| 摘要 |
A method and apparatus for scheduling processing jobs is described. In one embodiment, a scheduler receives a request to process one or more computation jobs. The scheduler generates a size metric corresponding to a size of an executable image of each computation job and a corresponding data set associated with each computation job. The scheduler adjusts a priority of each computation job based on a system configuration setting and schedules the process of each computation job according to the priority of each computation job. In another embodiment, the scheduler distributes the plurality of computation jobs on one or more processor of a computing system, where the system configuration setting prioritizes a computation job with a smaller size metric than a computation job with a larger size metric. In another embodiment, the scheduler distributes the computation jobs across a network of computing systems with one or more computation jobs distributed over one or more computing systems, where the system configuration setting prioritizes a computation job with a smaller size metric than a computation job with a larger size metric. |
