Output file of lammps - number of GPUs

LAMMPS LAMMPS General Discussion
1

Hey guys,

How to find out number of gpus being used in my simulation?
I am running like:
mpirun -n 16 lmp_cuda -sf gpu -pk gpu 4 blocksize 128 neigh yes newton on split 1 tpa 4 device_type nvidiagpu
and reserved 4 nodes each has 4 GPU cards
but in output file I can see only 8 GPUs

Device 0: NVIDIA A100-PCIE-40GB, 108 CUs, 38/39 GB, 1.4 GHZ (Mixed Precision)

Device 1: NVIDIA A100-PCIE-40GB, 108 CUs, 1.4 GHZ (Mixed Precision)

Device 2: NVIDIA A100-PCIE-40GB, 108 CUs, 1.4 GHZ (Mixed Precision)

Device 3: NVIDIA A100-PCIE-40GB, 108 CUs, 1.4 GHZ (Mixed Precision)

Initializing Device and compiling on process 0…Done.

Initializing Devices 0-3 on core 0…Done.

Initializing Devices 0-3 on core 1…Done.

Initializing Devices 0-3 on core 2…Done.

Initializing Devices 0-3 on core 3…Done.

System init for write_data …

  • Using acceleration for sw/gpu:

  • with 4 proc(s) per device.

  • Horizontal vector operations: ENABLED

  • Shared memory system: No

Device 0: NVIDIA A100-PCIE-40GB, 108 CUs, 38/39 GB, 1.4 GHZ (Mixed Precision)

Device 1: NVIDIA A100-PCIE-40GB, 108 CUs, 1.4 GHZ (Mixed Precision)

Device 2: NVIDIA A100-PCIE-40GB, 108 CUs, 1.4 GHZ (Mixed Precision)

Device 3: NVIDIA A100-PCIE-40GB, 108 CUs, 1.4 GHZ (Mixed Precision)

2

You should ‘see’ only the number of GPUs on MPI rank 0, that is 4.

Do you have other output duplicated as well?
Or do you have multiple run/minimize commands?