Sustained Quantum System Performance (SQSP)

• SQSP is a unified metric that quantifies the annual throughput of quantum systems by aggregating diverse benchmarks across materials science, quantum chemistry, and high-energy physics.
• It evaluates key device capabilities—qubit count, gate depth, and clock speed—while incorporating algorithmic innovations that reduce resource needs.
• SQSP informs system planning and vendor roadmaps by enabling direct comparisons of practical quantum performance for high-impact scientific workloads.

Sustained Quantum System Performance (SQSP) is a system-level metric that quantifies the throughput of a quantum computer when executing a heterogeneous workload of benchmark scientific applications. Designed as an analogue to classical metrics such as Sustained System Performance (SSP) used in high-performance computing, SQSP aggregates the execution rates across various quantum benchmarks—representing materials science, quantum chemistry, and high-energy physics—to yield a single figure that reflects how many quantum jobs can be completed in a given time window, such as one year. This approach captures not only device capabilities (qubit count, gate depth, clock speed) but also algorithmic advances that have reduced the quantum resources needed for challenging problems (Camps et al., 11 Sep 2025).

1. Definition and Calculation of SQSP

SQSP is defined for fault-tolerant quantum systems as the geometric mean of per-application throughput values, providing a robust singular metric for system-level performance assessment. For n benchmark quantum applications, each with throughput $T_i$ calculated as the ratio of a fixed time window (e.g., one year) to its estimated execution time $t_{\text{exe},i}$, the SQSP is given by:

$SQSP = \left( \prod_{i=1}^n T_i \right)^{1/n}$

where

$T_i = \frac{1\,\text{yr}}{t_{\text{exe},i}}$

$t_{\text{exe},i} = \frac{N_s \cdot n_G}{f}$

Here, $N_s$ is the number of circuit repetitions (shots) required to achieve the desired output fidelity, $n_G$ is the number of dominant quantum gates (e.g., T gates, Toffoli gates) in the circuit, and $f$ denotes the logical clock speed of the quantum processor (Camps et al., 11 Sep 2025). This formulation allows direct comparison of overall productivity between quantum systems regardless of technology or application domain.

2. Quantum Capabilities and Vendor Roadmaps

SQSP is sensitive to prevailing and projected hardware capabilities. Vendor technology roadmaps for platforms based on superconducting circuits, trapped ions, and neutral atoms project:

• Exponential increases in qubit counts, leading to more logical qubits capable of encoded fault-tolerant operations.
• Orders-of-magnitude growth in reliable gate depth, progressing from “megaquop” ($\sim10^6$ gates) to “gigaquop” ($\sim10^9$ gates) and “teraquop” ($t_{\text{exe},i}$0 gates).
• Faster clock speeds, ranging from kilohertz (kHz) for certain architectures up to gigahertz (GHz) for others.

These advancements directly reduce $t_{\text{exe},i}$1 (by increasing $t_{\text{exe},i}$2 and allowing higher $t_{\text{exe},i}$3) and thereby boost $t_{\text{exe},i}$4 for each application, resulting in higher overall SQSP. The anticipated convergence of technology capabilities and benchmark algorithm requirements within five to ten years predicts a dramatic improvement in system throughput for quantum scientific workloads (Camps et al., 11 Sep 2025).

3. Algorithmic Advances and Resource Compression

Recent quantum algorithmic innovations have substantially lowered the resource requirements for key scientific benchmarks. For example, ground state energy estimation (GSEE) for quantum chemistry targets such as FeMoco has seen a reduction in resource estimates by three orders of magnitude in gate count and fivefold in logical qubit count, mainly due to:

• Advanced Hamiltonian simulation methods (qubitization, quantum signal processing)
• Tensor factorization strategies (single/double factorization, tensor hypercontraction)

Because $t_{\text{exe},i}$5 depends critically on $t_{\text{exe},i}$6, these reductions translate directly to higher throughput and improved SQSP. Constant improvements in $t_{\text{exe},i}$7, via methods that require fewer shots per application (such as improved measurement and error mitigation protocols), also contribute to accelerated execution times across the benchmark suite (Camps et al., 11 Sep 2025).

4. Scientific Domains and Benchmark Selection

SQSP assessments focus on application benchmarks from the scientific domains most relevant to national research workloads:

Domain	Benchmark Focus	Resource Range (P-vector)
Materials Science/Condensed Matter	Spin models, magnetism	$t_{\text{exe},i}$8
Quantum Chemistry	Ground state energies, reaction pathways	$t_{\text{exe},i}$9
High Energy/Nuclear Physics	Lattice gauge simulations	$SQSP = \left( \prod_{i=1}^n T_i \right)^{1/n}$0

Where $SQSP = \left( \prod_{i=1}^n T_i \right)^{1/n}$1 denotes the spatial (logical qubit count) and temporal (gate count) requirements. SQSP synthesizes throughput across these varied domains, accounting for differences in circuit complexity, gate fidelity, shot count, and device speed.

5. Comparison to Classical System Performance Metrics

Unlike classical HPC throughput metrics, which are primarily measures of floating-point operation rates (FLOPS) and problem completion times in a homogeneous software landscape, SQSP addresses the unique heterogeneity of quantum workloads. Specifically, quantum workloads differ in circuit depth ($SQSP = \left( \prod_{i=1}^n T_i \right)^{1/n}$2), qubit width ($SQSP = \left( \prod_{i=1}^n T_i \right)^{1/n}$3), the need for repeated circuit execution ($SQSP = \left( \prod_{i=1}^n T_i \right)^{1/n}$4), and the constraint of device clock speed ($SQSP = \left( \prod_{i=1}^n T_i \right)^{1/n}$5). SQSP offers a unified measure to compare how quantum devices perform, over representative workloads, with respect to the practical time required to solve classically intractable scientific problems (Camps et al., 11 Sep 2025).

6. Impact and Strategic Use for System Planning

The introduction of SQSP supports quantitative comparisons for system procurement and deployment planning in science-driven quantum computing centers. By using formulas:

$SQSP = \left( \prod_{i=1}^n T_i \right)^{1/n}$6

$SQSP = \left( \prod_{i=1}^n T_i \right)^{1/n}$7

$SQSP = \left( \prod_{i=1}^n T_i \right)^{1/n}$8

decision-makers obtain a practical evaluation tool to weigh the productivity and sustained performance of candidate quantum systems under real, diverse scientific workloads. This enables predictive modeling of research output, guides hardware/software investment, and informs future benchmarking protocol refinement.

7. Future Directions

The projected improvements in hardware (higher logical qubit counts, faster gates, deeper circuits) and continued algorithmic innovation (lower $SQSP = \left( \prod_{i=1}^n T_i \right)^{1/n}$9, $T_i = \frac{1\,\text{yr}}{t_{\text{exe},i}}$0, and $T_i = \frac{1\,\text{yr}}{t_{\text{exe},i}}$1 per application) indicate that SQSP will rise by multiple orders of magnitude over the next decade. This suggests near-term quantum systems will become increasingly competitive and relevant for high-impact scientific computing tasks. A plausible implication is that SQSP will become an essential metric alongside classical benchmarks for quantifying system readiness, efficacy, and scientific value in hybrid quantum-classical computing environments (Camps et al., 11 Sep 2025).