The compilation process produces an executable called `main`, run it with `mpirun`
The compilation process produces an executable called `main`, run it with `mpirun`
The suggestion is to run it with one mpi task per socket.
The suggestion is to run it with one mpi task per socket.
...
@@ -13,7 +13,7 @@ The suggestion is to run it with one mpi task per socket.
...
@@ -13,7 +13,7 @@ The suggestion is to run it with one mpi task per socket.
- -o --out-data *(optional)* output path for the data, the datafile is shuffled between mpi tasks and datapoints are ordered default is `out_data`
- -o --out-data *(optional)* output path for the data, the datafile is shuffled between mpi tasks and datapoints are ordered default is `out_data`
- -a --out-assignment *(optional)* output path for the cluster assignment output ranges [0 ... Nc - 1] for core points halo points have indices [-Nc ... -1] conversion of idx for an halo point is `cluster_idx = -halo_idx - 1`, default is `out_assignment`
- -a --out-assignment *(optional)* output path for the cluster assignment output ranges [0 ... Nc - 1] for core points halo points have indices [-Nc ... -1] conversion of idx for an halo point is `cluster_idx = -halo_idx - 1`, default is `out_assignment`