Lightning

Usage

Development

API

  1. Docs
  2. Lightning AMD GPU device

Lightning AMD GPU device

The lightning.amdgpu device is an extension of PennyLane’s built-in lightning.qubit device. It extends the CPU-focused Lightning simulator to enable GPU-accelerated simulation of quantum state-vector evolution specifically on AMD GPUs.

A lightning.amdgpu device can be loaded using:

import pennylane as qml
dev = qml.device("lightning.amdgpu", wires=2)

The pre-built lightning.amdgpu device wheels are available for ROCm 7.0+ and MI300 series GPUs. For other versions of ROCm or AMD GPUs series, please refer to the Lightning-AMDGPU installation guide for instructions on building from source.

The lightning.amdgpu device is an instantiation of the lightning.kokkos device specifically for AMD GPUs. It inherits all features from the lightning.kokkos device, including support for a wide range of quantum operations and observables, as well as compatibility with PennyLane’s quantum functions and QNodes. For a comprehensive list of supported operations and observables, please refer to the Lightning Kokkos device documentation.

Note

  • It is not recommended to install both lightning.amdgpu and lightning.kokkos devices in the same environment, as this may lead to conflicts. Choose the device that best fits your hardware and simulation needs.

  • lightning.amdgpu requires libomp.so, which may present conflict with libraries requiring libgomp.so. Please report if you encounter such issues.

Previous
Next