Code execution
When using OpenMPI, binding is controlled using the –-bind-to parameter. To bind processes to cores, use --bind-to core. Possible other values can be found in the man page.
| Codeblock |
|---|
mpirun --bind-to core ./yourprogram |
Our Per default our hardware supports hyperthreading, allowing you start 192 processes on Cascade Lake machines (*96 partitions) and 80 on Skylake machines.If no specific requests regarding the number of tasks has been done, mpirun defaults to hyperthreading and starts cores*2 processes. If a number of tasks has been specified (either with -n or --tasks-per-node), mpirun honors this. For example:
| Codeblock |
|---|
| language | bash |
|---|
| title | Hyperthreading active, 80/192 Processes per Node will be started |
|---|
| collapse | true |
|---|
|
#!/bin/bash
#SBATCH -N 4
module load gcc/9.2.0
openmpi/gcc.9/3.1.5
cpus_per_node=${SLURM_JOB_CPUS_PER_NODE%\(*}
export OMP_NUM_THREADS=$(($cpus_per_node*$SLURM_JOB_NUM_NODES))
mpirun# Run multi-threaded application ./yourexe |
Note that more than one node, as in the example set with `-N 4` can only be used with a hybrid MPI-OpenMP job. For pure OpenMP programs, one '-N 1' is meaningful.
| Codeblock |
|---|
| language | bash |
|---|
| title | Hyperthreading disabled, 96 processes per node are started. |
|---|
| collapse | true |
|---|
|
#!/bin/bash
#SBATCH -N 4
#SBATCH --tasks-per-node 96
module load gcc/9.2.0
openmpi/gcc.9/3.1.5
tasks_per_node=${SLURM_TASKS_PER_NODE%\(*}
export OMP_NUM_THREADS=$(($tasks_per_node*$SLURM_JOB_NUM_NODES))
# Run mpirunmulti-threaded application ./yourexe |