====== Performance of the LiH-HFX Benchmark on Piz Daint ======

^ Nodes ^ Cores ^ Best time (No GPU) (s) ^ Config ^ Time with GPU, 8 OMP threads per MPI task (s) ^
| 128 | 1024 | 1240.84 | 4 OMP thread per MPI task | 2390.252 |
| 256 | 2048 | 464.047 | 1 OMP thread per MPI task | 693.845 |
| 512 | 4096 | 253.007 | 1 OMP threads per MPI task | 313.34 |
| 1024 | 8192 | 149.176 | 2 OMP threads per MPI task | 201.779 |
| 2048 | 16384 | 97.25 | 2 OMP threads per MPI task | - |
| 4096 | 32768 | 66.051 | 4 OMP threads per MPI task | - |

{{:performance:lih-hfx-comparison-piz-daint-piz-daint-gpu.png?direct|}}