You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The atmos_products job is running wgrib2 without specifying -ncpu, which is causing wgrib2 to over-thread when called during MPMD jobs, resulting in OpenMP errors on S4.
What should have happened?
OMP_NUM_THREADS should be specified using the env/<MACHINE>.env file specified thread value NTHREADS_UPP. -ncpu 1 should be specified inside of the ex script for atmos_products when running on S4 (and perhaps other machines as well).
What machines are impacted?
All or N/A
Steps to reproduce
Run an atmos_products job on S4 and check for OpenMP errors.
What is wrong?
The atmos_products job is running wgrib2 without specifying
-ncpu
, which is causing wgrib2 to over-thread when called during MPMD jobs, resulting in OpenMP errors on S4.What should have happened?
OMP_NUM_THREADS
should be specified using theenv/<MACHINE>.env
file specified thread valueNTHREADS_UPP
.-ncpu 1
should be specified inside of the ex script for atmos_products when running on S4 (and perhaps other machines as well).What machines are impacted?
All or N/A
Steps to reproduce
Run an atmos_products job on S4 and check for OpenMP errors.
Additional information
Reported by @souopgui on S4.
Do you have a proposed solution?
Addexport OMP_NUM_THREADS=${OMP_NUM_THREADS:-${NTHREADS_UPP:-1}}
to JGLOBAL_ATMOS_UPP.Add
-ncpu 1
towgrib2
calls inside of MPMD jobs.The text was updated successfully, but these errors were encountered: