Tutorial: Benchmark a Backend

Measure the performance of a quantum backend using standardized benchmark suites. Generate PDF reports with evidence.

Time: 15 minutes · Prerequisites: API key configured

---

Step 1 — Discover Available Backends

import Tabs from '@theme/Tabs'; import TabItem from '@theme/TabItem';

from softqcos_sdk import QCOSClient

client = QCOSClient()

backends = client.backends.list()
for b in backends:
    print(f"{b['name']:20s} | {b['num_qubits']} qubits | {b['status']}")

softqcos backends list

---

Step 2 — Check Available Benchmark Suites

suites = client.bench.suites()
for s in suites:
    print(f"{s['suite_id']:15s} | {s['description']}")

softqcos bench suites

Available suites include:

• standard — General quantum computing performance (20 circuits)
• volumetric — Quantum volume and related metrics
• application — Real-world algorithm benchmarks
• noise — Noise characterization

---

Step 3 — Run the Benchmark

report = client.bench.run(
    backend="aer_simulator",
    suite_id="standard",
    shots=4096,
    repetitions=5
)

print(f"Benchmark ID: {report['benchmark_id']}")
print(f"Score: {report['score']:.2f}")
print(f"Fidelity: {report['fidelity']:.4f}")
print(f"Depth ratio: {report['depth_ratio']:.2f}")

softqcos bench run aer_simulator --suite standard --shots 4096 --reps 5

---

Step 4 — Download the Report

pdf = client.bench.download(report["benchmark_id"])
with open("benchmark_report.pdf", "wb") as f:
    f.write(pdf)
print("Report saved to benchmark_report.pdf")

softqcos bench download bench_abc123 --output benchmark_report.pdf

---

Step 5 — Verify Benchmark Integrity

Benchmark results are evidence-backed — verify with:

verification = client.bench.verify(report["benchmark_id"])
print(f"Valid: {verification['valid']}")
print(f"Hash: {verification['hash']}")

softqcos bench verify bench_abc123
# Output: ✓ Benchmark evidence valid

---

Comparing Multiple Backends

Benchmark multiple backends and compare:

backends_to_test = ["aer_simulator", "ibm_brisbane", "ionq_harmony"]
reports = []

for backend in backends_to_test:
    report = client.bench.run(
        backend=backend,
        suite_id="standard",
        shots=4096,
        repetitions=3
    )
    reports.append(report)
    print(f"{backend:20s} | Score: {report['score']:.2f} | Fidelity: {report['fidelity']:.4f}")

# Sort by score
reports.sort(key=lambda r: r["score"], reverse=True)
print(f"\nBest: {reports[0]['backend']} (score: {reports[0]['score']:.2f})")

---

What's Next?

• Calibration Flow — Check device calibration before running
• ROI Analysis — Understand cost efficiency
• Bench API Reference — All bench methods