| Table of Contents | ||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
|
...
| Expand | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| ||||||||||
|
...
Command | Total MPI processes | CPU per MPI process | MPI processes per node | GPU card |
|---|---|---|---|---|
srun | -n, --ntasks | -c, --cpus-per-task | --ntasks-per-node | -G, --gpus |
mpirun/mpiexec | -n, -np | --map-by socket:PE=N | --map-by ppr:N:node | - |
aprun | -n, --pes | -d, --cpus-per-pe | -N, --pes-per-node | - |
Nevertheless, there There is usually no needs need to explicitly add these options to srun, since most srun options are automatically derived from sbatch, except for --cpus-per-task and a few others since, by default, Slurm will automatically derive them from sbatch. However, we recommend explicitly adding GPU binding options such as --gpus-per-task or --ntasks-per-gpu according to each software specification to srun. Please visit Slurm srun for more details.
| Note |
|---|
For multi-threaded applications, it is essential to specify |
...