发明名称 Routing image data across on-chip networks
摘要 A network of switches may be adapted to route image data to one or more processor cores based on tags associated with data samples, where each tag includes at least one reference-space coordinate value. When image data is received by the network, the image data may be spatially transformed to a reference space, e.g., the physical space that is represented by the image data, to generate the data samples and each data sample may be tagged with a corresponding reference space coordinate value and routed through the network to one or more of the processors according to the tag.
申请公布号 US9189448(B2) 申请公布日期 2015.11.17
申请号 US200912545063 申请日期 2009.08.20
申请人 EMPIRE TECHNOLOGY DEVELOPMENT LLC 发明人 Wolfe Andrew;Conte Tom
分类号 G06F15/173 主分类号 G06F15/173
代理机构 Ren-Sheng International 代理人 Ren-Sheng International
主权项 1. An apparatus that is arranged to process image data, the apparatus comprising: a plurality of processors; and a plurality of switches arranged in a network, wherein the network is arranged to use the switches to selectively route data samples through the network to one or more of the plurality of processors based on tags associated with the data samples, wherein each tag includes at least one reference-space coordinate value of a reference space that is spatially transformed from an image space, the data samples each represents a different portion of the reference space, and each of the plurality of processors is assigned to one or more of the different portions of the reference space, and wherein each data sample is self-routed through the network by use of a self-routing process in which the switches are programmed with a set of distribution rules that map reference-space coordinate values, contained in the tags of the data samples, to the plurality of processors, and wherein the network is arranged to use the switches to modify routing of the data samples based on time-dependent changes, including at least one of: predicted changes in the reference space, andchanges in the reference space that have skewed a distribution of computational workload among the plurality of processors.
地址 Wilmington DE US