发明名称 Consolidated crossbar that supports a multitude of traffic types
摘要 One embodiment of the present invention sets forth a crossbar unit that is coupled to a plurality of client subsystems. The crossbar unit is configured to transmit data packets between the client subsystems and includes a high-bandwidth channel and a narrow-bandwidth channel. The high-bandwidth channel is used for transmitting large data packets, while the narrow-bandwidth is used for transmitting smaller data packets. The transmission of data packets may be prioritized based on the source and destination clients as well as the type of data being transmitted. Further, the crossbar unit includes a buffer mechanism for buffering data packets received from source clients until those data packets can be received by the destination clients.
申请公布号 US9098383(B1) 申请公布日期 2015.08.04
申请号 US201012792684 申请日期 2010.06.02
申请人 NVIDIA CORPORATION 发明人 Treichler Sean J.;Mrazek Dane T.;Tang Yin Fung (David);Glasco David B.;Case Colyn Scott;Kilgariff Emmett M.
分类号 G06F13/00;H04J3/14;G06F5/00;H04L12/26;H04J3/16;H04J3/22;H04L12/54 主分类号 G06F13/00
代理机构 Artegis Law Group, LLP 代理人 Artegis Law Group, LLP
主权项 1. A method for transmitting data packets between a plurality of client subsystems, the method comprising: receiving a plurality of higher-bandwidth data packets and a plurality of lower-bandwidth data packets from the plurality of client subsystems, wherein each higher-bandwidth data packet and each lower-bandwidth data packet is associated with a source client subsystem, one or more destination client subsystems and a data type, and wherein each client subsystem is associated with a plurality of different higher-bandwidth memory buffers for storing higher-bandwidth data packets and a plurality of different lower-bandwidth memory buffers for storing lower-bandwidth data packets; storing the plurality of higher-bandwidth data packets and the plurality of lower-bandwidth data packets in memory buffers by: storing each higher-bandwidth data packet in a higher-bandwidth memory buffer associated with the source client subsystem associated with the higher-bandwidth data packet; andstoring each lower-bandwidth data packet in a lower-bandwidth memory buffer associated with the source client subsystem associated with the lower-bandwidth data packet; and at a first clock cycle, selecting a first higher-bandwidth data packet for transmission to the one or more destination client subsystems associated with the first higher-bandwidth data packet based on one or more pre-determined conditions; or at a first clock cycle, selecting a first lower-bandwidth data packet for transmission to the one or more destination client subsystems associated with the first lower-bandwidth based on the one or more pre-determined conditions, and, after the first clock cycle, determining whether a second higher-bandwidth data packet corresponding to the first lower-bandwidth data packet is available in the buffer memories for transmission.
地址 Santa Clara CA US