Benchmarks
Each test measures something different. Start with quantum benchmarks explained, or go straight to a test.
Single operations
- Randomized benchmarking Benchmark
The average error per gate, found by running random sequences of gates of growing length and measuring how fast the results decay.
1 result
- Two-qubit gate fidelity Measurement
How often an operation on two qubits gives the right result. 99.9% means about one error in a thousand operations.
14 results
Whole machines
- Quantum Volume Benchmark
The largest square circuit, with as many layers as qubits, that a machine runs reliably; reported as 2 to the power of that size.
7 results
- QUOPS Benchmark
How large a random circuit a machine can run successfully (the QUOPS score) and how many such operations it completes per second (the QUOPS rate).
5 results
- Circuit layer operations per second Benchmark
How many layers of a standard test circuit a machine runs per second, including the time spent by the classical computers around it.
4 results
- Cross-entropy benchmarking Benchmark
How closely a machine's output from random circuits matches the ideal, expressed as a fidelity between 0 and 1.
1 result
- Layer fidelity (error per layered gate) Measurement
The error per two-qubit gate when many gates run at once across a chain of qubits, which captures interference between neighbouring qubits.
5 results
- Logical error rate Measurement
How often an error-corrected (logical) qubit still fails, usually per round of error correction.
5 results
- Physical qubits Measurement
How many working physical qubits the machine has.
6 results
- Logical qubits Measurement
How many error-corrected (logical) qubits the machine has run at once, as demonstrated in a published experiment.
4 results
Real tasks
- Algorithmic qubits Benchmark
The number of qubits a machine can use successfully across a set of standard algorithms, based on the QED-C application benchmarks.
3 results
- Q-score Benchmark
The largest Max-Cut optimisation problem, in number of variables, that a machine solves noticeably better than random guessing.
3 results