Hi guys,
I trying to use the rankfile in OpenMPI 4.1 but it crashes. I got the following
message: [genji608:451977] [[64242,0],0] ORTE_ERROR_LOG: Data unpack would read
past end of buffer in file base/odls_base_default_fns.c at line 612
I use the following script. It's fine if using the last command of the script.
I may do something wrong. In that case, please let me know !
Thanks in advance for the time spend on that problem,
Ludovic
[cid:image001.gif@01D088CD.E4C3E120]
Ludovic HABLOT
HPC Applications Expert
Applications & Performance Group
Big Data & Security
BULL, an ATOS company
#!/bin/bash
#SBATCH -N 2
#SBATCH -n 2
#SBATCH --exclusive
module load intel/2018.5.274
module load openmpi-hpcx/hpcx-2.3.0/4.1.0a1/icc-2018
rank=0
rm -Rf rankfile
for node in $(nodeset -e $SLURM_NODELIST)
do
echo rank $rank=$node slot=0-4 >> rankfile
rank=$[rank + 1]
done
echo "" >> rankfile
cat rankfile
export bin=hostname
mpirun -V
mpirun -rf rankfile -n $SLURM_NTASKS --oversubscribe --report-bindings $bin
#mpirun -host $(nodeset -S',' -e $SLURM_NODELIST) --map-by socket:PE=5 -n
$SLURM_NTASKS --report-bindings --oversubscribe $bin
Output with the first command
rank 0=genji608 slot=0-4
rank 1=genji609 slot=0-4
mpirun (Open MPI) 4.1.0a1
Report bugs to http://www.open-mpi.org/community/help/
[genji608:451977] [[64242,0],0] ORTE_ERROR_LOG: Data unpack would read past end
of buffer in file base/odls_base_default_fns.c at line 612
Output with the second command
ank 0=genji608 slot=0-4
rank 1=genji609 slot=0-4
mpirun (Open MPI) 4.1.0a1
Report bugs to http://www.open-mpi.org/community/help/
[genji609:444658] MCW rank 1 bound to socket 0[core 0[hwt 0-1]], socket 0[core
1[hwt 0-1]], socket 0[core 2[hwt 0-1]], socket 0[core 3[hwt 0-1]], socket
0[core 4[hwt 0-1]]:
[BB/BB/BB/BB/BB/../../../../../../../../../../../../../../..][../../../../../../../../../../../../../../../../../../../..]
[genji608:456450] MCW rank 0 bound to socket 0[core 0[hwt 0-1]], socket 0[core
1[hwt 0-1]], socket 0[core 2[hwt 0-1]], socket 0[core 3[hwt 0-1]], socket
0[core 4[hwt 0-1]]:
[BB/BB/BB/BB/BB/../../../../../../../../../../../../../../..][../../../../../../../../../../../../../../../../../../../..]
genji609
genji608
[cid:image001.gif@01D088CD.E4C3E120]
Ludovic HABLOT
HPC Applications Expert
Applications & Performance Group
Big Data & Security
BULL, an ATOS company
ludovic.hab...@atos.net<mailto:ludovic.hab...@atos.net>
_______________________________________________
users mailing list
users@lists.open-mpi.org
https://lists.open-mpi.org/mailman/listinfo/users
