If you want one core per job, you could do 240 jobs simultaneously. Whether the machines can actually handle the jobs is another question - you say nothing of what is in you’re pipeline and how much multithreading is done within any of the stages.
you need to use the --sshloginfile and associated args > On May 2, 2020, at 9:03 AM, Nyasimi Festus <nfes...@yahoo.com> wrote: > > Hi, > > I have a few issues in using GNU Parallel. > > Problem > I have 150 jobs to run through the same pipeline and each job has different > core requirements but utilizes the same pipeline. > I am interested in using GNU parallel to run the jobs in parallel at least 10 > at each single time ensuring I utilize all the cores available at my disposal. > Is there a way I have to write my parallel command to ensure I utilize all > the cores available to me? > > I have 10 nodes each with 24 cores available for this work. > > I will really appreciate assistance on this. > >
