Our benchmarking method for quantum emulators is to run growing circuits until we reach hardware limitations. We create square circuits, e.g. 24x24 or 30x30, denoting the number of qubits and the number of quantum gates (depth) respectively. These gates are randomly selected from Pauli X, Y, Z and Hadamar.
These squared-circuits, designed on Perceval SDK, were executed on popular local setups and on our Quantum as a Service across different platforms.
As depicted in Figure 2, our GPU-accelerated platforms exhibit a substantial computation speedup for equivalent circuit size, taking less than a second compared to 241 seconds for Apple M2 or 695 seconds for an Intel i7.
Moreover, in local setups we encounter limitations by running squared-circuits with more than 11 qubits due to excessive memory requirement. In contrast our H100 GPU-accelerated platform, enables us to extend up to 31 qubits in 2h.