Qulacs vs TensorCircuit
The same circuits, one host
Qulacs and TensorCircuit both finished 151 of the 175 workloads measured here. Qulacs was faster in 100 of them and TensorCircuit in 51. At the median Qulacs was 1.3× faster.
By circuit family
Who was faster where
| Circuit family | Workloads both finished | Qulacs faster in | TensorCircuit faster in | Faster at the median |
|---|---|---|---|---|
| Workloads both finished 16 | Qulacs faster in 9 | TensorCircuit faster in 7 | Faster at the median Qulacs 1.3× | |
| Workloads both finished 24 | Qulacs faster in 13 | TensorCircuit faster in 11 | Faster at the median Qulacs 1.1× | |
| Workloads both finished 24 | Qulacs faster in 22 | TensorCircuit faster in 2 | Faster at the median Qulacs 1.7× | |
| Workloads both finished 0 | Qulacs faster in 0 | TensorCircuit faster in 0 | Faster at the median – | |
| Workloads both finished 24 | Qulacs faster in 22 | TensorCircuit faster in 2 | Faster at the median Qulacs 1.7× | |
| Workloads both finished 11 | Qulacs faster in 7 | TensorCircuit faster in 4 | Faster at the median Qulacs 1.9× | |
| Workloads both finished 16 | Qulacs faster in 14 | TensorCircuit faster in 2 | Faster at the median Qulacs 1.4× | |
| Workloads both finished 16 | Qulacs faster in 8 | TensorCircuit faster in 8 | Faster at the median Qulacs 1.0× | |
| Workloads both finished 20 | Qulacs faster in 5 | TensorCircuit faster in 15 | Faster at the median TensorCircuit 2.5× |
Local circuit dynamics
Brickwork nearest-neighbour layers: lattice models, quench dynamics
| Workload | Qubits | Qulacs | TensorCircuit | Faster |
|---|---|---|---|---|
Expectation value, local observable, depth 3 | Qubits 4 | Qulacs 32.1 µs | TensorCircuit 159 µs | Faster Qulacs 5.0× |
Expectation value, local observable, depth 3 | Qubits 8 | Qulacs 73.1 µs | TensorCircuit 249 µs | Faster Qulacs 3.4× |
Expectation value, local observable, depth 3 | Qubits 12 | Qulacs 585 µs | TensorCircuit 378 µs | Faster TensorCircuit 1.6× |
Expectation value, local observable, depth 3 | Qubits 16 | Qulacs 10.7 ms | TensorCircuit 2.17 ms | Faster TensorCircuit 4.9× |
Expectation value, local observable, depth 3 | Qubits 20 | Qulacs 206 ms | TensorCircuit 63.5 ms | Faster TensorCircuit 3.2× |
Expectation value, local observable, depth 3 | Qubits 24 | Qulacs 4.76 s | TensorCircuit 1.16 s | Faster TensorCircuit 4.1× |
Expectation value, global observable (all_z), depth 3 | Qubits 4 | Qulacs 42.5 µs | TensorCircuit 181 µs | Faster Qulacs 4.3× |
Expectation value, global observable (all_z), depth 3 | Qubits 8 | Qulacs 119 µs | TensorCircuit 285 µs | Faster Qulacs 2.4× |
Expectation value, global observable (all_z), depth 3 | Qubits 12 | Qulacs 1.38 ms | TensorCircuit 848 µs | Faster TensorCircuit 1.6× |
Expectation value, global observable (all_z), depth 3 | Qubits 16 | Qulacs 26.2 ms | TensorCircuit 33.5 ms | Faster Qulacs 1.3× |
Expectation value, global observable (all_z), depth 3 | Qubits 20 | Qulacs 535 ms | TensorCircuit 735 ms | Faster Qulacs 1.4× |
Expectation value, global observable (all_z), depth 3 | Qubits 24 | Qulacs 10.7 s | TensorCircuit 16.2 s | Faster Qulacs 1.5× |
Gradient, local observable, depth 3 | Qubits 4 | Qulacs 117 µs | TensorCircuit 253 µs | Faster Qulacs 2.2× |
Gradient, local observable, depth 3 | Qubits 8 | Qulacs 307 µs | TensorCircuit 426 µs | Faster Qulacs 1.4× |
Gradient, local observable, depth 3 | Qubits 12 | Qulacs 3.00 ms | TensorCircuit 758 µs | Faster TensorCircuit 4.0× |
Gradient, local observable, depth 3 | Qubits 16 | Qulacs 59.6 ms | TensorCircuit 5.49 ms | Faster TensorCircuit 10.9× |
Hamiltonian time evolution
Trotterised spin chain: ZZ + XX couplings with local fields
| Workload | Qubits | Qulacs | TensorCircuit | Faster |
|---|---|---|---|---|
Expectation value, local observable, depth 3 | Qubits 4 | Qulacs 104 µs | TensorCircuit 267 µs | Faster Qulacs 2.6× |
Expectation value, local observable, depth 3 | Qubits 8 | Qulacs 276 µs | TensorCircuit 468 µs | Faster Qulacs 1.7× |
Expectation value, local observable, depth 3 | Qubits 12 | Qulacs 1.85 ms | TensorCircuit 749 µs | Faster TensorCircuit 2.5× |
Expectation value, local observable, depth 3 | Qubits 16 | Qulacs 34.0 ms | TensorCircuit 4.46 ms | Faster TensorCircuit 7.6× |
Expectation value, local observable, depth 3 | Qubits 20 | Qulacs 711 ms | TensorCircuit 164 ms | Faster TensorCircuit 4.3× |
Expectation value, local observable, depth 3 | Qubits 24 | Qulacs 21.1 s | TensorCircuit 6.20 s | Faster TensorCircuit 3.4× |
Expectation value, global observable (all_z), depth 3 | Qubits 4 | Qulacs 113 µs | TensorCircuit 275 µs | Faster Qulacs 2.4× |
Expectation value, global observable (all_z), depth 3 | Qubits 8 | Qulacs 321 µs | TensorCircuit 522 µs | Faster Qulacs 1.6× |
Expectation value, global observable (all_z), depth 3 | Qubits 12 | Qulacs 2.51 ms | TensorCircuit 1.42 ms | Faster TensorCircuit 1.8× |
Expectation value, global observable (all_z), depth 3 | Qubits 16 | Qulacs 48.2 ms | TensorCircuit 45.6 ms | Faster TensorCircuit 1.1× |
Expectation value, global observable (all_z), depth 3 | Qubits 20 | Qulacs 1.02 s | TensorCircuit 1.43 s | Faster Qulacs 1.4× |
Expectation value, global observable (all_z), depth 3 | Qubits 24 | Qulacs 26.9 s | TensorCircuit 45.3 s | Faster Qulacs 1.7× |
Expectation value, global observable (pauli_mix), depth 3 | Qubits 8 | Qulacs 349 µs | TensorCircuit 593 µs | Faster Qulacs 1.7× |
Expectation value, global observable (pauli_mix), depth 3 | Qubits 12 | Qulacs 3.17 ms | TensorCircuit 2.12 ms | Faster TensorCircuit 1.5× |
Expectation value, global observable (pauli_mix), depth 3 | Qubits 16 | Qulacs 61.5 ms | TensorCircuit 78.8 ms | Faster Qulacs 1.3× |
Expectation value, global observable (pauli_mix), depth 3 | Qubits 20 | Qulacs 1.28 s | TensorCircuit 2.65 s | Faster Qulacs 2.1× |
Expectation value, global observable (zz_sum), depth 3 | Qubits 8 | Qulacs 305 µs | TensorCircuit 537 µs | Faster Qulacs 1.8× |
Expectation value, global observable (zz_sum), depth 3 | Qubits 12 | Qulacs 2.50 ms | TensorCircuit 1.74 ms | Faster TensorCircuit 1.4× |
Expectation value, global observable (zz_sum), depth 3 | Qubits 16 | Qulacs 48.1 ms | TensorCircuit 60.3 ms | Faster Qulacs 1.3× |
Expectation value, global observable (zz_sum), depth 3 | Qubits 20 | Qulacs 1.01 s | TensorCircuit 1.90 s | Faster Qulacs 1.9× |
Gradient, local observable, depth 3 | Qubits 4 | Qulacs 424 µs | TensorCircuit 442 µs | Faster Qulacs 1.0× |
Gradient, local observable, depth 3 | Qubits 8 | Qulacs 1.17 ms | TensorCircuit 962 µs | Faster TensorCircuit 1.2× |
Gradient, local observable, depth 3 | Qubits 12 | Qulacs 9.19 ms | TensorCircuit 1.61 ms | Faster TensorCircuit 5.7× |
Gradient, local observable, depth 3 | Qubits 16 | Qulacs 181 ms | TensorCircuit 10.7 ms | Faster TensorCircuit 16.9× |
MaxCut on a ring
QAOA over a cycle graph
| Workload | Qubits | Qulacs | TensorCircuit | Faster |
|---|---|---|---|---|
Expectation value, local observable, depth 3 | Qubits 4 | Qulacs 63.1 µs | TensorCircuit 199 µs | Faster Qulacs 3.1× |
Expectation value, local observable, depth 3 | Qubits 8 | Qulacs 154 µs | TensorCircuit 387 µs | Faster Qulacs 2.5× |
Expectation value, local observable, depth 3 | Qubits 12 | Qulacs 1.14 ms | TensorCircuit 1.43 ms | Faster Qulacs 1.3× |
Expectation value, local observable, depth 3 | Qubits 16 | Qulacs 20.9 ms | TensorCircuit 21.2 ms | Faster Qulacs 1.0× |
Expectation value, local observable, depth 3 | Qubits 20 | Qulacs 433 ms | TensorCircuit 596 ms | Faster Qulacs 1.4× |
Expectation value, local observable, depth 3 | Qubits 24 | Qulacs 10.6 s | TensorCircuit 12.8 s | Faster Qulacs 1.2× |
Expectation value, global observable (all_z), depth 3 | Qubits 4 | Qulacs 72.0 µs | TensorCircuit 207 µs | Faster Qulacs 2.9× |
Expectation value, global observable (all_z), depth 3 | Qubits 8 | Qulacs 197 µs | TensorCircuit 428 µs | Faster Qulacs 2.2× |
Expectation value, global observable (all_z), depth 3 | Qubits 12 | Qulacs 1.91 ms | TensorCircuit 2.01 ms | Faster Qulacs 1.1× |
Expectation value, global observable (all_z), depth 3 | Qubits 16 | Qulacs 36.9 ms | TensorCircuit 58.3 ms | Faster Qulacs 1.6× |
Expectation value, global observable (all_z), depth 3 | Qubits 20 | Qulacs 740 ms | TensorCircuit 1.66 s | Faster Qulacs 2.2× |
Expectation value, global observable (all_z), depth 3 | Qubits 24 | Qulacs 16.6 s | TensorCircuit 42.9 s | Faster Qulacs 2.6× |
Expectation value, global observable (pauli_mix), depth 3 | Qubits 8 | Qulacs 234 µs | TensorCircuit 541 µs | Faster Qulacs 2.3× |
Expectation value, global observable (pauli_mix), depth 3 | Qubits 12 | Qulacs 2.61 ms | TensorCircuit 2.90 ms | Faster Qulacs 1.1× |
Expectation value, global observable (pauli_mix), depth 3 | Qubits 16 | Qulacs 52.1 ms | TensorCircuit 96.2 ms | Faster Qulacs 1.8× |
Expectation value, global observable (pauli_mix), depth 3 | Qubits 20 | Qulacs 1.03 s | TensorCircuit 2.50 s | Faster Qulacs 2.4× |
Expectation value, global observable (zz_sum), depth 3 | Qubits 8 | Qulacs 191 µs | TensorCircuit 502 µs | Faster Qulacs 2.6× |
Expectation value, global observable (zz_sum), depth 3 | Qubits 12 | Qulacs 1.95 ms | TensorCircuit 2.53 ms | Faster Qulacs 1.3× |
Expectation value, global observable (zz_sum), depth 3 | Qubits 16 | Qulacs 38.6 ms | TensorCircuit 77.6 ms | Faster Qulacs 2.0× |
Expectation value, global observable (zz_sum), depth 3 | Qubits 20 | Qulacs 762 ms | TensorCircuit 2.25 s | Faster Qulacs 3.0× |
Gradient, local observable, depth 3 | Qubits 4 | Qulacs 255 µs | TensorCircuit 372 µs | Faster Qulacs 1.5× |
Gradient, local observable, depth 3 | Qubits 8 | Qulacs 681 µs | TensorCircuit 858 µs | Faster Qulacs 1.3× |
Gradient, local observable, depth 3 | Qubits 12 | Qulacs 6.05 ms | TensorCircuit 4.31 ms | Faster TensorCircuit 1.4× |
Gradient, local observable, depth 3 | Qubits 16 | Qulacs 120 ms | TensorCircuit 104 ms | Faster TensorCircuit 1.2× |
Statevector, global observable (all_z), depth 3 | Qubits 8 | Qulacs 155 µs | TensorCircuit unsupported | Faster – |
Statevector, global observable (all_z), depth 3 | Qubits 12 | Qulacs 1.13 ms | TensorCircuit unsupported | Faster – |
Statevector, global observable (all_z), depth 3 | Qubits 16 | Qulacs 21.0 ms | TensorCircuit unsupported | Faster – |
MaxCut, arbitrary graph
QAOA over a random-edge graph
| Workload | Qubits | Qulacs | TensorCircuit | Faster |
|---|---|---|---|---|
Expectation value, local observable, depth 3 | Qubits 4 | Qulacs 71.4 µs | TensorCircuit error | Faster – |
Expectation value, local observable, depth 3 | Qubits 8 | Qulacs 188 µs | TensorCircuit error | Faster – |
Expectation value, local observable, depth 3 | Qubits 12 | Qulacs 1.51 ms | TensorCircuit error | Faster – |
Expectation value, local observable, depth 3 | Qubits 16 | Qulacs 25.8 ms | TensorCircuit error | Faster – |
Expectation value, local observable, depth 3 | Qubits 20 | Qulacs 588 ms | TensorCircuit error | Faster – |
Expectation value, local observable, depth 3 | Qubits 24 | Qulacs 16.1 s | TensorCircuit error | Faster – |
Expectation value, global observable (all_z), depth 3 | Qubits 4 | Qulacs 81.0 µs | TensorCircuit error | Faster – |
Expectation value, global observable (all_z), depth 3 | Qubits 8 | Qulacs 239 µs | TensorCircuit error | Faster – |
Expectation value, global observable (all_z), depth 3 | Qubits 12 | Qulacs 2.39 ms | TensorCircuit error | Faster – |
Expectation value, global observable (all_z), depth 3 | Qubits 16 | Qulacs 46.3 ms | TensorCircuit error | Faster – |
Expectation value, global observable (all_z), depth 3 | Qubits 20 | Qulacs 1.00 s | TensorCircuit error | Faster – |
Expectation value, global observable (all_z), depth 3 | Qubits 24 | Qulacs 24.2 s | TensorCircuit error | Faster – |
Gradient, local observable, depth 3 | Qubits 4 | Qulacs 287 µs | TensorCircuit error | Faster – |
Gradient, local observable, depth 3 | Qubits 8 | Qulacs 873 µs | TensorCircuit error | Faster – |
Gradient, local observable, depth 3 | Qubits 12 | Qulacs 8.07 ms | TensorCircuit error | Faster – |
Gradient, local observable, depth 3 | Qubits 16 | Qulacs 148 ms | TensorCircuit error | Faster – |
Sampling / IQP
Diagonal phase circuit between Hadamard layers
| Workload | Qubits | Qulacs | TensorCircuit | Faster |
|---|---|---|---|---|
Expectation value, local observable, depth 3 | Qubits 4 | Qulacs 68.8 µs | TensorCircuit 209 µs | Faster Qulacs 3.0× |
Expectation value, local observable, depth 3 | Qubits 8 | Qulacs 156 µs | TensorCircuit 425 µs | Faster Qulacs 2.7× |
Expectation value, local observable, depth 3 | Qubits 12 | Qulacs 1.04 ms | TensorCircuit 1.49 ms | Faster Qulacs 1.4× |
Expectation value, local observable, depth 3 | Qubits 16 | Qulacs 18.5 ms | TensorCircuit 21.1 ms | Faster Qulacs 1.1× |
Expectation value, local observable, depth 3 | Qubits 20 | Qulacs 392 ms | TensorCircuit 633 ms | Faster Qulacs 1.6× |
Expectation value, local observable, depth 3 | Qubits 24 | Qulacs 10.7 s | TensorCircuit 13.7 s | Faster Qulacs 1.3× |
Expectation value, global observable (all_z), depth 3 | Qubits 4 | Qulacs 78.2 µs | TensorCircuit 229 µs | Faster Qulacs 2.9× |
Expectation value, global observable (all_z), depth 3 | Qubits 8 | Qulacs 203 µs | TensorCircuit 464 µs | Faster Qulacs 2.3× |
Expectation value, global observable (all_z), depth 3 | Qubits 12 | Qulacs 1.97 ms | TensorCircuit 2.04 ms | Faster Qulacs 1.0× |
Expectation value, global observable (all_z), depth 3 | Qubits 16 | Qulacs 38.7 ms | TensorCircuit 59.5 ms | Faster Qulacs 1.5× |
Expectation value, global observable (all_z), depth 3 | Qubits 20 | Qulacs 797 ms | TensorCircuit 1.53 s | Faster Qulacs 1.9× |
Expectation value, global observable (all_z), depth 3 | Qubits 24 | Qulacs 19.1 s | TensorCircuit 45.0 s | Faster Qulacs 2.4× |
Expectation value, global observable (pauli_mix), depth 3 | Qubits 8 | Qulacs 239 µs | TensorCircuit 568 µs | Faster Qulacs 2.4× |
Expectation value, global observable (pauli_mix), depth 3 | Qubits 12 | Qulacs 2.67 ms | TensorCircuit 2.94 ms | Faster Qulacs 1.1× |
Expectation value, global observable (pauli_mix), depth 3 | Qubits 16 | Qulacs 53.4 ms | TensorCircuit 96.1 ms | Faster Qulacs 1.8× |
Expectation value, global observable (pauli_mix), depth 3 | Qubits 20 | Qulacs 1.09 s | TensorCircuit 2.76 s | Faster Qulacs 2.5× |
Expectation value, global observable (zz_sum), depth 3 | Qubits 8 | Qulacs 196 µs | TensorCircuit 525 µs | Faster Qulacs 2.7× |
Expectation value, global observable (zz_sum), depth 3 | Qubits 12 | Qulacs 2.04 ms | TensorCircuit 2.56 ms | Faster Qulacs 1.3× |
Expectation value, global observable (zz_sum), depth 3 | Qubits 16 | Qulacs 39.9 ms | TensorCircuit 79.0 ms | Faster Qulacs 2.0× |
Expectation value, global observable (zz_sum), depth 3 | Qubits 20 | Qulacs 856 ms | TensorCircuit 2.22 s | Faster Qulacs 2.6× |
Gradient, local observable, depth 3 | Qubits 4 | Qulacs 272 µs | TensorCircuit 425 µs | Faster Qulacs 1.6× |
Gradient, local observable, depth 3 | Qubits 8 | Qulacs 701 µs | TensorCircuit 898 µs | Faster Qulacs 1.3× |
Gradient, local observable, depth 3 | Qubits 12 | Qulacs 5.38 ms | TensorCircuit 4.39 ms | Faster TensorCircuit 1.2× |
Gradient, local observable, depth 3 | Qubits 16 | Qulacs 105 ms | TensorCircuit 104 ms | Faster TensorCircuit 1.0× |
Quantum kernel feature map
Data-encoding circuit for kernel methods
| Workload | Qubits | Qulacs | TensorCircuit | Faster |
|---|---|---|---|---|
Expectation value, global observable (all_z), depth 2 | Qubits 8 | Qulacs 107 µs | TensorCircuit 288 µs | Faster Qulacs 2.7× |
Expectation value, global observable (all_z), depth 2 | Qubits 12 | Qulacs 895 µs | TensorCircuit 871 µs | Faster TensorCircuit 1.0× |
Expectation value, global observable (all_z), depth 2 | Qubits 16 | Qulacs 16.9 ms | TensorCircuit 30.7 ms | Faster Qulacs 1.8× |
Expectation value, global observable (all_z), depth 2, batch of 32 | Qubits 8 | Qulacs 3.36 ms | TensorCircuit 1.28 ms | Faster TensorCircuit 2.6× |
Expectation value, global observable (all_z), depth 2, batch of 32 | Qubits 12 | Qulacs 28.3 ms | TensorCircuit 52.5 ms | Faster Qulacs 1.9× |
Expectation value, global observable (all_z), depth 2, batch of 32 | Qubits 16 | Qulacs 539 ms | TensorCircuit 1.09 s | Faster Qulacs 2.0× |
Expectation value, global observable (all_z), depth 2, batch of 256 | Qubits 8 | Qulacs 26.9 ms | TensorCircuit 8.45 ms | Faster TensorCircuit 3.2× |
Expectation value, global observable (all_z), depth 2, batch of 256 | Qubits 12 | Qulacs 226 ms | TensorCircuit 428 ms | Faster Qulacs 1.9× |
Expectation value, global observable (all_z), depth 2, batch of 256 | Qubits 16 | Qulacs 4.31 s | TensorCircuit 8.81 s | Faster Qulacs 2.0× |
Expectation value, global observable (all_z), depth 2, batch of 1024 | Qubits 8 | Qulacs 109 ms | TensorCircuit 79.4 ms | Faster TensorCircuit 1.4× |
Expectation value, global observable (all_z), depth 2, batch of 1024 | Qubits 12 | Qulacs 915 ms | TensorCircuit 1.79 s | Faster Qulacs 2.0× |
Expectation value, global observable (all_z), depth 2, batch of 1024 | Qubits 16 | Qulacs 17.3 s | TensorCircuit error | Faster – |
Hardware-efficient VQE
RY/RZ layers with a CNOT ring
| Workload | Qubits | Qulacs | TensorCircuit | Faster |
|---|---|---|---|---|
Expectation value, local observable, depth 3 | Qubits 4 | Qulacs 52.5 µs | TensorCircuit 211 µs | Faster Qulacs 4.0× |
Expectation value, local observable, depth 3 | Qubits 8 | Qulacs 126 µs | TensorCircuit 386 µs | Faster Qulacs 3.1× |
Expectation value, local observable, depth 3 | Qubits 12 | Qulacs 1.03 ms | TensorCircuit 1.50 ms | Faster Qulacs 1.5× |
Expectation value, local observable, depth 3 | Qubits 16 | Qulacs 19.2 ms | TensorCircuit 20.7 ms | Faster Qulacs 1.1× |
Expectation value, local observable, depth 3 | Qubits 20 | Qulacs 386 ms | TensorCircuit 488 ms | Faster Qulacs 1.3× |
Expectation value, local observable, depth 3 | Qubits 24 | Qulacs 8.90 s | TensorCircuit 13.7 s | Faster Qulacs 1.5× |
Expectation value, global observable (all_z), depth 3 | Qubits 4 | Qulacs 63.7 µs | TensorCircuit 212 µs | Faster Qulacs 3.3× |
Expectation value, global observable (all_z), depth 3 | Qubits 8 | Qulacs 172 µs | TensorCircuit 454 µs | Faster Qulacs 2.6× |
Expectation value, global observable (all_z), depth 3 | Qubits 12 | Qulacs 1.97 ms | TensorCircuit 2.02 ms | Faster Qulacs 1.0× |
Expectation value, global observable (all_z), depth 3 | Qubits 16 | Qulacs 40.2 ms | TensorCircuit 53.9 ms | Faster Qulacs 1.3× |
Expectation value, global observable (all_z), depth 3 | Qubits 20 | Qulacs 808 ms | TensorCircuit 1.09 s | Faster Qulacs 1.4× |
Expectation value, global observable (all_z), depth 3 | Qubits 24 | Qulacs 17.1 s | TensorCircuit 27.5 s | Faster Qulacs 1.6× |
Gradient, local observable, depth 3 | Qubits 4 | Qulacs 206 µs | TensorCircuit 340 µs | Faster Qulacs 1.6× |
Gradient, local observable, depth 3 | Qubits 8 | Qulacs 577 µs | TensorCircuit 830 µs | Faster Qulacs 1.4× |
Gradient, local observable, depth 3 | Qubits 12 | Qulacs 5.51 ms | TensorCircuit 4.05 ms | Faster TensorCircuit 1.4× |
Gradient, local observable, depth 3 | Qubits 16 | Qulacs 110 ms | TensorCircuit 66.6 ms | Faster TensorCircuit 1.7× |
Statevector, global observable (all_z), depth 3 | Qubits 8 | Qulacs 129 µs | TensorCircuit unsupported | Faster – |
Statevector, global observable (all_z), depth 3 | Qubits 12 | Qulacs 1.00 ms | TensorCircuit unsupported | Faster – |
Statevector, global observable (all_z), depth 3 | Qubits 16 | Qulacs 18.9 ms | TensorCircuit unsupported | Faster – |
Strongly-entangling ansatz
RX/RY/RZ layers with a CNOT ring
| Workload | Qubits | Qulacs | TensorCircuit | Faster |
|---|---|---|---|---|
Expectation value, local observable, depth 3 | Qubits 4 | Qulacs 68.2 µs | TensorCircuit 224 µs | Faster Qulacs 3.3× |
Expectation value, local observable, depth 3 | Qubits 8 | Qulacs 177 µs | TensorCircuit 418 µs | Faster Qulacs 2.4× |
Expectation value, local observable, depth 3 | Qubits 12 | Qulacs 1.56 ms | TensorCircuit 1.46 ms | Faster TensorCircuit 1.1× |
Expectation value, local observable, depth 3 | Qubits 16 | Qulacs 30.0 ms | TensorCircuit 20.7 ms | Faster TensorCircuit 1.4× |
Expectation value, local observable, depth 3 | Qubits 20 | Qulacs 595 ms | TensorCircuit 467 ms | Faster TensorCircuit 1.3× |
Expectation value, local observable, depth 3 | Qubits 24 | Qulacs 13.2 s | TensorCircuit 13.0 s | Faster TensorCircuit 1.0× |
Expectation value, global observable (all_z), depth 3 | Qubits 4 | Qulacs 77.7 µs | TensorCircuit 229 µs | Faster Qulacs 2.9× |
Expectation value, global observable (all_z), depth 3 | Qubits 8 | Qulacs 222 µs | TensorCircuit 458 µs | Faster Qulacs 2.1× |
Expectation value, global observable (all_z), depth 3 | Qubits 12 | Qulacs 2.50 ms | TensorCircuit 2.01 ms | Faster TensorCircuit 1.2× |
Expectation value, global observable (all_z), depth 3 | Qubits 16 | Qulacs 50.6 ms | TensorCircuit 50.3 ms | Faster TensorCircuit 1.0× |
Expectation value, global observable (all_z), depth 3 | Qubits 20 | Qulacs 1.02 s | TensorCircuit 1.13 s | Faster Qulacs 1.1× |
Expectation value, global observable (all_z), depth 3 | Qubits 24 | Qulacs 21.4 s | TensorCircuit 26.2 s | Faster Qulacs 1.2× |
Gradient, local observable, depth 3 | Qubits 4 | Qulacs 278 µs | TensorCircuit 395 µs | Faster Qulacs 1.4× |
Gradient, local observable, depth 3 | Qubits 8 | Qulacs 822 µs | TensorCircuit 986 µs | Faster Qulacs 1.2× |
Gradient, local observable, depth 3 | Qubits 12 | Qulacs 8.63 ms | TensorCircuit 4.28 ms | Faster TensorCircuit 2.0× |
Gradient, local observable, depth 3 | Qubits 16 | Qulacs 180 ms | TensorCircuit 69.9 ms | Faster TensorCircuit 2.6× |
Data re-uploading QML
Repeated encode-and-rotate classifier
| Workload | Qubits | Qulacs | TensorCircuit | Faster |
|---|---|---|---|---|
Expectation value, global observable (all_z), depth 2 | Qubits 8 | Qulacs 185 µs | TensorCircuit 388 µs | Faster Qulacs 2.1× |
Expectation value, global observable (all_z), depth 2 | Qubits 12 | Qulacs 2.06 ms | TensorCircuit 1.20 ms | Faster TensorCircuit 1.7× |
Expectation value, global observable (all_z), depth 2 | Qubits 16 | Qulacs 41.8 ms | TensorCircuit 43.8 ms | Faster Qulacs 1.0× |
Expectation value, global observable (all_z), depth 2, batch of 32 | Qubits 8 | Qulacs 6.97 ms | TensorCircuit 2.00 ms | Faster TensorCircuit 3.5× |
Expectation value, global observable (all_z), depth 2, batch of 32 | Qubits 12 | Qulacs 67.1 ms | TensorCircuit 59.8 ms | Faster TensorCircuit 1.1× |
Expectation value, global observable (all_z), depth 2, batch of 32 | Qubits 16 | Qulacs 1.33 s | TensorCircuit 2.02 s | Faster Qulacs 1.5× |
Expectation value, global observable (all_z), depth 2, batch of 256 | Qubits 8 | Qulacs 48.4 ms | TensorCircuit 14.7 ms | Faster TensorCircuit 3.3× |
Expectation value, global observable (all_z), depth 2, batch of 256 | Qubits 12 | Qulacs 529 ms | TensorCircuit 497 ms | Faster TensorCircuit 1.1× |
Expectation value, global observable (all_z), depth 2, batch of 256 | Qubits 16 | Qulacs 10.6 s | TensorCircuit 16.9 s | Faster Qulacs 1.6× |
Expectation value, global observable (all_z), depth 2, batch of 1024 | Qubits 8 | Qulacs 193 ms | TensorCircuit 107 ms | Faster TensorCircuit 1.8× |
Expectation value, global observable (all_z), depth 2, batch of 1024 | Qubits 12 | Qulacs 2.13 s | TensorCircuit 2.17 s | Faster Qulacs 1.0× |
Expectation value, global observable (all_z), depth 2, batch of 1024 | Qubits 16 | Qulacs 42.7 s | TensorCircuit error | Faster – |
Gradient, local observable, depth 2 | Qubits 8 | Qulacs 606 µs | TensorCircuit 596 µs | Faster TensorCircuit 1.0× |
Gradient, local observable, depth 2 | Qubits 12 | Qulacs 5.81 ms | TensorCircuit 1.16 ms | Faster TensorCircuit 5.0× |
Gradient, local observable, depth 2 | Qubits 16 | Qulacs 117 ms | TensorCircuit 7.10 ms | Faster TensorCircuit 16.4× |
Gradient, local observable, depth 2, batch of 32 | Qubits 8 | Qulacs 19.3 ms | TensorCircuit 3.46 ms | Faster TensorCircuit 5.6× |
Gradient, local observable, depth 2, batch of 32 | Qubits 12 | Qulacs 189 ms | TensorCircuit 19.9 ms | Faster TensorCircuit 9.5× |
Gradient, local observable, depth 2, batch of 32 | Qubits 16 | Qulacs 3.76 s | TensorCircuit 424 ms | Faster TensorCircuit 8.9× |
Gradient, local observable, depth 2, batch of 256 | Qubits 8 | Qulacs 153 ms | TensorCircuit 26.0 ms | Faster TensorCircuit 5.9× |
Gradient, local observable, depth 2, batch of 256 | Qubits 12 | Qulacs 1.47 s | TensorCircuit 243 ms | Faster TensorCircuit 6.1× |
Gradient, local observable, depth 2, batch of 256 | Qubits 16 | Qulacs 30.0 s | TensorCircuit 3.51 s | Faster TensorCircuit 8.5× |
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.