HPC Acceleration (GPU & QuEST)

GPU Backend (NVIDIA cuStateVec & Custom Kernels)

Maestro supports hardware-accelerated simulation on NVIDIA GPUs through cuStateVec and custom CUDA kernels.

Availability & Initialization

import maestro
 
# Check availability
if maestro.is_gpu_available():
    print("GPU acceleration available")
    maestro.init_gpu()

GPU Simulation

# GPU Statevector
gpu_sv_config = maestro.SimulatorConfig(
    simulator_type=maestro.SimulatorType.Gpu,
    simulation_type=maestro.SimulationType.Statevector,
)
result = maestro.simple_execute(qasm, config=gpu_sv_config, shots=1000)
 
# GPU Matrix Product State (MPS)
gpu_mps_config = maestro.SimulatorConfig(
    simulator_type=maestro.SimulatorType.Gpu,
    simulation_type=maestro.SimulationType.MatrixProductState,
    max_bond_dimension=64,
)
result_mps = maestro.simple_execute(qasm, config=gpu_mps_config, shots=1000)

---

QuEST Backend (MPI Distributed Simulation)

Maestro integrates the QuEST engine for distributed statevector simulation across cluster nodes.

Running with QuEST

import maestro
 
if maestro.is_quest_available():
    maestro.init_quest()
 
quest_config = maestro.SimulatorConfig(
    simulator_type=maestro.SimulatorType.QuestSim,
    simulation_type=maestro.SimulationType.Statevector,
)
result = maestro.simple_execute(qasm, config=quest_config, shots=1024)

Note

QuEST only supports SimulationType.Statevector. Attempting to use other simulation methods will raise an exception.

---

Graceful Hardware Fallback Pattern

For portable workflows across GPU workstations, clusters, and developer laptops:

import maestro
 
if maestro.is_gpu_available() and maestro.init_gpu():
    backend = maestro.SimulatorType.Gpu
elif maestro.is_quest_available() and maestro.init_quest():
    backend = maestro.SimulatorType.QuestSim
else:
    backend = maestro.SimulatorType.QCSim
 
config = maestro.SimulatorConfig(simulator_type=backend)
result = maestro.simple_execute(qasm, config=config, shots=1024)