发明名称 Signaling, ordering, and execution of dynamically generated tasks in a processing system
摘要 One embodiment of the present invention sets forth a technique for enabling the insertion of generated tasks into a scheduling pipeline of a multiple processor system allows a compute task that is being executed to dynamically generate a dynamic task and notify a scheduling unit of the multiple processor system without intervention by a CPU. A reflected notification signal is generated in response to a write request when data for the dynamic task is written to a queue. Additional reflected notification signals are generated for other events that occur during execution of a compute task, e.g., to invalidate cache entries storing data for the compute task and to enable scheduling of another compute task.
申请公布号 US8984183(B2) 申请公布日期 2015.03.17
申请号 US201113329169 申请日期 2011.12.16
申请人 Nvidia Corporation 发明人 Purcell Timothy John;Shah Lacky V.;Duluk, Jr. Jerome F.;Treichler Sean J.;Abdalla Karim M.;Cuadra Philip Alexander;Pharris Brian
分类号 G06F3/00;G06F9/46;G06F9/48 主分类号 G06F3/00
代理机构 Artegis Law Group, LLP 代理人 Artegis Law Group, LLP
主权项 1. A method of notifying a task manager that data is available for processing, the method comprising: generating a first notification signal that is transmitted to the task manager, wherein the notification signal specifies a first offset and a first delta value; storing first data in one or more entries indicated by the first offset and the first delta value in a queue corresponding to a first processing task; generating a second notification signal that is transmitted to the task manager, wherein the notification signal comprises a second offset and a second delta value; storing second data in one or more entries indicated by the second offset and the second delta value in the queue corresponding to the first processing task; receiving the first notification signal and a second notification signal; determining if the first offset and the first delta value can be coalesced with the second offset and the second delta value, and, if so, generating a coalesced offset and a coalesced delta value; and scheduling, by the task manager, processing of at least one of the first data and the second data for the first processing task in a processing cluster array.
地址 Santa Clara CA US