Strong scaling performance, ReaxFF HNS benchmark, P100, 64K atoms
Performance in millions of atom-timesteps / second / node
Nodes | Kokkos/Cuda-1 (mpi) | Kokkos/Cuda-2 (mpi) | Kokkos/Cuda-4 (mpi) | |
1 | 0.5986 (1) | 0.9801 (2) | 1.366 (4) | |
2 | 0.4765 (1) | 0.6596 (2) | 0.761 (4) | |
4 | 0.3313 (1) | 0.377 (2) | 0.3907 (4) | |
8 | 0.1885 (1) | 0.1979 (2) | 0.205 (4) |
1 | mpirun -np 1 --npernode 1 --bind-to core lmp_ride100_kokkos_cuda -sf kk -k on g 1 -pk kokkos neigh half neigh/qeq full newton on comm device binsize 11.0 -v x 4 -v y 8 -v z 6 -v t 100 -in in.reaxc.hns.kokkos_cuda.steps -nocite -log log.lammps.date=17Jan18.model=hns.machine=ride100.pkg=kokkos_cuda.kind=strong.size=64K.node=1.mpi=1.gpu=1 |
2 | mpirun -np 2 --npernode 1 --bind-to core lmp_ride100_kokkos_cuda -sf kk -k on g 1 -pk kokkos neigh half neigh/qeq full newton on comm device binsize 11.0 -v x 4 -v y 8 -v z 6 -v t 100 -in in.reaxc.hns.kokkos_cuda.steps -nocite -log log.lammps.date=17Jan18.model=hns.machine=ride100.pkg=kokkos_cuda.kind=strong.size=64K.node=2.mpi=1.gpu=1 |
4 | mpirun -np 4 --npernode 1 --bind-to core lmp_ride100_kokkos_cuda -sf kk -k on g 1 -pk kokkos neigh half neigh/qeq full newton on comm device binsize 11.0 -v x 4 -v y 8 -v z 6 -v t 100 -in in.reaxc.hns.kokkos_cuda.steps -nocite -log log.lammps.date=17Jan18.model=hns.machine=ride100.pkg=kokkos_cuda.kind=strong.size=64K.node=4.mpi=1.gpu=1 |
8 | mpirun -np 8 --npernode 1 --bind-to core lmp_ride100_kokkos_cuda -sf kk -k on g 1 -pk kokkos neigh half neigh/qeq full newton on comm device binsize 11.0 -v x 4 -v y 8 -v z 6 -v t 100 -in in.reaxc.hns.kokkos_cuda.steps -nocite -log log.lammps.date=17Jan18.model=hns.machine=ride100.pkg=kokkos_cuda.kind=strong.size=64K.node=8.mpi=1.gpu=1 |
1 | mpirun -np 2 --npersocket 1 --bind-to core lmp_ride100_kokkos_cuda -sf kk -k on g 2 -pk kokkos neigh half neigh/qeq full newton on comm device binsize 11.0 -v x 4 -v y 8 -v z 6 -v t 100 -in in.reaxc.hns.kokkos_cuda.steps -nocite -log log.lammps.date=17Jan18.model=hns.machine=ride100.pkg=kokkos_cuda.kind=strong.size=64K.node=1.mpi=2.gpu=2 |
2 | mpirun -np 4 --npersocket 1 --bind-to core lmp_ride100_kokkos_cuda -sf kk -k on g 2 -pk kokkos neigh half neigh/qeq full newton on comm device binsize 11.0 -v x 4 -v y 8 -v z 6 -v t 100 -in in.reaxc.hns.kokkos_cuda.steps -nocite -log log.lammps.date=17Jan18.model=hns.machine=ride100.pkg=kokkos_cuda.kind=strong.size=64K.node=2.mpi=2.gpu=2 |
4 | mpirun -np 8 --npersocket 1 --bind-to core lmp_ride100_kokkos_cuda -sf kk -k on g 2 -pk kokkos neigh half neigh/qeq full newton on comm device binsize 11.0 -v x 4 -v y 8 -v z 6 -v t 100 -in in.reaxc.hns.kokkos_cuda.steps -nocite -log log.lammps.date=17Jan18.model=hns.machine=ride100.pkg=kokkos_cuda.kind=strong.size=64K.node=4.mpi=2.gpu=2 |
8 | mpirun -np 16 --npersocket 1 --bind-to core lmp_ride100_kokkos_cuda -sf kk -k on g 2 -pk kokkos neigh half neigh/qeq full newton on comm device binsize 11.0 -v x 4 -v y 8 -v z 6 -v t 100 -in in.reaxc.hns.kokkos_cuda.steps -nocite -log log.lammps.date=17Jan18.model=hns.machine=ride100.pkg=kokkos_cuda.kind=strong.size=64K.node=8.mpi=2.gpu=2 |
1 | mpirun -np 4 --npersocket 2 --bind-to core lmp_ride100_kokkos_cuda -sf kk -k on g 4 -pk kokkos neigh half neigh/qeq full newton on comm device binsize 11.0 -v x 4 -v y 8 -v z 6 -v t 100 -in in.reaxc.hns.kokkos_cuda.steps -nocite -log log.lammps.date=17Jan18.model=hns.machine=ride100.pkg=kokkos_cuda.kind=strong.size=64K.node=1.mpi=4.gpu=4 |
2 | mpirun -np 8 --npersocket 2 --bind-to core lmp_ride100_kokkos_cuda -sf kk -k on g 4 -pk kokkos neigh half neigh/qeq full newton on comm device binsize 11.0 -v x 4 -v y 8 -v z 6 -v t 100 -in in.reaxc.hns.kokkos_cuda.steps -nocite -log log.lammps.date=17Jan18.model=hns.machine=ride100.pkg=kokkos_cuda.kind=strong.size=64K.node=2.mpi=4.gpu=4 |
4 | mpirun -np 16 --npersocket 2 --bind-to core lmp_ride100_kokkos_cuda -sf kk -k on g 4 -pk kokkos neigh half neigh/qeq full newton on comm device binsize 11.0 -v x 4 -v y 8 -v z 6 -v t 100 -in in.reaxc.hns.kokkos_cuda.steps -nocite -log log.lammps.date=17Jan18.model=hns.machine=ride100.pkg=kokkos_cuda.kind=strong.size=64K.node=4.mpi=4.gpu=4 |
8 | mpirun -np 32 --npersocket 2 --bind-to core lmp_ride100_kokkos_cuda -sf kk -k on g 4 -pk kokkos neigh half neigh/qeq full newton on comm device binsize 11.0 -v x 4 -v y 8 -v z 6 -v t 100 -in in.reaxc.hns.kokkos_cuda.steps -nocite -log log.lammps.date=17Jan18.model=hns.machine=ride100.pkg=kokkos_cuda.kind=strong.size=64K.node=8.mpi=4.gpu=4 |