I am doing distributed spectre simulation through an OCEAN script and i need to run close to 270 jobs in one go.
What i am noticing is that the most of the simulations are going through fine.But due to some LSF setting issue
few jobs are getting failed (not getting submitted)and it fails to generate even the spectre.out file.
I know it is more a problem in our LSF setup.
But i had couple of queries related to this (though easy to find a fix here rather than fixing our setup which may take more
1,Is there any way to provide the number of job submitted at a time to the LSF using Ocean? I checked the documentation
couldn't find from my side.
2,Due to this issue what is happening is that I am forced to re-submit the jobs which may again fail or forced to split my
runs into blocks there by editing my setup files.
So is there a way to tell spectre environment to re-run only the ones which have not even RUN (no spectre.out)?
This way may be in two steps I may be able to finish the set of runs.(without editing the setup files).
Any help is really appreciated.