Data processing

Abstract

Applications often run multiple tasks that need to share data with each other. When these tasks run repeatedly—like in a loop or across multiple iterations—keeping track of which data belongs to which iteration can be challenging. This disclosure describes techniques for using timestamps to help coordinate data between tasks. For example, when a first task generates data, that data gets tagged with a timestamp. Later, when a second task needs to use data from the first task, the second task may look at the timestamps to help it identify and use data from the matching iteration. This timestamp-based approach may help tasks stay synchronized even when running on different processors or computing platforms.

Dependencies are inferred from the claim text.