Originally, I have used
CURRENT_JOB = mdb.Job(atTime=None, contactPrint=OFF, description='', echoPrint=OFF, # CREATE JOB
explicitPrecision=SINGLE, getMemoryFromAnalysis=True, historyPrint=OFF,
memory=90, memoryUnits=PERCENTAGE, model=NAME, modelPrint=OFF,
multiprocessingMode=DEFAULT, name='Job-{0}'.format(i), nodalOutputPrecision=SINGLE,
numCpus=4, numDomains=4, numGPUs=0, queue=None, resultsFormat=ODB, scratch=
'', type=ANALYSIS, userSubroutine='', waitHours=0, waitMinutes=0)
CURRENT_JOB.submit() # SUBMIT
CURRENT_JOB.waitForCompletion()
I have tried with MPI parallel, however the tasks are still taking quite a while, even using many CPUs:
CURRENT_JOB = mdb.Job(atTime=None, contactPrint=OFF, description='', echoPrint=OFF, # CREATE JOB
explicitPrecision=SINGLE, getMemoryFromAnalysis=True, historyPrint=OFF,
memory=90, memoryUnits=PERCENTAGE, model=NAME, modelPrint=OFF,
multiprocessingMode=MPI, name=NAME, nodalOutputPrecision=SINGLE,parallelizationMethodExplicit=DOMAIN,
numCpus=20, numDomains=20, numGPUs=0, queue=None,resultsFormat=ODB, scratch=
'', type=ANALYSIS, userSubroutine='', waitHours=0, waitMinutes=0)
CURRENT_JOB.submit() # SUBMIT
CURRENT_JOB.waitForCompletion()
Is there a better way to send these jobs to the hpc? Thank you