes, or star topologies) to see which best suits specific algorithm types.
  • Developing traffic-aware compilers that use the tool's metrics to automatically minimize hotspots and burstiness during the mapping process.

Summary updated

Abstract

Quantum many-core processors are envisioned as the ultimate solution for the scalability of quantum computers. Based upon Noisy Intermediate-Scale Quantum (NISQ) chips interconnected in a sort of quantum intranet, they enable large algorithms to be executed on current and close future technology. In order to optimize such architectures, it is crucial to develop tools that allow specific design space explorations. To this aim, in this paper we present a technique to perform a spatio-temporal characterization of quantum circuits running in multi-chip quantum computers. Specifically, we focus on the analysis of the qubit traffic resulting from operations that involve qubits residing in different cores, and hence quantum communication across chips, while also giving importance to the amount of intra-core operations that occur in between those communications. Using specific multi-core performance metrics and a complete set of benchmarks, our analysis showcases the opportunities that the proposed approach may provide to guide the design of multi-core quantum computers and their interconnects.
BibTeX
@inproceedings{rodrigo2022characterizing,
  author = {Santiago Rodrigo and Domenico Spanò and Medina Bandić and Sergi Abadal and Hans van Someren and Anabel Ovide and Sebastian Feld and Carmen G. Almudéver and Eduard Alarcón},
  title = {{Characterizing the spatio-temporal qubit traffic of a quantum intranet aiming at modular quantum computer architectures}},
  booktitle = {NANOCOM},
  year = {2022},
  url = {https://openreview.net/forum?id=g46SpE05ZK},
}
Figure