storage icon indicating copy to clipboard operation
storage copied to clipboard

mpirun: Error: unknown option "-o"

Open Linzsd opened this issue 1 year ago • 2 comments

Linzsd avatar Apr 08 '24 05:04 Linzsd

I've also seen this issue when running with multiple hosts. The error looks to come from

benchmark.sh. line 280

mpirun -hosts $hosts -np $num_accelerators python3 dlio_benchmark/dlio_benchmark/main.py --config-path=$CONFIG_PATH workload=$config_name ++workload.workflow.generate_data=False ++workload.workflow.train=True ${prefixed_array[@ ]} ${EXTRA_PARAMS[@]}

man mpirun ... Specifying Host Nodes Host nodes can be identified on the mpirun command line with the -host option

Note -host vs -hosts

wpower-work avatar May 17 '24 15:05 wpower-work

The mpirun/mpiexec command differs from system to system. It is good to modify ./benchmark.sh accordingly.

Maybe Johnu can add this note in the documentation.

Huihuo

On May 17, 2024, at 10:41 AM, William Power @.***> wrote:

I've also seen this issue when running with multiple hosts. The error looks to come from

benchmark.sh. line 280

mpirun -hosts $hosts -np $num_accelerators python3 dlio_benchmark/dlio_benchmark/main.py --config-path=$CONFIG_PATH workload=$config_name ++workload.workflow.generate_data=False ++workload.workflow.train=True ${prefixed_array[@ ]} ${EXTRA_PARAMS[@]}

man mpirun ... Specifying Host Nodes Host nodes can be identified on the mpirun command line with the -host option

Note -host vs -hosts

— Reply to this email directly, view it on GitHub https://github.com/mlcommons/storage/issues/59#issuecomment-2117873344, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABMCS3IY4XBXU5SH2VAK4W3ZCYQLTAVCNFSM6AAAAABF37MAFKVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCMJXHA3TGMZUGQ. You are receiving this because you are subscribed to this thread.

zhenghh04 avatar May 17 '24 16:05 zhenghh04