Nitro, a high-throughput job scheduler from Adaptive Computing, is now available on Flux. Nitro is designed to schedule thousands to millions of tasks very quickly, working in conjunction with the existing Torque scheduler.
Nitro speeds up the process of scheduling very short jobs. Nitro lets researchers with large numbers of such jobs to submit one PBS job that will then run through a list of shorter tasks.
Adaptive Computing’s website describes it this way: “Nitro facilitates the execution of small compute tasks on a very large scale and without the overhead of individual scheduler jobs. Instead of creating individual jobs, Nitro combines all of the compute tasks into a single file. The file is then sent to Nitro as part of a job, and Nitro distributes the compute tasks across the allocated nodes. Tasks are executed on multiple threads on each compute node. Since the overhead of managing these tasks is small, most of the allocated compute resources can be spent executing the desired tasks.”