Quantum Simulation Infrastructure Calculator
Calculate the memory a classical simulator needs to hold a quantum state — 2ⁿ amplitudes for a state vector, 4ⁿ entries for a density matrix — and the nodes required to hold it, plus the largest system one node can simulate.
Results
- State memory (34 qubits)
- 256 GiBcalculated
- Nodes required1 if the simulator needs a power-of-two rank count
- 1calculated
- Largest system on one node
- 34 qubitscalculated
| Qubits | Memory | Nodes |
|---|---|---|
| 30 | 16 GiB | 1 |
| 32 | 64 GiB | 1 |
| 34 | 256 GiB | 1 |
| 36 | 1 TiB | 3 |
| 38 | 4 TiB | 10 |
Each added qubit doubles state-vector memory (density matrices quadruple). Distributed simulation is dominated by all-to-all communication for gates on high-order qubits.
Tensor-network and stabiliser simulators can handle more qubits when entanglement or gate sets are restricted.
What it calculates
- State memory
- Nodes required
- Maximum qubits on one node
- Memory table by qubit count
Worked example
Using the default inputs below; change any input above to recalculate.
| Qubits | 34 |
|---|---|
| Method | State vector (pure states) |
| Amplitude precision | complex128 (16 bytes) |
| Simulator working-memory factor | 1 |
| Memory per node (RAM or GPU) | 512 GiB |
| Usable fraction | 85 % |
| State memory (34 qubits) | 256 GiB | calculated |
|---|---|---|
| Nodes required | 1 | calculated |
| Largest system on one node | 34 qubits | calculated |
Methodology
A pure n-qubit state is 2ⁿ complex amplitudes; a density matrix is 2ⁿ × 2ⁿ. Memory = entries × bytes per complex number × working-memory factor, divided by usable memory per node.
Formulas
- State vector
- M = 2ⁿ × b
- Density matrix
- M = 4ⁿ × b
Assumptions
- complex64 = 8 bytes, complex128 = 16 bytes per entry.
Limitations
- Run time depends on gate count and communication, not modelled here.