Qiskit vs Qulacs
The same circuits, one host
Qiskit and Qulacs both finished 121 of the 175 workloads measured here. Qiskit was faster in 0 of them and Qulacs in 121. At the median Qulacs was 6.6× faster.
By circuit family
Who was faster where
| Circuit family | Workloads both finished | Qiskit faster in | Qulacs faster in | Faster at the median |
|---|---|---|---|---|
| Workloads both finished 10 | Qiskit faster in 0 | Qulacs faster in 10 | Faster at the median Qulacs 5.5× | |
| Workloads both finished 18 | Qiskit faster in 0 | Qulacs faster in 18 | Faster at the median Qulacs 9.7× | |
| Workloads both finished 21 | Qiskit faster in 0 | Qulacs faster in 21 | Faster at the median Qulacs 6.0× | |
| Workloads both finished 10 | Qiskit faster in 0 | Qulacs faster in 10 | Faster at the median Qulacs 8.8× | |
| Workloads both finished 18 | Qiskit faster in 0 | Qulacs faster in 18 | Faster at the median Qulacs 6.4× | |
| Workloads both finished 8 | Qiskit faster in 0 | Qulacs faster in 8 | Faster at the median Qulacs 11.9× | |
| Workloads both finished 14 | Qiskit faster in 0 | Qulacs faster in 14 | Faster at the median Qulacs 7.7× | |
| Workloads both finished 10 | Qiskit faster in 0 | Qulacs faster in 10 | Faster at the median Qulacs 5.2× | |
| Workloads both finished 12 | Qiskit faster in 0 | Qulacs faster in 12 | Faster at the median Qulacs 3.7× |
Local circuit dynamics
Brickwork nearest-neighbour layers: lattice models, quench dynamics
| Workload | Qubits | Qiskit | Qulacs | Faster |
|---|---|---|---|---|
Expectation value, local observable, depth 3 | Qubits 4 | Qiskit 865 µs | Qulacs 32.1 µs | Faster Qulacs 27.0× |
Expectation value, local observable, depth 3 | Qubits 8 | Qiskit 1.80 ms | Qulacs 73.1 µs | Faster Qulacs 24.7× |
Expectation value, local observable, depth 3 | Qubits 12 | Qiskit 3.75 ms | Qulacs 585 µs | Faster Qulacs 6.4× |
Expectation value, local observable, depth 3 | Qubits 16 | Qiskit 35.7 ms | Qulacs 10.7 ms | Faster Qulacs 3.3× |
Expectation value, local observable, depth 3 | Qubits 20 | Qiskit 936 ms | Qulacs 206 ms | Faster Qulacs 4.5× |
Expectation value, local observable, depth 3 | Qubits 24 | Qiskit skip | Qulacs 4.76 s | Faster – |
Expectation value, global observable (all_z), depth 3 | Qubits 4 | Qiskit 996 µs | Qulacs 42.5 µs | Faster Qulacs 23.4× |
Expectation value, global observable (all_z), depth 3 | Qubits 8 | Qiskit 2.11 ms | Qulacs 119 µs | Faster Qulacs 17.7× |
Expectation value, global observable (all_z), depth 3 | Qubits 12 | Qiskit 4.26 ms | Qulacs 1.38 ms | Faster Qulacs 3.1× |
Expectation value, global observable (all_z), depth 3 | Qubits 16 | Qiskit 39.7 ms | Qulacs 26.2 ms | Faster Qulacs 1.5× |
Expectation value, global observable (all_z), depth 3 | Qubits 20 | Qiskit 947 ms | Qulacs 535 ms | Faster Qulacs 1.8× |
Expectation value, global observable (all_z), depth 3 | Qubits 24 | Qiskit skip | Qulacs 10.7 s | Faster – |
Gradient, local observable, depth 3 | Qubits 4 | Qiskit unsupported | Qulacs 117 µs | Faster – |
Gradient, local observable, depth 3 | Qubits 8 | Qiskit unsupported | Qulacs 307 µs | Faster – |
Gradient, local observable, depth 3 | Qubits 12 | Qiskit unsupported | Qulacs 3.00 ms | Faster – |
Gradient, local observable, depth 3 | Qubits 16 | Qiskit unsupported | Qulacs 59.6 ms | Faster – |
Hamiltonian time evolution
Trotterised spin chain: ZZ + XX couplings with local fields
| Workload | Qubits | Qiskit | Qulacs | Faster |
|---|---|---|---|---|
Expectation value, local observable, depth 3 | Qubits 4 | Qiskit 3.80 ms | Qulacs 104 µs | Faster Qulacs 36.4× |
Expectation value, local observable, depth 3 | Qubits 8 | Qiskit 10.1 ms | Qulacs 276 µs | Faster Qulacs 36.5× |
Expectation value, local observable, depth 3 | Qubits 12 | Qiskit 24.5 ms | Qulacs 1.85 ms | Faster Qulacs 13.3× |
Expectation value, local observable, depth 3 | Qubits 16 | Qiskit 259 ms | Qulacs 34.0 ms | Faster Qulacs 7.6× |
Expectation value, local observable, depth 3 | Qubits 20 | Qiskit 6.72 s | Qulacs 711 ms | Faster Qulacs 9.4× |
Expectation value, local observable, depth 3 | Qubits 24 | Qiskit skip | Qulacs 21.1 s | Faster – |
Expectation value, global observable (all_z), depth 3 | Qubits 4 | Qiskit 3.88 ms | Qulacs 113 µs | Faster Qulacs 34.4× |
Expectation value, global observable (all_z), depth 3 | Qubits 8 | Qiskit 10.6 ms | Qulacs 321 µs | Faster Qulacs 33.1× |
Expectation value, global observable (all_z), depth 3 | Qubits 12 | Qiskit 25.7 ms | Qulacs 2.51 ms | Faster Qulacs 10.3× |
Expectation value, global observable (all_z), depth 3 | Qubits 16 | Qiskit 262 ms | Qulacs 48.2 ms | Faster Qulacs 5.4× |
Expectation value, global observable (all_z), depth 3 | Qubits 20 | Qiskit 7.22 s | Qulacs 1.02 s | Faster Qulacs 7.1× |
Expectation value, global observable (all_z), depth 3 | Qubits 24 | Qiskit skip | Qulacs 26.9 s | Faster – |
Expectation value, global observable (pauli_mix), depth 3 | Qubits 8 | Qiskit 10.7 ms | Qulacs 349 µs | Faster Qulacs 30.7× |
Expectation value, global observable (pauli_mix), depth 3 | Qubits 12 | Qiskit 26.0 ms | Qulacs 3.17 ms | Faster Qulacs 8.2× |
Expectation value, global observable (pauli_mix), depth 3 | Qubits 16 | Qiskit 271 ms | Qulacs 61.5 ms | Faster Qulacs 4.4× |
Expectation value, global observable (pauli_mix), depth 3 | Qubits 20 | Qiskit 6.93 s | Qulacs 1.28 s | Faster Qulacs 5.4× |
Expectation value, global observable (zz_sum), depth 3 | Qubits 8 | Qiskit 10.4 ms | Qulacs 305 µs | Faster Qulacs 34.2× |
Expectation value, global observable (zz_sum), depth 3 | Qubits 12 | Qiskit 24.9 ms | Qulacs 2.50 ms | Faster Qulacs 9.9× |
Expectation value, global observable (zz_sum), depth 3 | Qubits 16 | Qiskit 261 ms | Qulacs 48.1 ms | Faster Qulacs 5.4× |
Expectation value, global observable (zz_sum), depth 3 | Qubits 20 | Qiskit 6.65 s | Qulacs 1.01 s | Faster Qulacs 6.6× |
Gradient, local observable, depth 3 | Qubits 4 | Qiskit unsupported | Qulacs 424 µs | Faster – |
Gradient, local observable, depth 3 | Qubits 8 | Qiskit unsupported | Qulacs 1.17 ms | Faster – |
Gradient, local observable, depth 3 | Qubits 12 | Qiskit unsupported | Qulacs 9.19 ms | Faster – |
Gradient, local observable, depth 3 | Qubits 16 | Qiskit unsupported | Qulacs 181 ms | Faster – |
MaxCut on a ring
QAOA over a cycle graph
| Workload | Qubits | Qiskit | Qulacs | Faster |
|---|---|---|---|---|
Expectation value, local observable, depth 3 | Qubits 4 | Qiskit 2.01 ms | Qulacs 63.1 µs | Faster Qulacs 31.9× |
Expectation value, local observable, depth 3 | Qubits 8 | Qiskit 4.71 ms | Qulacs 154 µs | Faster Qulacs 30.6× |
Expectation value, local observable, depth 3 | Qubits 12 | Qiskit 9.95 ms | Qulacs 1.14 ms | Faster Qulacs 8.7× |
Expectation value, local observable, depth 3 | Qubits 16 | Qiskit 105 ms | Qulacs 20.9 ms | Faster Qulacs 5.0× |
Expectation value, local observable, depth 3 | Qubits 20 | Qiskit 2.86 s | Qulacs 433 ms | Faster Qulacs 6.6× |
Expectation value, local observable, depth 3 | Qubits 24 | Qiskit skip | Qulacs 10.6 s | Faster – |
Expectation value, global observable (all_z), depth 3 | Qubits 4 | Qiskit 2.13 ms | Qulacs 72.0 µs | Faster Qulacs 29.6× |
Expectation value, global observable (all_z), depth 3 | Qubits 8 | Qiskit 4.72 ms | Qulacs 197 µs | Faster Qulacs 24.0× |
Expectation value, global observable (all_z), depth 3 | Qubits 12 | Qiskit 10.5 ms | Qulacs 1.91 ms | Faster Qulacs 5.5× |
Expectation value, global observable (all_z), depth 3 | Qubits 16 | Qiskit 107 ms | Qulacs 36.9 ms | Faster Qulacs 2.9× |
Expectation value, global observable (all_z), depth 3 | Qubits 20 | Qiskit 2.81 s | Qulacs 740 ms | Faster Qulacs 3.8× |
Expectation value, global observable (all_z), depth 3 | Qubits 24 | Qiskit skip | Qulacs 16.6 s | Faster – |
Expectation value, global observable (pauli_mix), depth 3 | Qubits 8 | Qiskit 5.70 ms | Qulacs 234 µs | Faster Qulacs 24.4× |
Expectation value, global observable (pauli_mix), depth 3 | Qubits 12 | Qiskit 12.4 ms | Qulacs 2.61 ms | Faster Qulacs 4.7× |
Expectation value, global observable (pauli_mix), depth 3 | Qubits 16 | Qiskit 121 ms | Qulacs 52.1 ms | Faster Qulacs 2.3× |
Expectation value, global observable (pauli_mix), depth 3 | Qubits 20 | Qiskit 2.83 s | Qulacs 1.03 s | Faster Qulacs 2.8× |
Expectation value, global observable (zz_sum), depth 3 | Qubits 8 | Qiskit 5.27 ms | Qulacs 191 µs | Faster Qulacs 27.6× |
Expectation value, global observable (zz_sum), depth 3 | Qubits 12 | Qiskit 11.8 ms | Qulacs 1.95 ms | Faster Qulacs 6.0× |
Expectation value, global observable (zz_sum), depth 3 | Qubits 16 | Qiskit 113 ms | Qulacs 38.6 ms | Faster Qulacs 2.9× |
Expectation value, global observable (zz_sum), depth 3 | Qubits 20 | Qiskit 2.84 s | Qulacs 762 ms | Faster Qulacs 3.7× |
Gradient, local observable, depth 3 | Qubits 4 | Qiskit unsupported | Qulacs 255 µs | Faster – |
Gradient, local observable, depth 3 | Qubits 8 | Qiskit unsupported | Qulacs 681 µs | Faster – |
Gradient, local observable, depth 3 | Qubits 12 | Qiskit unsupported | Qulacs 6.05 ms | Faster – |
Gradient, local observable, depth 3 | Qubits 16 | Qiskit unsupported | Qulacs 120 ms | Faster – |
Statevector, global observable (all_z), depth 3 | Qubits 8 | Qiskit 4.86 ms | Qulacs 155 µs | Faster Qulacs 31.3× |
Statevector, global observable (all_z), depth 3 | Qubits 12 | Qiskit 11.1 ms | Qulacs 1.13 ms | Faster Qulacs 9.8× |
Statevector, global observable (all_z), depth 3 | Qubits 16 | Qiskit 113 ms | Qulacs 21.0 ms | Faster Qulacs 5.4× |
MaxCut, arbitrary graph
QAOA over a random-edge graph
| Workload | Qubits | Qiskit | Qulacs | Faster |
|---|---|---|---|---|
Expectation value, local observable, depth 3 | Qubits 4 | Qiskit 2.23 ms | Qulacs 71.4 µs | Faster Qulacs 31.2× |
Expectation value, local observable, depth 3 | Qubits 8 | Qiskit 5.90 ms | Qulacs 188 µs | Faster Qulacs 31.5× |
Expectation value, local observable, depth 3 | Qubits 12 | Qiskit 15.3 ms | Qulacs 1.51 ms | Faster Qulacs 10.2× |
Expectation value, local observable, depth 3 | Qubits 16 | Qiskit 150 ms | Qulacs 25.8 ms | Faster Qulacs 5.8× |
Expectation value, local observable, depth 3 | Qubits 20 | Qiskit 4.43 s | Qulacs 588 ms | Faster Qulacs 7.5× |
Expectation value, local observable, depth 3 | Qubits 24 | Qiskit skip | Qulacs 16.1 s | Faster – |
Expectation value, global observable (all_z), depth 3 | Qubits 4 | Qiskit 2.40 ms | Qulacs 81.0 µs | Faster Qulacs 29.6× |
Expectation value, global observable (all_z), depth 3 | Qubits 8 | Qiskit 6.21 ms | Qulacs 239 µs | Faster Qulacs 25.9× |
Expectation value, global observable (all_z), depth 3 | Qubits 12 | Qiskit 15.9 ms | Qulacs 2.39 ms | Faster Qulacs 6.6× |
Expectation value, global observable (all_z), depth 3 | Qubits 16 | Qiskit 150 ms | Qulacs 46.3 ms | Faster Qulacs 3.2× |
Expectation value, global observable (all_z), depth 3 | Qubits 20 | Qiskit 4.30 s | Qulacs 1.00 s | Faster Qulacs 4.3× |
Expectation value, global observable (all_z), depth 3 | Qubits 24 | Qiskit skip | Qulacs 24.2 s | Faster – |
Gradient, local observable, depth 3 | Qubits 4 | Qiskit unsupported | Qulacs 287 µs | Faster – |
Gradient, local observable, depth 3 | Qubits 8 | Qiskit unsupported | Qulacs 873 µs | Faster – |
Gradient, local observable, depth 3 | Qubits 12 | Qiskit unsupported | Qulacs 8.07 ms | Faster – |
Gradient, local observable, depth 3 | Qubits 16 | Qiskit unsupported | Qulacs 148 ms | Faster – |
Sampling / IQP
Diagonal phase circuit between Hadamard layers
| Workload | Qubits | Qiskit | Qulacs | Faster |
|---|---|---|---|---|
Expectation value, local observable, depth 3 | Qubits 4 | Qiskit 2.20 ms | Qulacs 68.8 µs | Faster Qulacs 32.0× |
Expectation value, local observable, depth 3 | Qubits 8 | Qiskit 4.96 ms | Qulacs 156 µs | Faster Qulacs 31.9× |
Expectation value, local observable, depth 3 | Qubits 12 | Qiskit 11.3 ms | Qulacs 1.04 ms | Faster Qulacs 10.9× |
Expectation value, local observable, depth 3 | Qubits 16 | Qiskit 119 ms | Qulacs 18.5 ms | Faster Qulacs 6.4× |
Expectation value, local observable, depth 3 | Qubits 20 | Qiskit 3.26 s | Qulacs 392 ms | Faster Qulacs 8.3× |
Expectation value, local observable, depth 3 | Qubits 24 | Qiskit skip | Qulacs 10.7 s | Faster – |
Expectation value, global observable (all_z), depth 3 | Qubits 4 | Qiskit 2.33 ms | Qulacs 78.2 µs | Faster Qulacs 29.8× |
Expectation value, global observable (all_z), depth 3 | Qubits 8 | Qiskit 5.29 ms | Qulacs 203 µs | Faster Qulacs 26.1× |
Expectation value, global observable (all_z), depth 3 | Qubits 12 | Qiskit 11.7 ms | Qulacs 1.97 ms | Faster Qulacs 6.0× |
Expectation value, global observable (all_z), depth 3 | Qubits 16 | Qiskit 121 ms | Qulacs 38.7 ms | Faster Qulacs 3.1× |
Expectation value, global observable (all_z), depth 3 | Qubits 20 | Qiskit 3.09 s | Qulacs 797 ms | Faster Qulacs 3.9× |
Expectation value, global observable (all_z), depth 3 | Qubits 24 | Qiskit skip | Qulacs 19.1 s | Faster – |
Expectation value, global observable (pauli_mix), depth 3 | Qubits 8 | Qiskit 6.21 ms | Qulacs 239 µs | Faster Qulacs 26.0× |
Expectation value, global observable (pauli_mix), depth 3 | Qubits 12 | Qiskit 13.8 ms | Qulacs 2.67 ms | Faster Qulacs 5.2× |
Expectation value, global observable (pauli_mix), depth 3 | Qubits 16 | Qiskit 138 ms | Qulacs 53.4 ms | Faster Qulacs 2.6× |
Expectation value, global observable (pauli_mix), depth 3 | Qubits 20 | Qiskit 3.26 s | Qulacs 1.09 s | Faster Qulacs 3.0× |
Expectation value, global observable (zz_sum), depth 3 | Qubits 8 | Qiskit 6.01 ms | Qulacs 196 µs | Faster Qulacs 30.7× |
Expectation value, global observable (zz_sum), depth 3 | Qubits 12 | Qiskit 13.0 ms | Qulacs 2.04 ms | Faster Qulacs 6.4× |
Expectation value, global observable (zz_sum), depth 3 | Qubits 16 | Qiskit 134 ms | Qulacs 39.9 ms | Faster Qulacs 3.3× |
Expectation value, global observable (zz_sum), depth 3 | Qubits 20 | Qiskit 3.18 s | Qulacs 856 ms | Faster Qulacs 3.7× |
Gradient, local observable, depth 3 | Qubits 4 | Qiskit unsupported | Qulacs 272 µs | Faster – |
Gradient, local observable, depth 3 | Qubits 8 | Qiskit unsupported | Qulacs 701 µs | Faster – |
Gradient, local observable, depth 3 | Qubits 12 | Qiskit unsupported | Qulacs 5.38 ms | Faster – |
Gradient, local observable, depth 3 | Qubits 16 | Qiskit unsupported | Qulacs 105 ms | Faster – |
Quantum kernel feature map
Data-encoding circuit for kernel methods
| Workload | Qubits | Qiskit | Qulacs | Faster |
|---|---|---|---|---|
Expectation value, global observable (all_z), depth 2 | Qubits 8 | Qiskit 2.04 ms | Qulacs 107 µs | Faster Qulacs 19.2× |
Expectation value, global observable (all_z), depth 2 | Qubits 12 | Qiskit 4.12 ms | Qulacs 895 µs | Faster Qulacs 4.6× |
Expectation value, global observable (all_z), depth 2 | Qubits 16 | Qiskit skip | Qulacs 16.9 ms | Faster – |
Expectation value, global observable (all_z), depth 2, batch of 32 | Qubits 8 | Qiskit 64.7 ms | Qulacs 3.36 ms | Faster Qulacs 19.2× |
Expectation value, global observable (all_z), depth 2, batch of 32 | Qubits 12 | Qiskit 132 ms | Qulacs 28.3 ms | Faster Qulacs 4.7× |
Expectation value, global observable (all_z), depth 2, batch of 32 | Qubits 16 | Qiskit skip | Qulacs 539 ms | Faster – |
Expectation value, global observable (all_z), depth 2, batch of 256 | Qubits 8 | Qiskit 521 ms | Qulacs 26.9 ms | Faster Qulacs 19.3× |
Expectation value, global observable (all_z), depth 2, batch of 256 | Qubits 12 | Qiskit 1.04 s | Qulacs 226 ms | Faster Qulacs 4.6× |
Expectation value, global observable (all_z), depth 2, batch of 256 | Qubits 16 | Qiskit skip | Qulacs 4.31 s | Faster – |
Expectation value, global observable (all_z), depth 2, batch of 1024 | Qubits 8 | Qiskit 2.12 s | Qulacs 109 ms | Faster Qulacs 19.4× |
Expectation value, global observable (all_z), depth 2, batch of 1024 | Qubits 12 | Qiskit 4.16 s | Qulacs 915 ms | Faster Qulacs 4.5× |
Expectation value, global observable (all_z), depth 2, batch of 1024 | Qubits 16 | Qiskit skip | Qulacs 17.3 s | Faster – |
Hardware-efficient VQE
RY/RZ layers with a CNOT ring
| Workload | Qubits | Qiskit | Qulacs | Faster |
|---|---|---|---|---|
Expectation value, local observable, depth 3 | Qubits 4 | Qiskit 1.61 ms | Qulacs 52.5 µs | Faster Qulacs 30.7× |
Expectation value, local observable, depth 3 | Qubits 8 | Qiskit 3.51 ms | Qulacs 126 µs | Faster Qulacs 27.8× |
Expectation value, local observable, depth 3 | Qubits 12 | Qiskit 7.81 ms | Qulacs 1.03 ms | Faster Qulacs 7.6× |
Expectation value, local observable, depth 3 | Qubits 16 | Qiskit 74.9 ms | Qulacs 19.2 ms | Faster Qulacs 3.9× |
Expectation value, local observable, depth 3 | Qubits 20 | Qiskit 1.90 s | Qulacs 386 ms | Faster Qulacs 4.9× |
Expectation value, local observable, depth 3 | Qubits 24 | Qiskit skip | Qulacs 8.90 s | Faster – |
Expectation value, global observable (all_z), depth 3 | Qubits 4 | Qiskit 1.80 ms | Qulacs 63.7 µs | Faster Qulacs 28.3× |
Expectation value, global observable (all_z), depth 3 | Qubits 8 | Qiskit 3.83 ms | Qulacs 172 µs | Faster Qulacs 22.3× |
Expectation value, global observable (all_z), depth 3 | Qubits 12 | Qiskit 8.17 ms | Qulacs 1.97 ms | Faster Qulacs 4.1× |
Expectation value, global observable (all_z), depth 3 | Qubits 16 | Qiskit 79.1 ms | Qulacs 40.2 ms | Faster Qulacs 2.0× |
Expectation value, global observable (all_z), depth 3 | Qubits 20 | Qiskit 2.04 s | Qulacs 808 ms | Faster Qulacs 2.5× |
Expectation value, global observable (all_z), depth 3 | Qubits 24 | Qiskit 134 s | Qulacs 17.1 s | Faster Qulacs 7.8× |
Gradient, local observable, depth 3 | Qubits 4 | Qiskit unsupported | Qulacs 206 µs | Faster – |
Gradient, local observable, depth 3 | Qubits 8 | Qiskit unsupported | Qulacs 577 µs | Faster – |
Gradient, local observable, depth 3 | Qubits 12 | Qiskit unsupported | Qulacs 5.51 ms | Faster – |
Gradient, local observable, depth 3 | Qubits 16 | Qiskit unsupported | Qulacs 110 ms | Faster – |
Statevector, global observable (all_z), depth 3 | Qubits 8 | Qiskit 3.68 ms | Qulacs 129 µs | Faster Qulacs 28.5× |
Statevector, global observable (all_z), depth 3 | Qubits 12 | Qiskit 7.98 ms | Qulacs 1.00 ms | Faster Qulacs 8.0× |
Statevector, global observable (all_z), depth 3 | Qubits 16 | Qiskit 77.5 ms | Qulacs 18.9 ms | Faster Qulacs 4.1× |
Strongly-entangling ansatz
RX/RY/RZ layers with a CNOT ring
| Workload | Qubits | Qiskit | Qulacs | Faster |
|---|---|---|---|---|
Expectation value, local observable, depth 3 | Qubits 4 | Qiskit 1.96 ms | Qulacs 68.2 µs | Faster Qulacs 28.8× |
Expectation value, local observable, depth 3 | Qubits 8 | Qiskit 4.37 ms | Qulacs 177 µs | Faster Qulacs 24.7× |
Expectation value, local observable, depth 3 | Qubits 12 | Qiskit 9.48 ms | Qulacs 1.56 ms | Faster Qulacs 6.1× |
Expectation value, local observable, depth 3 | Qubits 16 | Qiskit 101 ms | Qulacs 30.0 ms | Faster Qulacs 3.4× |
Expectation value, local observable, depth 3 | Qubits 20 | Qiskit 2.61 s | Qulacs 595 ms | Faster Qulacs 4.4× |
Expectation value, local observable, depth 3 | Qubits 24 | Qiskit skip | Qulacs 13.2 s | Faster – |
Expectation value, global observable (all_z), depth 3 | Qubits 4 | Qiskit 2.09 ms | Qulacs 77.7 µs | Faster Qulacs 26.9× |
Expectation value, global observable (all_z), depth 3 | Qubits 8 | Qiskit 4.65 ms | Qulacs 222 µs | Faster Qulacs 20.9× |
Expectation value, global observable (all_z), depth 3 | Qubits 12 | Qiskit 9.75 ms | Qulacs 2.50 ms | Faster Qulacs 3.9× |
Expectation value, global observable (all_z), depth 3 | Qubits 16 | Qiskit 96.2 ms | Qulacs 50.6 ms | Faster Qulacs 1.9× |
Expectation value, global observable (all_z), depth 3 | Qubits 20 | Qiskit 2.62 s | Qulacs 1.02 s | Faster Qulacs 2.6× |
Expectation value, global observable (all_z), depth 3 | Qubits 24 | Qiskit skip | Qulacs 21.4 s | Faster – |
Gradient, local observable, depth 3 | Qubits 4 | Qiskit unsupported | Qulacs 278 µs | Faster – |
Gradient, local observable, depth 3 | Qubits 8 | Qiskit unsupported | Qulacs 822 µs | Faster – |
Gradient, local observable, depth 3 | Qubits 12 | Qiskit unsupported | Qulacs 8.63 ms | Faster – |
Gradient, local observable, depth 3 | Qubits 16 | Qiskit unsupported | Qulacs 180 ms | Faster – |
Data re-uploading QML
Repeated encode-and-rotate classifier
| Workload | Qubits | Qiskit | Qulacs | Faster |
|---|---|---|---|---|
Expectation value, global observable (all_z), depth 2 | Qubits 8 | Qiskit 3.65 ms | Qulacs 185 µs | Faster Qulacs 19.7× |
Expectation value, global observable (all_z), depth 2 | Qubits 12 | Qiskit 7.61 ms | Qulacs 2.06 ms | Faster Qulacs 3.7× |
Expectation value, global observable (all_z), depth 2 | Qubits 16 | Qiskit 72.6 ms | Qulacs 41.8 ms | Faster Qulacs 1.7× |
Expectation value, global observable (all_z), depth 2, batch of 32 | Qubits 8 | Qiskit 116 ms | Qulacs 6.97 ms | Faster Qulacs 16.6× |
Expectation value, global observable (all_z), depth 2, batch of 32 | Qubits 12 | Qiskit 246 ms | Qulacs 67.1 ms | Faster Qulacs 3.7× |
Expectation value, global observable (all_z), depth 2, batch of 32 | Qubits 16 | Qiskit 2.26 s | Qulacs 1.33 s | Faster Qulacs 1.7× |
Expectation value, global observable (all_z), depth 2, batch of 256 | Qubits 8 | Qiskit 941 ms | Qulacs 48.4 ms | Faster Qulacs 19.4× |
Expectation value, global observable (all_z), depth 2, batch of 256 | Qubits 12 | Qiskit 1.94 s | Qulacs 529 ms | Faster Qulacs 3.7× |
Expectation value, global observable (all_z), depth 2, batch of 256 | Qubits 16 | Qiskit 17.9 s | Qulacs 10.6 s | Faster Qulacs 1.7× |
Expectation value, global observable (all_z), depth 2, batch of 1024 | Qubits 8 | Qiskit 3.74 s | Qulacs 193 ms | Faster Qulacs 19.4× |
Expectation value, global observable (all_z), depth 2, batch of 1024 | Qubits 12 | Qiskit 7.93 s | Qulacs 2.13 s | Faster Qulacs 3.7× |
Expectation value, global observable (all_z), depth 2, batch of 1024 | Qubits 16 | Qiskit 72.3 s | Qulacs 42.7 s | Faster Qulacs 1.7× |
Gradient, local observable, depth 2 | Qubits 8 | Qiskit unsupported | Qulacs 606 µs | Faster – |
Gradient, local observable, depth 2 | Qubits 12 | Qiskit unsupported | Qulacs 5.81 ms | Faster – |
Gradient, local observable, depth 2 | Qubits 16 | Qiskit unsupported | Qulacs 117 ms | Faster – |
Gradient, local observable, depth 2, batch of 32 | Qubits 8 | Qiskit unsupported | Qulacs 19.3 ms | Faster – |
Gradient, local observable, depth 2, batch of 32 | Qubits 12 | Qiskit unsupported | Qulacs 189 ms | Faster – |
Gradient, local observable, depth 2, batch of 32 | Qubits 16 | Qiskit unsupported | Qulacs 3.76 s | Faster – |
Gradient, local observable, depth 2, batch of 256 | Qubits 8 | Qiskit unsupported | Qulacs 153 ms | Faster – |
Gradient, local observable, depth 2, batch of 256 | Qubits 12 | Qiskit unsupported | Qulacs 1.47 s | Faster – |
Gradient, local observable, depth 2, batch of 256 | Qubits 16 | Qiskit unsupported | Qulacs 30.0 s | Faster – |
How this was measured
One host, one harness
Every simulator ran the same circuits on one host, single-threaded and core-pinned, driven through one common harness, so any overhead the harness adds is charged to the simulator. Each result was checked against a reference implementation: a simulator that ran fast but returned the wrong answer is left out of the ranking. Nothing is interpolated or rescaled. A time was measured, or it is not shown.
Run 2026-08-10, 8 cores, threads=1. numpy 2.4.6, scipy 1.18.0, torch 2.13.0, numba 0.66.0, pennylane 0.45.1, pennylane_lightning 0.45.0, qiskit 2.5.1, qulacs 0.6.14, tensorcircuit 1.8.0, cudaq 0.15.1. One host and one thread make the comparison reproducible. It is not a prediction for your laptop.
The interactive tool shows the same measurements for any workload you pick, with the quantum hardware that could run it and what a run would cost.