This repository serves accompanying data and documentation for the manuscript "Hybrid quantum programming with PennyLane Lightning on HPC platforms"
The structures of this repository is as follows:
DataCollection/
├─ distributed/
| ├─ LUMI_LKOKKOS_VQE
| | # Data, notebooks and scripts used for the distributed VQE workload
| | # running on 8 - 256 AMD MI250X GPUs on LUMI, using Lightning-Kokkos,
| | # OpenMPI+UCX and mpi4py.
| |
| ├─ Perlmutter_LGPU_MPI/
| | | # All materials related to fully-distributed Lightning-GPU workloads.
| | ├─ sampling_results/
| | | # Data and scripts for the 33-41 qubit sampling workload running on
| | | # 4 - 512 NVIDIA A100 (80GB) GPUs on Perlmutter, using Lightning-GPU
| | | # Cray-MPICH 8.1.25, mpi4py and CUDA 11.7.
| | |
| | ├─ sel_gradient_results/
| | | # Data and scripts for the 33-37 qubit distributed quantum gradient workload
| | | # running on 16 - 512 NVIDIA A100 (80GB) GPUs on Perlmutter, using
| | | # Lightning-GPU, Cray-MPICH 8.1.25, mpi4py and CUDA 11.7.
| | |
| ├─ Perlmutter_LGPU_QAOA/
| | | # All materials related to task-based circuit cutting Lightning-GPU workloads.
| | ├─ QAOA_QCUT_BATCHED_CUDA12/
| | | # Data and scripts for the 79-qubit circuit cutting workload, with 40k+
| | | # intermediate circuits, using batched execution, and running on between
| | | # 4 - 256 NVIDIA A100 (80GB) GPUs on Perlmutter, using Lightning-GPU
| | | # CUDA 12.2, and Ray.
| | |
| | ├─ QAOA_QCUT_ONDEMAND_CUDA12/
| | | # Data and scripts for the 79-qubit circuit cutting workload, with 40k+
| | | # intermediate circuits, using on-demand execution, and running on between
| | | # 4 - 256 NVIDIA A100 (80GB) GPUs on Perlmutter, using Lightning-GPU
| | | # CUDA 12.2, and Ray.
| | |
├─ single_node/
| ├─ GateKernels_LQUBIT/
| | | # Targeted kernel evaluation for LM, AVX2 and AVX512 implementations, with
| | | # comparisons against other HPC-focused simulators.
| ├─ VQE/
| | | # Comparison of Lightning suite across CPU (Lightning-Qubit, Lightning-Kokkos+OpenMP on AMD 3rd gen. Epyc)
| | | # and GPU (Lightning-GPU and Lightning-Kokkos+CUDA on NVIDIA A100) backends, using
| | | # VQE and a range of molecules from PennyLane's datasets.
The included requirements.txt contains all used dependencies for running the attached notebooks and plotting scripts.