WO2024016825A1 - 一种通信芯片及数据交换装置 - Google Patents

一种通信芯片及数据交换装置 Download PDF

Info

Publication number
WO2024016825A1
WO2024016825A1 PCT/CN2023/095556 CN2023095556W WO2024016825A1 WO 2024016825 A1 WO2024016825 A1 WO 2024016825A1 CN 2023095556 W CN2023095556 W CN 2023095556W WO 2024016825 A1 WO2024016825 A1 WO 2024016825A1
Authority
WO
WIPO (PCT)
Prior art keywords
die
message
switching
network processing
communication chip
Prior art date
Application number
PCT/CN2023/095556
Other languages
English (en)
French (fr)
Inventor
张先富
Original Assignee
华为技术有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 华为技术有限公司 filed Critical 华为技术有限公司
Publication of WO2024016825A1 publication Critical patent/WO2024016825A1/zh

Links

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L45/00—Routing or path finding of packets in data switching networks
    • H04L45/60—Router architectures
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L49/00—Packet switching elements
    • H04L49/10—Packet switching elements characterised by the switching fabric construction

Definitions

  • the present application relates to the field of communication technology, and in particular, to a communication chip and a data exchange device.
  • a router is a hardware device that connects two or more networks. It acts as a gateway between networks. It is a network device that can read the destination address in a data packet and decide how to transmit it. The router can identify different network protocols, convert the destination addresses of data packets of different network protocols, and then transmit the data packets to the designated location according to the best route according to the corresponding routing algorithm.
  • routers include multiple network processing (NP) chips and multiple switching (SW) chips.
  • the multiple NP chips are respectively installed on multiple line cards (line cards). , LC), the plurality of switching chips are provided on the backplane (BP), and the backplane is also provided with a board-level serial (serdes) interface.
  • the multiple line cards communicate with the board-level serial interface through the board-level serial interface.
  • Multiple switch chips are connected on the backplane.
  • the NP chip can be used to receive data from the external network and forward the data to the switching chip, or to receive data output by the switching chip and forward the data to the external network.
  • the switching chip is used to receive the data forwarded by the NP chip, and after data exchange, the data is sent to the corresponding NP chip.
  • This application provides a communication chip and data exchange device for reducing the power consumption, cost and volume of the data exchange network.
  • a communication chip in a first aspect, includes: a plurality of switching dies and a plurality of network processing dies, and any one of the plurality of switching dies is connected to other switching dies to achieve distribution.
  • any one of the plurality of network processing dies includes an external port and an internal port, and the internal port is connected to other network processing dies, or with other network processing dies and the plurality of switching dies.
  • One of the exchanged grains is connected.
  • any network processing die among the plurality of network processing die is used to: receive the first message from outside the communication chip through the external port, obtain the purpose information of the first message, and obtain the purpose information of the first message through the internal port. Send a second message.
  • the second message includes the first message and the destination information.
  • the destination information is used to indicate the destination network processing chip of the first message.
  • the destination information is the destination address or the destination network processing chip. Grain identification. Any switching die among the plurality of switching die is used to: receive the second message, and when it is determined according to the purpose information that the preset condition is not met, send the second message to the connected switching die, or when When it is determined that the preset condition is met, a second message is sent to the connected first network processing die.
  • the first network processing die is configured to receive a second message from a connected switching die and send the second message to the outside through the external port, or to send the second message to other devices through the internal port.
  • the preset condition Including: the switching die among the plurality of switching die that is connected to the destination network processing die, or the one that is connected to the destination network processing die through other network processing die and has the smallest number of connected network processing die.
  • the communication chip includes a plurality of switching dies and a plurality of network processing dies.
  • the plurality of switching dies and the plurality of network processing dies can be connected within the communication chip, and the multiple networks
  • the processing die can be used to receive external messages or send messages to the outside.
  • the multiple switching die can be used to realize distributed switching of messages, thereby realizing data exchange within the communication chip, which is different from the existing technology through multiple Compared with a data exchange network composed of multiple switching chips, a network processing chip can greatly reduce power consumption and volume, while reducing the heat dissipation cost and deployment space of equipment using this communication chip to meet the current requirements for green, low-carbon and sustainable the need for sustainable development.
  • the plurality of exchange grains include a first exchange grain and a second exchange grain, the second exchange grain is connected to the first exchange grain, and the first exchange grain
  • the second switching die is closest to the source network processing die, and the second switching die is closest to the destination network processing die; the first switching die is used to: receive the second message, and determine based on the destination information that the first switching die does not satisfy the When the conditions are preset, send the second message to the second switching die; the second switching die is used to: receive the second message from the first switching die, and determine the second switching die based on the purpose information When the preset condition is met, the second message is sent to the first network processing die.
  • the first switching die when the first switching die closest to the source network processing die of the message and the second switching die closest to the destination network processing die of the message are not connected, the first switching die After receiving the message, the chip can send it to the second switching chip, so that the second switching chip switches the message to the destination network processing chip.
  • the plurality of exchange grains also include a third exchange grain, and the third exchange grain is connected to the second exchange grain, that is, the first exchange grain, the second exchange grain
  • the die and the third exchange die are connected in sequence.
  • the first exchange die is closest to the source network processing die, and the third exchange die is closest to the destination network processing die.
  • the second exchange die is also used: when determining the third When the second switching die does not meet the preset condition, it sends a second message to the third switching die; the third switching die is used to: receive the second message from the second switching die and process it according to the purpose information. When it is determined that the third switching chip meets the preset condition, the second message is sent to the first network processing chip.
  • the first switching die when the first switching die closest to the source network processing die of the message and the third switching die closest to the destination network processing die of the message are not connected, and these two switches When the die are all connected to the first switching die, the first switching die can send the message to the second switching die after receiving the message, and the second switching die sends the message to the third switching die, so that The third switching chip switches the message to the destination network processing chip.
  • the first switching die is closest to the source network processing die, and the third switching die is closest to the destination network processing die.
  • the first switching die is specifically configured to: when sending a second message to the second switching die, send the second message to one of the at least two second switching die.
  • the first switching die can flexibly select one of the second switching die to send the second message. , thus avoiding the problem of long packet transmission delays caused by congestion on certain transmission paths.
  • any switching die among the plurality of switching die is also used to: when sending the second message to the first network processing die, send the second message to the first network processing die.
  • the processing die sends path indication information, and the path indication information is used to indicate the sending path of the second message to control the transmission of the second message in the first network processing die;
  • the first network processing die is also used to: Receive the path indication information and pass the external port according to the path indication information Send the second message to the outside, or send the second message to other network processing chips through the internal port according to the path indication information.
  • the switching die can control the first network processing by controlling the sending path of the second message in the first network processing die.
  • the chip forwards the message to the destination network processing chip to avoid incorrect exchange of messages.
  • the first network processing die includes a first selection switch.
  • the first selection switch includes a connection end, a first selection end and a second selection end.
  • the connection end of the first selection switch Connected to one of the plurality of switching dies, the first selection end of the first selection switch is connected to the external port, the second selection end of the first selection switch is connected to other network processing dies; the first network processing die It is also used for: when sending the second message to the outside through the external port according to the path indication information, connecting the connection end of the first selection switch and the first selection end of the first selection switch; or, the first network processing chip
  • the chip is also used to connect the connection terminal of the first selection switch and the second selection terminal of the first selection switch when sending the second message to other network processing chips through the internal port according to the path indication information.
  • the switching chip controls the first selection switch to forward the message to the destination network processing chip by controlling different selection terminals of the first selection switch in the first network processing chip to avoid Incorrect exchange of messages.
  • the first selection switch includes at least two second selection terminals, and the at least two second selection terminals are respectively connected to at least two internal ports of the first network processing die. , the at least two internal ports are respectively connected to at least two of the plurality of network processing chips; the first network processing chip is also used to: send the second message through the internal port according to the path indication information.
  • the connection terminal of the first selection switch is turned on and the second selection terminal corresponding to the transmission path indicated by the path indication information.
  • the switching die controls different selection ends of the first selection switch in the first network processing die to select a corresponding transmission path and forward the message to the destination network processing die.
  • the target network processing die is not connected to any exchange die among the plurality of exchange die, and there are at least two first network processing die among the plurality of network processing die.
  • the network processing chips are connected to the second switching chip and the destination network processing chip; the second switching chip is also used to: when sending the second message to the first network processing chip, send the second message to the at least two network processing chips.
  • One of the first network processing dies sends the second message.
  • the second switching die can flexibly select the first network processing die for forwarding the second message. granules, thereby avoiding the problem of long packet transmission delays due to congestion on certain transmission paths.
  • a second network processing die among the plurality of network processing die is not connected to any of the plurality of switching die; the second network processing die, Specifically used for: after receiving the first message through the external port and sending the second message through the internal port, sending the second message to the third network processing die through the internal port, and the third network processing die Connected to both one switching die and the second network processing die among the plurality of switching die; the third network processing die is used for: receiving the second message and forwarding the second message to the switching die .
  • the second network processing die when the second network processing die is not connected to any switching die, the second message can be forwarded to a certain switching die through the third network processing die, thereby being consistent with the existing technology. Compared with this, there is no need to set up corresponding wiring on the backplane, which can shorten the transmission path, thereby reducing the delay and power consumption of transmitting the message and the destination information.
  • the second network processing die is specifically used for: transmitting data to the third network processing die through the internal port.
  • the third network processing die sends the second message
  • the second message is sent to one of the at least two third network processing die.
  • the second switching die can flexibly Selecting a third network processing chip for forwarding the second message can avoid the problem of long transmission delay of the message due to congestion on certain transmission paths.
  • the second network processing die further includes a second selection switch.
  • the second selection switch includes a connection end and at least two selection ends.
  • the connection end of the second selection switch is connected to the external The ports are connected, and the at least two selection terminals of the second selection switch are respectively connected to the at least two third network processing chips; the second network processing chip is also specifically used to: connect the connection terminal of the second selection switch and One of the at least two selection terminals selects one of the at least two third network processing dies to send the second message.
  • the second network processing die is further configured to: when the destination network processing die is the second network processing die, send the first message through the external port.
  • the second network processing die receives the first message through the external port, if the second network processing die is the destination network processing die of the first message, it sends the message through the external port. first message, thereby reducing the transmission delay of the first message.
  • the communication chip includes a plurality of groups, any one of the plurality of groups includes at least one network processing die among the plurality of network processing die, and the plurality of network processing die Network processing dies in any two groups in the group are not connected; wherein, at least one network processing die in the same group is connected to one of the plurality of switching dies through the internal port, or through the internal port.
  • the ports connect adjacent network processing dies in the same group. That is, when the multiple network processing dies belong to multiple groups, the network processing dies in the multiple groups are not connected across groups.
  • the wiring in the communication chip can be reduced, the transmission routing in the communication chip can also be simplified, the processing efficiency can be improved, and the power of the communication chip can be reduced. consumption, volume and cost.
  • the plurality of switching dies and the plurality of network processing dies are located in the same wafer. That is, there may be only one wafer in the communication chip.
  • the unit includes the plurality of switching dies and the plurality of network processing dies.
  • the communication chip further includes: a plurality of storage dies, any one of the network processing dies and one of the plurality of storage dies. Or connect multiple memory dies.
  • a network processing die is coupled to one or more storage die, so that during the data exchange process, the storage die can be used to cache the corresponding exchange messages, thus improving the efficiency to a certain extent. Data exchange capabilities of communication chips.
  • a data exchange device in a second aspect, includes a backplane and at least one communication chip provided on the backplane.
  • the communication chip is any one of the above-mentioned first aspect or the first possible aspect.
  • the communication chip provided by the implementation method.
  • the data switching device includes a switching box or a switching frame.
  • the switching frame when the data switching device includes a switching frame, the switching frame includes at least one line card corresponding to the at least one communication chip, and the at least one communication chip passes through the at least one communication chip. Line cards are mounted on the backplane.
  • a switching device in a third aspect, includes the above second aspect or any one of the second aspects.
  • the data exchange device provided by the functional implementation method.
  • Figure 1 is a schematic structural diagram of a router provided by the prior art
  • Figure 2 is an application schematic diagram of a data switching network provided by an embodiment of the present application.
  • Figure 3 is a schematic structural diagram of a data switching network provided by an embodiment of the present application.
  • Figure 4 is a schematic structural diagram of a communication chip provided by an embodiment of the present application.
  • Figure 5 is a schematic structural diagram of another communication chip provided by an embodiment of the present application.
  • Figure 6 is a schematic structural diagram of another communication chip provided by an embodiment of the present application.
  • Figure 7 is a schematic structural diagram of another communication chip provided by an embodiment of the present application.
  • Figure 8 is a schematic diagram of data exchange provided by an embodiment of the present application.
  • FIG. 9 is a schematic diagram of another data exchange provided by an embodiment of the present application.
  • FIG. 10 is a schematic diagram of another data exchange provided by the embodiment of the present application.
  • Figure 11 is a schematic structural diagram of another communication chip provided by an embodiment of the present application.
  • At least one of a, b or c can mean: a, b, c, a and b, a and c, b and c or a, b and c, where a, b and c can It can be single or multiple.
  • words such as “first” and “second” do not limit the number and order.
  • the technical solutions provided by the embodiments of this application can be applied to a variety of different data exchange networks.
  • the data exchange network can also be called a data exchange system. It can be used to implement data exchange between different devices in the same data network. It can also be used to implement Data exchange across different data networks.
  • the data switching network may be a switch, a router, or a switching network composed of multiple switches or routers.
  • the different data networks may include local area networks and the Internet.
  • the local area network and the Internet may exchange data through routers.
  • the local area network may include terminal devices such as mobile phones, tablet computers, and personal computers.
  • the Internet It can include one or more servers. The following is an example of the structure of the data switching network.
  • FIG. 3 is a schematic structural diagram of an m ⁇ n (both m and n are integers greater than 1) data switching network provided by an embodiment of the present application.
  • the data exchange network includes m source nodes (source, S) and n destination nodes (destination, D), and at least one switch node (switch, SW) located between the m source nodes S and the n destination nodes. ).
  • the m source nodes S are represented as S1 to Sm
  • the n destination nodes D are represented as D1 to Dn
  • the at least one switching node SW includes multiple switching nodes and is represented as SW1 to SWi as an example. illustrate.
  • each of the m source nodes S and the n destination nodes D may include one or more ports.
  • the one or more ports are input ports, which can be used to receive data packets or cells from outside the switching system and send them to the at least one switching node SW.
  • the at least one switching node SW can be used to perform switching according to the destination information of the received data packets or cells (for example, destination IP address, destination port or destination node information), and send the switched data packets or cells to The corresponding destination node D.
  • the one or more ports are output ports, which can be used to send completed exchanged data packets or cells to the outside of the switching system.
  • the data switching network can be used to implement the exchange of variable-length packets (variable-length packets) or the exchange of fixed-length cells (cells).
  • the source node S when the above-mentioned source node S receives the data packet, it can keep the original format of the variable-length data packet and send it, or it can divide the data packet into cells and send it; if the source node S is sent according to the The data packet is sent in the form of cells, and the destination node D can reassemble the complete data packet after receiving all the cells.
  • the source node S and the destination node D are two independent nodes. In practical applications, the source node S and the destination node D can also be the same node and have inputs at the same time. ports and output ports, the above-mentioned Figure 3 does not limit the embodiment of the present application.
  • the source node S, the destination node D and the switching node SW may be dies.
  • the source node S and the destination node D may be network processing (network processing, NP) dies (also may be called an NP grain), and the switching node SW may be a switching grain (which may also be called a SW grain).
  • NP network processing
  • the above-mentioned data exchange network may be a communication chip (chip) including multiple NP dies and one or more SW dies, or a chipset obtained by merging and packaging multiple such communication chips together. The structure of the communication chip is described in detail below.
  • FIG. 4 is a schematic structural diagram of a communication chip provided by an embodiment of the present application.
  • the communication chip includes a plurality of SW die 10 and a plurality of NP die 20. Any SW die in the plurality of SW die 10 is connected to other SW die, and any one of the plurality of NP die 20 is connected to other SW die.
  • An NP die includes an internal port and an external port. The internal port is used to connect to other NP die, or to both other NP die and a SW die. The external port can be used to receive messages from the outside, or Send messages to the outside.
  • the plurality of SW die 10 includes two SW die and are respectively represented as 11 to 12
  • the plurality of NP die 20 includes four NP grains and are respectively represented as 21 to 24 as an example.
  • the above-mentioned crystal grains may also be called bare crystals or bare wafers, so the SW crystal grains may also be called SW bare crystals or SW bare wafers, and the NP crystal grains may also be called NP bare crystals or NP bare wafers.
  • the plurality of SW die 10 and the plurality of NP die 20 may be multiple independent die cut from a wafer, or may be multiple uncut die located on the same wafer. of grains.
  • the NP die when the NP die is used to receive a certain message from the outside through the external port, the NP die can be called the source of the message.
  • the NP die when the NP die is used to output to the outside through an external port, the NP die can be called the destination NP die of the message.
  • the source NP grain and the destination NP grain of the same message can be the same NP grain, or they can be different NP grains.
  • the die is NP die 21 as an example), which is used to: receive the first message from outside the communication chip through the external port, obtain the destination information of the first message, and send the second message through the internal port.
  • the message includes the first message and the destination information, and the destination information is used to indicate the destination network processing chip of the first message.
  • the NP chip 21 includes an external port (which can also be called an input port), which can be used to receive a first message from outside the communication chip through the external port and parse the destination address or purpose of the first message. port to determine the destination information based on the destination address or destination port.
  • the destination information can be the identification of the destination NP crystal grain corresponding to the destination address or destination port; the NP crystal grain 21 can also send the first packet including the first packet through the internal port. message and the second message of the destination information.
  • each NP die in the plurality of NP die 20 can store a preset routing table corresponding to the communication chip, and the preset routing table can be used to indicate any two of the plurality of NP die 20. Routing information between NP grains, so that when any NP grain parses and obtains the destination address or destination port of the first message, it can determine and send the second message based on the corresponding routing information.
  • Any SW die among the plurality of SW die 10 is configured to: receive the second message, and when it is determined according to the destination information that the preset condition is not met, send a message to the first NP among the plurality of NP die 20
  • the die (for example, NP die 22 or NP die 23) sends the second message, or when it is determined that the preset condition is met, sends the second message to the connected SW die.
  • the preset conditions include: the SW die among the plurality of SW die 10 is connected to the target NP die, or the SW is connected to the target NP die through other NP die and has the smallest number of connected NP die. grains.
  • each SW die in the plurality of SW die 10 may also store a preset routing table corresponding to the communication chip, so that any SW die can determine whether it satisfies the preset routing table based on the preset routing table. Preset conditions and send the second message.
  • the plurality of SW die 10 include a first SW die 11 and a second SW die 12 , and the first SW die 11 and the second SW die 12 are connected.
  • the first SW die 11 is used to: receive the second message, and determine according to the destination information that the first SW die 11 does not meet the preset condition, and send the second message to the second SW die 12 ;
  • the second SW die 12 is used to: receive the second message from the first SW die 11, and determine according to the destination information that when the second SW die 12 meets the preset condition, send the message to the first NP die 22 Send the second message.
  • each SW die in the plurality of SW die 10 may include a plurality of switching ports, and the plurality of switching ports may be connected to at least one NP die, or to at least one NP die and to another one or Multiple SW dies are connected, and each SW die can switch packets received by one switch port to another switch port for output.
  • the first SW die 11 can perform the following steps: receive the second message from the NP die 21 through the first switching port; determine based on the destination information that the SW die 11 satisfies the preset condition (for example, the NP die 22 is the destination NP die.
  • the second message is switched from the first switch port to the second switch port to send the second message to the NP die 22 through the second switch port.
  • the second message is switched from the first switch port to the third switch port, so as to send the second message to the second SW die 12 through the third switch port, so that the second The SW die 12 sends a second message to the NP die 23 when it is determined that the preset condition is met.
  • the first NP die (hereinafter taking NP die 23 as an example) is used to receive the second message from the connected SW die and send the second message to the outside through the external port, or to send the second message to the outside through the internal port.
  • the second message is sent to the NP die.
  • the first NP die 23 receives the second message from the switching die 12, if the destination NP die of the first message is the first NP die 23, the first NP die 23 passes The external port sends the second message; if the first The destination NP die of the message is not the first NP die 23.
  • the first NP die 23 sends the second message to other NP die through the internal port.
  • the destination NP die of the first message is the NP die.
  • the first NP die 23 sends the second message to the NP die 24 .
  • the destination NP die of the first message is not the first NP die 23, the destination NP die of the first message is different from the SW die 10 in the first message.
  • Any SW grain is not connected, and there are at least two first NP grains 23 in the plurality of NP grains 20 (taking two NP grains as an example and represented as 23a and 23b), both of which are connected to the second SW grain.
  • the second SW die 12 is connected to the target NP die, that is, there are multiple transmission paths between the second SW die 12 and the target NP die, then the second SW die 12 is specifically used to: transfer to the first NP die 22
  • the second message may be sent to one of at least two first NP dies 23 .
  • the second SW die 12 can send a second message to the first NP die 23a, or the second SW die 12 can send a second message to the first NP die 23b. arts.
  • the second SW die 12 can select one transmission path from the multiple transmission paths according to a certain flow control strategy. Transmit the second message to the destination NP die.
  • the traffic control policy may include one data stream being transmitted through one transmission path, or one data stream being transmitted through multiple transmission paths in a load balancing manner, etc. This embodiment of the present application does not specifically limit this.
  • the second NP die 21 i.e., the source NP grain
  • the second NP die 21 is specifically used to: after receiving the first message through the external port and sending the second message through the internal port, send the third NP die 22 through the internal port.
  • the third NP die 22 is connected to both one of the plurality of SW die 20 (for example, the first SW die 11) and the second NP die 21.
  • the third NP die 22 is used to receive the second message and forward the second message to the first SW die 11 .
  • the second NP crystal grain 21 is specifically used for:
  • the second message is sent to one of at least two third NP die 22 .
  • the second NP die 21 sends the second message to the third NP die 22a, or the second NP die 21 sends the second message to the third NP die 22b.
  • the second NP die 21 can select one transmission path from the multiple transmission paths to the first SW die according to a certain flow control strategy.
  • SW die 11 transmits the second message.
  • the traffic control policy may include one data stream being transmitted through one transmission path, or one data stream being transmitted through multiple transmission paths in a load balancing manner, etc. This embodiment of the present application does not specifically limit this.
  • the above-mentioned second SW die 12 may satisfy the preset condition, or may not satisfy the preset condition.
  • the second SW die 12 may send a second message to the connected first NP die 23 .
  • the second SW die 12 does not meet the preset condition, the second SW die 12 can send a second message to other SW die connected to the second SW die 12, so that the other SW die 12 connects to the second SW die 12.
  • the first NP die 23 sends the second message.
  • the plurality of SW die 10 also includes a third SW die 13 , and the third SW die 13 is connected to the second SW die 12 respectively. is connected to the first NP crystal grain 23, then the second SW crystal
  • the particle 12 is also used to send a second message to the third SW die 13 when it is determined that the second SW die 12 does not meet the preset condition.
  • the third SW die 13 is used to: receive the second message from the second SW die 12, and determine according to the destination information in the second message that the third SW die 13 meets the preset condition, and send the message to the third SW die 13.
  • the first NP die 23 sends the second message.
  • the first SW die 11 is specifically used for:
  • the second message may be sent to one of at least two second SW die 12 .
  • the first SW die 11 sends a second message to the second SW die 12a, or the first SW die 11 sends a second message to the second SW die 12b.
  • the first SW die 11 can select one transmission path from the multiple transmission paths according to a certain flow control strategy.
  • the path transmits the second message to the third SW die 23 .
  • the traffic control policy may include one data stream being transmitted through one transmission path, or one data stream being transmitted through multiple transmission paths in a load balancing manner, etc. This embodiment of the present application does not specifically limit this.
  • the first NP die 23 can not only be used to send the second message to the outside through the external port, but can also be used to send the second message to other NP die through the internal port.
  • the SW die connected to the first NP die 23 (or called the SW die that sends the second message to the first NP die 23 ) is also used to: send path indication information to the first NP die 23, where the path indication information is used to indicate the sending path of the second message.
  • the first NP chip 23 is also configured to receive the path indication information and send the second packet to the outside through the external port according to the path indication information, or to send the second packet to the outside through the internal port according to the path indication information.
  • the text is sent to other NP grains.
  • the path indication information can also be used to indicate that the second message is sent in the transmission path.
  • the sending paths in multiple NP die that is, the multiple NP die can forward the second message according to the path indication information until the second message is forwarded to the destination NP die.
  • the second message sent by the NP die 21 reaches the second SW die through the first SW die 11.
  • the second SW die 12 determines that the preset condition is met, and the second SW die 12 can send the second message and the path indication information to the first NP die 23 .
  • the path indication information can be used to instruct the first NP die 23 to send the second message to the outside through the external port, then the first NP die 23 receives the second message.
  • the second message is sent to the outside through the external port.
  • the path indication information can be used to instruct the first NP die 23 to forward the second message to the NP die 24 through the internal port, then the first NP die 23 receives the When receiving the second message and the path indication information, the second message is forwarded to the NP die 24 according to the path indication information.
  • the second message sent by the NP die 21 passes through the first SW die 11 and the second The SW die 12 reaches the third SW die 13.
  • the third SW die 13 determines that the preset condition is met, and the third SW die 13 can send the second message and path indication information to the first NP die 23a. If the destination NP die is the first NP die 23a, the path indication information can be used to instruct the first NP die 23a to send the second message to the outside through the external port, then the first NP die 23a receives the second message. When receiving the message and the path indication information, the second message is sent to the outside through the external port.
  • the path indication information can be used to instruct the first NP die 23a to forward the second message to the NP die 24 through the internal port, then the first NP die 23a receives the first When receiving the second message and the path indication information, the second message is forwarded to the NP die 24 according to the path indication information.
  • the communication chip shown in FIG. 7 may include four SW dies, denoted as 11, 12a, 12b, and 13 respectively, and 12 NP dies denoted as 21, 22a, 22b, 23a, 23b, 24, respectively. 25a, 25b, 26, 27a, 27b and 28.
  • SW crystal grain 11 and SW crystal grain 13 are both connected to SW crystal grain 12a and SW crystal grain 12b
  • NP crystal grain 21 is connected to SW crystal grain 11 through NP crystal grains 22a and 22b respectively
  • NP crystal grain 24 is respectively connected to SW crystal grain 11 through NP crystal grains 22a and 22b.
  • Dies 23a and 23b are connected to SW die 13
  • NP die 26 is connected to SW die 12a through NP die 25a and 25b, respectively
  • NP die 28 is connected to SW die 12b through NP die 27a and 27b, respectively.
  • the specific structure of the first NP crystal grain 23 and the operations performed by the first NP crystal grain 23 according to the path indication information are illustrated with examples.
  • the first NP die 23 a includes a first selection switch 231 .
  • the first selection switch 231 includes a connection terminal Q0, a first selection terminal Q1 and a second selection terminal Q2.
  • the connection terminal Q0 of the first selection switch 231 is connected to the third SW die 13.
  • the selection terminal Q1 and the second selection terminal Q2 are respectively connected to the external port of the first NP die 23a and the NP die 24.
  • the path indication information is used to indicate When the first NP chip 23a sends the second message to the outside through the external port, the first NP chip 23a can conduct the connection terminal Q0 of the first selection switch 231 and the first selection terminal Q1 of the first selection switch 231. If the destination NP die is NP die 24, the path indication information is used to instruct the first NP die 23a to forward the second message to the NP die 24 through the internal port, the first NP die 23a can cause Connect the connection terminal Q0 of the first selection switch 231 and the second selection terminal Q2 of the first selection switch 231 .
  • the first selection switch 231 may also include at least two second selection terminals Q2, and the at least two second selection terminals Q2 are respectively connected to at least two internal ports of the first NP die 23a.
  • the internal ports are respectively connected to at least two of the plurality of NP dies 20 .
  • the first NP die 23 is also used to connect the connection end of the first selection switch 221 with the path indication information when forwarding the second message to other NP die through the internal port according to the path indication information.
  • the second selection terminal Q2 corresponding to the indicated transmission path.
  • the specific structure of the second NP crystal grain 21 and the operation performed by the second NP crystal grain 21 to select the transmission path are illustrated with examples.
  • the second NP die 21 includes a second selection switch 211 .
  • the second selection switch 211 includes a connection terminal P0 and at least two selection terminals P1 to Pi (i is a positive integer greater than 1).
  • the connection terminal P0 of the second selection switch 211 is connected to the external port of the NP die 21.
  • At least two selection terminals P1 to Pi of the second selection switch 211 are respectively connected to at least two NP dies.
  • the at least two NP grains include two NP grains and are represented as 22a and 22b respectively.
  • the at least two selection terminals P1 to Pi include two selection terminals and are represented as P1 and P2 respectively.
  • the NP grains 22a is connected to the selection terminal P1, and the NP die 22b is connected to the selection terminal P2 for explanation.
  • the second NP chip 21 is also used to control the connection terminal P0 of the second selection switch 211 and one of the at least two selection terminals P1 to Pi of the second selection switch 211 to achieve the above-mentioned Select one NP die from at least two NP die to send the second message.
  • the NP die 21 can be used to control the connection terminal P0 of the second selection switch 211 and the selection terminal P1 of the second selection switch 211 to conduct, so as to obtain from the NP die 22 a and the NP die 22 b
  • the NP die 22a is selected to send the second message, that is, the NP die 21 chooses to forward the second message to the first SW die 11 through the NP die 22a.
  • the NP die 21 and the first NP die 23a may also include other functional units.
  • the NP die 21 and the first NP die 23a may also include a media access control (media access control) aggregation unit (aggregation), a network processor (network processor, NP) and a traffic manager (traffic manager) , TM), the media access control aggregation unit can be referred to as MAG.
  • MAG can be used to sense optical signals or electrical signals, forward the optical signals or electrical signals into data frames (ie, messages), and perform legality checks on the data frames.
  • NP's upstream processing can include packet parsing, data flow analysis, packet forwarding, and mapping MAC/IP addresses to TM queue numbers, etc.;
  • NP's downstream processing can include packet encapsulation and egress processing.
  • TM can be used to provide functions such as large-capacity memory, congestion avoidance, traffic shaping, multi-level traffic scheduling, and traffic congestion control. It is also used to map the TM queue number to the identification of the destination NP die.
  • any NP crystal among the plurality of NP crystals shown in FIG. 7 may include a media access control aggregation unit MAG, a network processor NP, and a traffic manager TM.
  • the above-mentioned NP may also include a selection switch similar to the first selection switch 231 or the second selection switch 211, which is not specifically limited in the embodiment of the present application.
  • the plurality of NP grains 20 may also include a larger number of SW grains and NP grains. Any SW grain Each grain can be connected to other SW grains. Any NP grain can be connected to other NP grains, or to other NP grains and SW grains at the same time. Each connected NP grain includes a selection switch. , MAG, NP and TM.
  • any one of the plurality of NP crystals 10 may perform the function of any one of the above-mentioned NP crystals 21 to 24, depending on the messages that need to be exchanged. It is related to the position (or connection relationship) of the NP grain in the communication chip.
  • the NP grain when the above NP grain includes multiple different functional units such as a selection switch, a media access control aggregation unit MAG, a network processor NP, and a traffic manager TM, the NP grain can also be split into multiple Different grains, each grain is used to correspond to one or more functional units mentioned above.
  • each NP grain may include two grains, and the two grains include grains corresponding to MAG, and grains corresponding to NP, TM, and the selection switch.
  • the structures of the plurality of SW crystal grains 10 in the embodiment of the present application may be the same or different; similarly, the structures of the plurality of NP crystal grains 20 may be the same or different.
  • the exact same plurality of die can be obtained through one tape-out or the same tape-out process, by configuring By rotating the plurality of crystal grains, SW crystal grains or NP crystal grains at different positions in the communication chip can be obtained, which can greatly reduce the development cost of SW crystal grains and NP crystal grains.
  • the source NP die can also be used to send the first message through the external port.
  • the second NP die 21 can also be used to: receive When the first message is received and the destination NP die is determined to be the second NP die 21, the first message is sent through the external port.
  • the source NP die and the destination NP die of the first message are two connected NP die
  • the source NP die can also be used to send the first message to the destination NP die through the internal port.
  • the second NP crystal 21 and the NP crystal 22a in Figure 5 or 6 as an example, the second NP crystal 21 is also used to: after receiving the first message and determine the destination NP crystal When the die is the NP die 22a, the first message is sent to the NP die 22a through the internal port.
  • the communication chip may also include a plurality of groups, and any one of the plurality of groups includes at least one NP die among the plurality of NP die 20 , and/or one of the plurality of SW die 10 . At least one SW die.
  • any one of the plurality of groups includes at least one SW die among the plurality of SW die 10 , or any one group includes at least one of the plurality of SW die 10
  • One SW die and at least one NP die Any NP die in the at least one NP die is connected to a SW die in the same group through an internal port, or to an adjacent NP in the same group through an internal port. The grains are connected.
  • the communication chip may include 4 groups.
  • the first group may include SW die 11, NP die 21, NP die 22a and NP die 22b.
  • the second group may include SW die 12a, NP die 26, NP die 25a and NP die 25b
  • the third group may include SW die 12b, NP die 28, NP die 27a and NP die.
  • the fourth group may include SW grains 13, NP grains 24, NP grains 23a and NP grains 23b.
  • the two adjacent NP crystal grains may not be connected.
  • the NP crystal grain 22a and the NP crystal grain 25a may not be connected
  • the NP crystal grain 22b and the NP crystal grain 27b may not be connected
  • the NP crystal grain 25b and the NP crystal grain may not be connected
  • 23b may not be connected
  • the NP crystal grain 27a and the NP crystal grain 23a may not be connected.
  • the two adjacent NP crystal grains can also be connected. In the above-mentioned FIG. 7, only in different groups Two adjacent NP grains are not connected as an example for illustration.
  • the message received by the source NP die can pass through the NP die and/or SW die in the group. Transmit to the destination NP die; when the source NP die and the destination NP die belong to different groups, the message received by the source NP die can be transmitted through the NP die and/or SW die in multiple groups. to the target NP grain.
  • the following takes the communication chip shown in Figures 8-10 as an example to illustrate the switching paths corresponding to messages of NP grains at different positions in the communication chip.
  • the different positions may include the source NP grain and the destination NP grain belonging to the same group or different groups.
  • the source NP grain is the NP grain directly connected to the SW grain (referred to as the directly connected NP grain).
  • the source NP grain is NP grains that are not directly connected to SW grains (referred to as indirect NP grains), the target NP grains are directly connected NP grains and indirect NP grains.
  • the communication chip includes 4 SW dies (denoted as SW1 to SW4) and 12 NP dies (denoted as NP1-NP12 respectively).
  • SW1 to SW4 and NP1-NP12 are divided into four Grouping, each group includes one SW die, two directly connected NP die and one non-directly connected NP die as an example for illustration.
  • both the source NP grain and the destination NP grain belong to the same group, and the source NP grain is a directly connected NP grain (for example, the source NP grain is An example of possible switching paths is provided when SW1 is directly connected to NP5).
  • the destination NP die is a directly connected NP die, such as NP5
  • the message corresponds to The switching path S11 is NP5-SW1-NP5, such as NP2, then the switching path S12 corresponding to the message is NP5-SW1-NP2; if the destination NP die is an indirectly connected NP die, such as NP1, then the corresponding The switching path S13a is NP5-SW1-NP2-NP1, or the switching path S13b is NP5-SW1-NP5-NP1.
  • the source NP die and the destination NP die belong to the same group, and the source NP die is an indirectly connected NP die (for example, the source NP die is NP9 directly connected to SW3).
  • the switching path S21a of the message is NP9-NP7-SW3-NP7-NP9, or the switching path S21b is NP9-NP10-SW3-NP10- NP9, or the switching path S21c is NP9-NP7-SW3-NP10-NP9 or NP9-NP10-SW3-NP7-NP9; if the destination NP die is directly connected to the NP die, such as NP7, then the switching path S22a of the message is NP9-NP7-SW3-NP7, or switching path S22b is NP9-NP10-SW3-NP7.
  • the source NP grain and the destination NP grain belong to different groups, and the source NP grain is a directly connected NP grain (for example, the source NP grain is SW1 Possible switching paths are given as examples for directly connected NP5).
  • the message exchange path S31 is NP5-SW1- SW2-NP3; if the destination NP die is an indirectly connected NP die, such as NP4, the switching path S32 of the message is NP5-SW1-SW2-NP6/NP3-NP4.
  • the switching path S41 of the message is NP5-SW1-SW2/ SW3-SW4-NP8; if the destination NP die is an indirectly connected NP die, such as NP12, the switching path S42 of the message is NP5-SW1-SW2/SW3-SW4-NP8/NP11-NP12.
  • the source NP grain and the destination NP grain belong to different groups, and the source NP grain is an indirectly connected NP grain (for example, the source NP grain is Possible switching paths in NP1) are given as examples. Specifically, taking the two groups to which the source NP die and the destination NP die belong to are adjacent as an example, if the destination NP die is directly connected to the NP die, such as NP3, then the exchange path S51 of the message is NP1-NP5/ NP2-SW1-SW2-NP3; if the destination NP die is an indirectly connected NP die, such as NP4, the switching path S52 of the message is NP1-NP5/NP2-SW1-SW2-NP6/NP3-NP4.
  • the switching path S61 of the message is NP1-NP5/NP2- SW1-SW2/SW3-SW4-NP8; if the destination NP die is an indirectly connected NP die, such as NP12, the message switching path S62 is NP1-NP5/NP2-SW1-SW2/SW3-SW4-NP8/ NP11-NP12.
  • the corresponding switching path needs to go through SW Crystal grains are taken as an example to illustrate.
  • the corresponding switching path can also directly pass through the SW grain without passing through the SW grain.
  • NP grain output For example, assuming that both the source NP die and the destination NP die are NP4, NP4 can pass the message through the corresponding external port after receiving the message and determining that the destination NP die of the message is NP4. output.
  • the communication chip may also include a larger number of NP grains and a larger number of SW grains.
  • the plurality of NP crystal grains can be used to receive external messages or send messages to the outside; a part of the multiple NP crystal grains can also be used to forward messages for another part of the NP crystal grains, so as to Forward the message received by the source NP die to the multiple SW die; some of the NP die among the multiple NP die can also be used to forward the message to the multiple SW die. Forwarded to the destination NP grain.
  • Each SW die in the multiple SW die can be used to exchange messages, and different SW die It can also be used to forward messages between them.
  • the communication chip includes a larger number of NP grains and a larger number of SW grains, the type and number of grains included in each group of the plurality of groups of the communication chip may be the same or different. .
  • the communication chip may include 9 SW dies (denoted as SW1 to SW9) and 16 NP dies (denoted as NP1 to NP16). Each SW die The grains can be connected to at least one NP grain, and each SW grain can also be connected to at least one SW grain. Each NP grain can be connected to one SW grain, or to at least one NP grain, or to both. One SW grain is connected to at least one NP grain.
  • the communication chip may include 9 groups, where each of the 4 groups may include three NP dies and one SW die, and the other 4 Each of the groups may include one NP die and one SW die, and the last group may include one SW die.
  • the communication chip may include 5 groups, where each of the 4 groups may include four NP dies and one SW die. , the last group can include one SW die.
  • the multiple groups shown in FIG. 11 are only exemplary. In practical applications, the multiple groups can also be divided in other ways, and this embodiment of the present application does not specifically limit this.
  • (a) in Figure 11 takes the example of two adjacent NP grains that are not connected when they belong to different groups; (b) in Figure 11 takes the example of two adjacent NP grains that are not connected. When two NP grains belong to different groups, the two adjacent NP grains are connected as an example for explanation.
  • the communication chip may further include: at least one memory die, and any one NP die among the plurality of NP die is connected to one or more memory die among the at least one memory die.
  • the communication chip includes a plurality of memory die, and the plurality of memory die corresponds to the plurality of NP die in a one-to-one correspondence, that is, each NP die is connected to a corresponding memory die.
  • the storage medium type of the storage chip includes but is not limited to: dynamic random access memory (dynamic random access memory, DRAM), static random access memory (static random access memory, SRAM), ferroelectric random access memory Memory (ferroelectric random access memory, FeRAM) or magnetic random access memory (magnetic random access memory, MRAM), etc.
  • the storage medium types of the multiple storage die may be the same or different; when an NP die is coupled with multiple storage die, the NP die
  • the storage medium types of the plurality of correspondingly coupled storage dies may be the same or different, and the embodiments of the present application do not specifically limit this.
  • the communication chip includes multiple SW die and multiple NP die.
  • the multiple SW die and the multiple NP die can be connected in the communication chip, and the multiple NP die can be used.
  • the multiple SW dies can be used to realize distributed exchange of messages, so that data exchange can be realized within the communication chip, which is similar to the existing technology through multiple NPs.
  • the chip can greatly reduce power consumption and volume, and at the same time reduce the heat dissipation cost and deployment space of equipment using this communication chip to meet the current requirements for green, low-carbon and sustainable development. need.
  • the data exchange device includes a backplane and at least one communication chip disposed on the backplane.
  • the communication chip is any of the communication chips mentioned above.
  • the data switching device may be a switching box; alternatively, the data switching device may be a switching frame.
  • the data switching device when the data switching device is a switching box, the data switching device may be in the form of a single device and does not support capacity expansion, so that no backplane interface may be provided on the backplane.
  • the data switching device is a switching frame, the data switching device may not be in the form of a single device and support capacity expansion, so that corresponding backplane interfaces can be set on the backplane.
  • the switching frame when the data switching device is a switching frame, the switching frame includes at least one communication chip Corresponding to at least one line card, the at least one communication chip is respectively connected to the backplane interface on the backplane through the at least one line card, so that the at least one communication chip is disposed on the backplane.
  • embodiments of the present application also provide a switching device, which may be a router or a switch.
  • the switching device includes the data switching device provided above.

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Data Exchanges In Wide-Area Networks (AREA)

Abstract

一种通信芯片及数据交换装置，涉及通信技术领域，用于降低数据交换网络的功耗、成本和体积。该通信芯片包括：多个交换晶粒、以及与多个交换晶粒直接或间接连接的多个网络处理晶粒；任一网络处理晶粒，用于：通过外部端口接收第一报文，并通过内部端口发送第二报文，第二报文包括第一报文和用于指示第一报文的目的网络处理晶粒的目的信息；任一交换晶粒，用于：接收第二报文，并根据目的信息确定不满足预设条件时，向相连的交换晶粒发送第二报文，或者在确定满足所述预设条件时，向相连的第一网络处理晶粒发送第二报文；第一网络处理晶粒，用于：接收第二报文，并向外部发送第二报文，或者将第二报文发送给其他网络处理晶粒。

Description

一种通信芯片及数据交换装置
本申请要求于2022年07月18日提交国家知识产权局、申请号为202210843102.4、申请名称为“一种通信芯片及数据交换装置”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
技术领域
本申请涉及通信技术领域，尤其涉及一种通信芯片及数据交换装置。
背景技术
路由器(router)是连接两个或多个网络的硬件设备，在网络间起网关的作用，可以读取数据包中的目的地址并决定如何传送的一种网络设备。路由器可以识别不同的网络协议，并对不同网络协议的数据包的目的地址进行转换，再根据相应的路由算法把数据包按最佳路线传送到指定位置。
目前，如图1所示，路由器包括多个网络处理(network processing，NP)芯片(chip)和多个交换(switching，SW)芯片，该多个NP芯片分别设置在多个线卡(line card，LC)上，该多个交换芯片设置在背板(backplane，BP)，该背板上还设置有板级串行(serdes)接口，该多个线卡通过该板级串行接口与该背板上的多个交换芯片连接。其中，该NP芯片可用于接收来自外部网络的数据并将该数据转发给交换芯片，或者接收交换芯片输出的数据并将该数据转发给外部网络。该交换芯片用于接收NP芯片转发的数据，并进行数据交换后将该数据发送给相应的NP芯片。
但是，通过板级串行接口连接上述多个线卡与背板上的多个交换芯片的方式，会造成功耗大、成本高、以及路由器的体积大的问题。
发明内容
本申请提供一种通信芯片及数据交换装置，用于降低数据交换网络的功耗、成本和体积。
为达到上述目的，本申请采用如下技术方案：
第一方面，提供一种通信芯片，该通信芯片包括：多个交换晶粒和多个网络处理晶粒，该多个交换晶粒中的任一交换晶粒与其它交换晶粒相连以实现分布式交换，该多个网络处理晶粒中的任一网络处理晶粒包括外部端口和内部端口，该内部端口与其他网络处理晶粒相连、或者与其它网络处理晶粒和该多个交换晶粒中的一个交换晶粒均相连。其中，该多个网络处理晶粒中的任一网络处理晶粒，用于：通过该外部端口接收来自该通信芯片外部的第一报文，获取第一报文的目的信息，通过该内部端口发送第二报文，第二报文包括第一报文和该目的信息，该目的信息用于指示第一报文的目的网络处理晶粒，比如，该目的信息为目的地址或目的网络处理晶粒的标识。该多个交换晶粒中的任一交换晶粒，用于：接收第二报文，并根据该目的信息确定不满足预设条件时，向相连的交换晶粒发送第二报文，或者在确定满足该预设条件时，向相连的第一网络处理晶粒发送第二报文。第一网络处理晶粒，用于：接收来自相连的交换晶粒的第二报文，并通过该外部端口向外部发送第二报文，或者，通过该内部端口将第二报文发送给其他网络处理晶粒；其中，该预设条件 包括：是该多个交换晶粒中与该目的网络处理晶粒相连的交换晶粒、或者是通过其它网络处理晶粒与该目的网络处理晶粒相连且连接的网络处理晶粒的数量最少的交换晶粒。
上述技术方案中，该通信芯片包括多个交换晶粒和多个网络处理晶粒，该多个交换晶粒和该多个网络处理晶粒在该通信芯片内可实现连接，且该多个网络处理晶粒可用于接收外部的报文或者向外部发送报文，该多个交换晶粒可用于实现报文的分布式交换，从而实现在通信芯片内部的数据交换，与现有技术中通过多个网络处理芯片和多个交换芯片组成的数据交换网络相比，能够大大降低功耗和体积，同时减小采用该通信芯片的设备的散热成本和部署空间，以满足当前对于绿色低碳和可持续发展的需求。
在第一方面的一种可能的实现方式中，该多个交换晶粒包括第一交换晶粒和第二交换晶粒，第二交换晶粒与第一交换晶粒相连，第一交换晶粒与源网络处理晶粒最近，第二交换晶粒与目的网络处理晶粒最近；第一交换晶粒，用于：接收第二报文，并根据该目的信息确定第一交换晶粒不满足该预设条件时，向第二交换晶粒发送第二报文；第二交换晶粒，用于：接收来自第一交换晶粒的第二报文，并根据该目的信息确定第二交换晶粒满足该预设条件时，向第一网络处理晶粒发送第二报文。上述可能的实现方式中，当与报文的源网络处理晶粒最近的第一交换晶粒、和与报文的目的网络处理晶粒最近的第二交换晶粒不相连时，第一交换晶粒在接收到该报文后，可以发送给第二交换晶粒，以使第二交换晶粒将报文交换至目的网络处理晶粒。
在第一方面的一种可能的实现方式中，该多个交换晶粒还包括第三交换晶粒，第三交换晶粒与第二交换晶粒相连，即第一交换晶粒、第二交换晶粒和第三交换晶粒依次相连，第一交换晶粒与源网络处理晶粒最近，第三交换晶粒与目的网络处理晶粒最近；第二交换晶粒，还用于：当确定第二交换晶粒不满足该预设条件时，向第三交换晶粒发送第二报文；第三交换晶粒用于：接收来自第二交换晶粒的第二报文，并根据该目的信息确定第三交换晶粒满足该预设条件时，向第一网络处理晶粒发送第二报文。上述可能的实现方式中，当与报文的源网络处理晶粒最近的第一交换晶粒、和与报文的目的网络处理晶粒最近的第三交换晶粒不相连，且这两个交换晶粒均与第一交换晶粒相连时，第一交换晶粒在接收到该报文后可以发送给第二交换晶粒，第二交换晶粒向第三交换晶粒发送报文，以使第三交换晶粒将报文交换至目的网络处理晶粒。
在第一方面的一种可能的实现方式中，该多个交换晶粒中存在至少两个第二交换晶粒与第一交换晶粒和第三交换晶粒均相连，即第一交换晶粒与源网络处理晶粒最近，第三交换晶粒与目的网络处理晶粒最近，且第一交换晶粒与第三交换晶粒之间存在多个传输路径，每个传输路径对应一个第二交换晶粒。第一交换晶粒，具体用于：在向第二交换晶粒发送第二报文时，向该至少两个第二交换晶粒中的一个发送第二报文。上述可能的实现方式中，当第一交换晶粒与第三交换晶粒之间存在多个传输路径时，第一交换晶粒能够灵活地选择其中的一个第二交换晶粒发送第二报文，从而能够避免因为某些传输路径拥塞而导致报文的传输时延较长的问题。
在第一方面的一种可能的实现方式中，该多个交换晶粒中的任一交换晶粒，还用于：在向第一网络处理晶粒发送第二报文时，向第一网络处理晶粒发送路径指示信息，该路径指示信息用于指示第二报文的发送路径，以控制第二报文在第一网络处理晶粒中的传输；第一网络处理晶粒还用于：接收该路径指示信息，并根据该路径指示信息通过该外部端口 向外部发送第二报文，或者，根据该路径指示信息通过该内部端口将第二报文发送给其他网络处理晶粒。上述可能的实现方式中，第一网络处理晶粒在传输第二报文时，该交换晶粒可通过控制第二报文在第一网络处理晶粒中的发送路径，以控制第一网络处理晶粒将该报文转发给目的网络处理晶粒，避免报文的误交换。
在第一方面的一种可能的实现方式中，第一网络处理晶粒包括第一选择开关，第一选择开关包括连接端、第一选择端和第二选择端，第一选择开关的连接端与该多个交换晶粒中的一个相连，第一选择开关的第一选择端与该外部端口相连，第一选择开关的第二选择端与其它网络处理晶粒相连；第一网络处理晶粒还用于：在根据该路径指示信息通过该外部端口向外部发送第二报文时，导通第一选择开关的连接端与第一选择开关的第一选择端；或者，第一网络处理晶粒，还用于：在根据该路径指示信息通过该内部端口将第二报文发送给其他网络处理晶粒时，导通第一选择开关的连接端与第一选择开关的第二选择端。上述可能的实现方式中，该交换晶粒通过控制第一网络处理晶粒中第一选择开关的不同选择端，以控制第一选择开关将该报文转发给该目的网络处理晶粒，以避免报文的误交换。
在第一方面的一种可能的实现方式中，第一选择开关包括至少两个第二选择端，该至少两个第二选择端分别与第一网络处理晶粒的至少两个该内部端口相连，该至少两个内部端口分别与该多个网络处理晶粒中的至少两个相连；第一网络处理晶粒，还用于：在根据该路径指示信息通过该内部端口将第二报文发送给其他网络处理晶粒时，导通第一选择开关的连接端与该路径指示信息指示的发送路径对应的第二选择端。上述可能的实现方式中，交换晶粒通过控制第一网络处理晶粒中第一选择开关的不同选择端，以选择对应的传输路径将该报文转发给目的网络处理晶粒。
在第一方面的一种可能的实现方式中，该目的网络处理晶粒与该多个交换晶粒中的任一交换晶粒不相连，该多个网络处理晶粒中存在至少两个第一网络处理晶粒均与第二交换晶粒和该目的网络处理晶粒相连；第二交换晶粒，还用于：在向第一网络处理晶粒发送第二报文时，向该至少两个第一网络处理晶粒中的一个发送第二报文。上述可能的实现方式中，当第二交换晶粒与目的网络处理晶粒之间存在多条传输路径时，第二交换晶粒能够灵活地选择用于转发第二报文的第一网络处理晶粒，从而能够避免因为某些传输路径拥塞而导致报文的传输时延较长的问题。
在第一方面的一种可能的实现方式中，该多个网络处理晶粒中存在第二网络处理晶粒与该多个交换晶粒中的任一个均不相连；第二网络处理晶粒，具体用于：在通过该外部端口接收第一报文后并通过该内部端口发送第二报文时，通过该内部端口向第三网络处理晶粒发送第二报文，第三网络处理晶粒与该多个交换晶粒中的一个交换晶粒和第二网络处理晶粒均相连；第三网络处理晶粒，用于：接收第二报文，并向该交换晶粒转发第二报文。上述可能的实现方式中，当第二网络处理晶粒与任一交换晶粒不相连时，可以通过第三网络处理晶粒向某一交换晶粒转发第二报文，从而与现有技术相比，无需在背板上设置相应的走线，能够缩短传输路径，从而减小传输该报文和该目的信息的时延和功耗。
在第一方面的一种可能的实现方式中，该多个网络处理晶粒中存在至少两个第三网络处理晶粒；第二网络处理晶粒，具体用于：在通过该内部端口向第三网络处理晶粒发送第二报文时，向该至少两个第三网络处理晶粒中的一个发送第二报文。上述可能的实现方式中，当第二交换晶粒与某一交换晶粒之间存在多条传输路径时，第二交换晶粒能够灵活地 选择用于转发第二报文的第三网络处理晶粒，从而能够避免因为某些传输路径拥塞而导致报文的传输时延较长的问题。
在第一方面的一种可能的实现方式中，第二网络处理晶粒还包括第二选择开关，第二选择开关包括连接端和至少两个选择端，第二选择开关的连接端与该外部端口相连，第二选择开关的该至少两个选择端分别与该至少两个第三网络处理晶粒相连；第二网络处理晶粒，还具体用于：导通第二选择开关的连接端与该至少两个选择端中的一个选择端，以从该至少两个第三网络处理晶粒中选择一个发送第二报文。上述可能的实现方式中，第二网络处理晶粒在进行第二报文转发时，可以通过选通第二选择开关的不同选择端，以灵活地选择不同的第三网络处理晶粒向某一交换晶粒转发该报文，从而能够避免因为某些传输路径拥塞而导致报文的传输时延较长的问题。
在第一方面的一种可能的实现方式中，第二网络处理晶粒，还用于：在该目的网络处理晶粒为第二网络处理晶粒时，通过该外部端口发送第一报文。上述可能的实现方式中，第二网络处理晶粒在通过外部端口接收到第一报文时，若第二网络处理晶粒为第一报文的目的网络处理晶粒，则通过该外部端口发送第一报文，从而能够减小第一报文的传输时延。
在第一方面的一种可能的实现方式中，该通信芯片包括多个分组，该多个分组中的任一分组包括该多个网络处理晶粒中的至少一个网络处理晶粒，该多个分组中任意两个分组中的网络处理晶粒不相连；其中，同一分组中的至少一个网络处理晶粒通过该内部端口连接同一该多个交换晶粒中的一个交换晶粒，或通过该内部端口连接同一分组中的相邻网络处理晶粒，也即是，当该多个网络处理晶粒属于多个分组时，该多个分组中的网络处理晶粒不会跨分组相连。上述可能的实现方式中，在不影响该通信芯片的性能的前提下，能够减少该通信芯片中的布线，还可以简化该通信芯片中的传输路由、提高处理效率，进而降低该通信芯片的功耗、体积和成本。
在第一方面的一种可能的实现方式中，该多个交换晶粒和该多个网络处理晶粒位于同一晶元中，也即是，该通信芯片中可以只存在一个晶元，该晶元中包括该多个交换晶粒和该多个网络处理晶粒。上述可能的实现方式，能够提高该多个交换晶粒和该多个网络处理晶粒集成度，从而进一步减小该通信芯片的体积和功耗。
在第一方面的一种可能的实现方式中，该通信芯片还包括：多个存储晶粒，该多个网络处理晶粒中的任意一个网络处理晶粒与该多个存储晶粒中的一个或者多个存储晶粒连接。上述可能的实现方式中，一个网络处理晶粒与一个或者多个存储晶粒耦合，从而在数据交换过程中，该存储晶粒可用于缓存相应的交换报文，从而在一定程度上能够提高该通信芯片的数据交换能力。
第二方面，提供一种数据交换装置，该数据交换装置包括背板、以及设置在该背板上的至少一个通信芯片，该通信芯片为上述第一方面或第一方面的任一种可能的实现方式所提供的通信芯片。
在第二方面的一种可能的实现方式中，该数据交换装置包括交换盒子或交换框。
在第二方面的一种可能的实现方式中，当该数据交换装置包括交换框时，该交换框包括与该至少一个通信芯片对应的至少一个线卡，该至少一个通信芯片分别通过该至少一个线卡设置在该背板上。
第三方面，提供一种交换设备，该交换设备包括上述第二方面或第二方面的任一种可 能的实现方式所提供的数据交换装置。
可以理解地，上述提供的任一种数据交换装置和交换设备，其所能达到的有益效果可对应参考上文所提供的通信芯片中的有益效果，此处不再赘述。
附图说明
图1为现有技术提供的一种路由器的结构示意图；
图2为本申请实施例提供的一种数据交换网络的应用示意图；
图3为本申请实施例提供的一种数据交换网络的结构示意图；
图4为本申请实施例提供的一种通信芯片的结构示意图；
图5为本申请实施例提供的另一种通信芯片的结构示意图；
图6为本申请实施例提供的又一种通信芯片的结构示意图；
图7为本申请实施例提供的另一种通信芯片的结构示意图；
图8为本申请实施例提供的一种数据交换的示意图；
图9为本申请实施例提供的另一种数据交换的示意图；
图10为本申请实施例提供的又一种数据交换的示意图；
图11为本申请实施例提供的另一种通信芯片的结构示意图。
具体实施方式
下面将结合本申请实施例中的附图，对本申请实施例中的技术方案进行描述。在本申请中，“至少一个”是指一个或者多个，“多个”是指两个或两个以上。“和/或”，描述关联对象的关联关系，表示可以存在三种关系，例如，A和/或B，可以表示：单独存在A，同时存在A和B，单独存在B的情况，其中A，B可以是单数或者复数。字符“/”一般表示前后关联对象是一种“或”的关系。“以下至少一项(个)”或其类似表达，是指的这些项中的任意组合，包括单项(个)或复数项(个)的任意组合。例如，a，b或c中的至少一项(个)，可以表示：a，b，c，a和b，a和c，b和c或a、b和c，其中a、b和c可以是单个，也可以是多个。另外，在本申请的实施例中，“第一”、“第二”等字样并不对数量和次序进行限定。
需要说明的是，本申请中，“示例性的”或者“例如”等词用于表示作例子、例证或说明。本申请中被描述为“示例性的”或者“例如”的任何实施例或设计方案不应被解释为比其他实施例或设计方案更优选或更具优势。确切而言，使用“示例性的”或者“例如”等词旨在以具体方式呈现相关概念。
本申请实施例提供的技术方案可以应用于多种不同的数据交换网络中，该数据交换网络也可以称为数据交换系统，可用于实现同一数据网络中不同设备间的数据交换，也可用于实现不同数据网络的数据交换。其中，该数据交换网络可以为交换机、路由器、或者由多个交换机或路由器组成的交换网络。示例性的，如图2所示，该不同数据网络可以包括局域网和因特网，该局域网和该因特网可以通过路由器进行数据交换，该局域网中可以包括手机、平板电脑和个人计算机等终端设备，该因特网中可以包括一个或者多个服务器。下面对该数据交换网络的结构进行举例说明。
图3为本申请实施例提供的一种m×n(m和n均为大于1的整数)的数据交换网络的结构示意图。该数据交换网络包括m个源节点(source，S)和n个目的节点(destination，D)，以及位于该m个源节点S与该n个目的节点之间的至少一个交换节点(switch，SW)。 图3中将该m个源节点S表示为S1至Sm，将该n个目的节点D表示为D1至Dn，以该至少一个交换节点SW包括多个交换节点且表示为SW1至SWi为例进行说明。
在该数据交换网络中，该m个源节点S和该n个目的节点D中的每个节点均可以包括一个或者多个端口。对于源节点S来说，该一个或者多个端口为输入端口(input port)，该输入端口可用于从该交换系统的外部接收数据包或信元，并发送至该至少一个交换节点SW。该至少一个交换节点SW可用于根据接收到的数据包或信元的目的信息(比如，目的IP地址、目的端口或者目的节点等信息)进行交换，并将交换后的数据包或信元发送给对应的目的节点D。对于目的节点D来说，该一个或者多个端口为输出端口(output port)，该输出端口可用于向该交换系统的外部发送已完成交换的数据包或信元。
可选的，该数据交换网络可用于实现可变长数据包(variable-length packet)的交换，也可以用于实现定长的信元(cell)的交换。进一步的，上述源节点S在接收到数据包时，可以保持原来的可变长数据包的格式进行发送，也可以将数据包切分成信元(cell)后发送；若源节点S是按照将数据包切分成信元的形式发送的，则目的节点D可以在接收到所有信元之后，再重组成完整的数据包。
需要说明的是，上述图3中以源节点S和目的节点D是两个独立的节点为例进行说明，在实际应用中，该源节点S和目的节点D也可以为同一节点，同时具有输入端口和输出端口，上述图3并不对本申请实施例构成限制。
在本申请实施例中，上述源节点S、目的节点D和交换节点SW可以是晶粒(die)，比如，源节点S和目的节点D可以是网络处理(network processing，NP)晶粒(也可以称为NP晶粒)，交换节点SW可以是交换晶粒(也可以称为SW晶粒)。因此，上述数据交换网络可以是一个包括多个NP晶粒和一个或者多个SW晶粒的通信芯片(chip)，或者是由多个这样的通信芯片合并封装在一起得到的芯片组。下面对该通信芯片的结构进行详细介绍说明。
图4为本申请实施例提供的一种通信芯片的结构示意图。该通信芯片包括多个SW晶粒10和多个NP晶粒20，该多个SW晶粒10中的任一SW晶粒与其它的SW晶粒相连，该多个NP晶粒20中的任一NP晶粒包括内部端口和外部端口，该内部端口用于与其它NP晶粒连接、或者与其它NP晶粒和一个SW晶粒均相连，该外部端口可用于接收来自外部的报文、或者向外部发送报文。图4中以该多个SW晶粒10包括两个SW晶粒且分别表示为11至12，该多个NP晶粒20包括四个NP晶粒且分别表示为21至24为例进行说明。
其中，上述晶粒也可以称为裸晶或裸片，从而SW晶粒也可以称为SW裸晶或SW裸片，NP晶粒也可以称为NP裸晶或NP裸片。可选的，该多个SW晶粒10和该多个NP晶粒20可以是从晶元(wafer)切割下来的多个独立的晶粒，也可以是位于同一晶元上的多个未切割的晶粒。
另外，对于该多个NP晶粒20中的每个NP晶粒，当该NP晶粒用于通过外部端口接收来自外部的某一报文时，该NP晶粒可以称为该报文的源NP晶粒，当该NP晶粒用于通过外部端口将输出至外部时，该NP晶粒可以称为该报文的目的NP晶粒。其中，同一报文的源NP晶粒和目的NP晶粒可以为同一NP晶粒，也可以为不同的NP晶粒。
该多个NP晶粒20中的任一NP晶粒(也可以称为第二NP晶粒，下文中以第二NP 晶粒为NP晶粒21为例)，用于：通过外部端口接收来自通信芯片外部的第一报文，获取第一报文的目的信息，并通过内部端口发送第二报文，第二报文包括第一报文和该目的信息，该目的信息用于指示第一报文的目的网络处理晶粒。示例性的，该NP晶粒21包括外部端口(也可以称为输入端口)，可用于通过该外部端口接收来自该通信芯片外部的第一报文，并解析第一报文的目的地址或目的端口，以根据该目的地址或目的端口确定该目的信息，该目的信息可以为目的地址或目的端口所对应的目的NP晶粒的标识；该NP晶粒21还可以通过内部端口发送包括第一报文和该目的信息的第二报文。可选的，该多个NP晶粒20中的每个NP晶粒中可以存储有该通信芯片对应的预设路由表，该预设路由表可用于指示该多个NP晶粒20中任意两个NP晶粒之间的路由信息，从而任一NP晶粒在解析得到第一报文的目的地址或目的端口时，可以根据对应的路由信息确定并发送第二报文。
该多个SW晶粒10中的任一SW晶粒，用于：接收第二报文，并根据该目的信息确定不满足预设条件时，向该多个NP晶粒20中的第一NP晶粒(比如，NP晶粒22或NP晶粒23)发送第二报文，或者在确定满足该预设条件时，向相连的SW晶粒发送第二报文。该预设条件包括：是该多个SW晶粒10中与目的NP晶粒相连的SW晶粒，或者是通过其它NP晶粒与目的NP晶粒相连且连接的NP晶粒的数量最少的SW晶粒。可选的，该多个SW晶粒10中的每个SW晶粒中也可以存储有该通信芯片对应的预设路由表，从而任一SW晶粒可以根据该预设路由表确定是否满足该预设条件，并发送第二报文。
可选的，如图4所示，该多个SW晶粒10包括第一SW晶粒11和第二SW晶粒12，第一SW晶粒11和第二SW晶粒12相连。具体的，第一SW晶粒11用于：接收第二报文，并根据该目的信息确定第一SW晶粒11不满足该预设条件时，向第二SW晶粒12发送第二报文；第二SW晶粒12用于：接收来自第一SW晶粒11的第二报文，并根据该目的信息确定第二SW晶粒12满足该预设条件时，向第一NP晶粒22发送第二报文。
其中，该多个SW晶粒10中的每个SW晶粒可以包括多个交换端口，该多个交换端口可以与至少一个NP晶粒相连，或者与至少一个NP晶粒和与其它的一个或者多个SW晶粒相连，每个SW晶粒可以将一个交换端口接收到的报文交换至另一交换端口输出。示例性的，假设第一SW晶粒11包括与NP晶粒21连接的第一交换端口、与NP晶粒22连接的第二交换端口、以及与第二SW晶粒12连接的第三交换端口，则第一SW晶粒11可以执行以下步骤：通过第一交换端口接收来自NP晶粒21的第二报文；根据该目的信息确定SW晶粒11满足该预设条件(比如，NP晶粒22为目的NP晶粒，第一SW晶粒11与NP晶粒22相连)时，将第二报文从第一交换端口交换至第二交换端口，以通过第二交换端口向NP晶粒22发送第二报文；或者，根据该目的信息确定第一SW晶粒11不满足该预设条件(比如，NP晶粒23为目的NP晶粒，第一SW晶粒11通过第二SW晶粒12与NP晶粒23相连)时，将第二报文从第一交换端口交换至第三交换端口，以通过第三交换端口向第二SW晶粒12发送第二报文，以使第二SW晶粒12在确定满足该预设条件时向NP晶粒23发送第二报文。
第一NP晶粒(下文中以NP晶粒23为例)用于：接收来自相连的SW晶粒的第二报文，并通过外部端口向外部发送第二报文，或者，通过内部端口将第二报文发送给NP晶粒。示例性的，第一NP晶粒23在接收到来自交换晶粒12的第二报文时，若第一报文的目的NP晶粒为第一NP晶粒23，第一NP晶粒23通过该外部端口发送第二报文；若第一 报文的目的NP晶粒不是第一NP晶粒23，第一NP晶粒23通过内部端口将第二报文发送给其他NP晶粒，比如，第一报文的目的NP晶粒为NP晶粒24，第一NP晶粒23将第二报文发送给该NP晶粒24。
进一步的，结合图4，如图5所示，若第一报文的目的NP晶粒不是第一NP晶粒23，第一报文的目的NP晶粒与该多个SW晶粒10中的任意一个SW晶粒均不相连，且该多个NP晶粒20中存在至少两个第一NP晶粒23(以两个NP晶粒为例且表示为23a和23b)均与第二SW晶粒12和该目的NP晶粒相连，即第二SW晶粒12与该目的NP晶粒之间存在多条传输路径，则第二SW晶粒12具体用于：在向第一NP晶粒22发送第二报文时，可以向至少两个第一NP晶粒23中的一个发送第二报文。比如，目的NP晶粒为NP晶粒24，第二SW晶粒12可以向第一NP晶粒23a发送第二报文，或者第二SW晶粒12向第一NP晶粒23b发送第二报文。
当第二SW晶粒12与第一报文的目的NP晶粒之间存在多条传输路径时，第二SW晶粒12可以按照一定的流量控制策略从该多条传输路径中选择一条传输路径向该目的NP晶粒传输第二报文。比如，该流量控制策略可以包括一条数据流通过一条传输路径传输，或者一条数据流按照负载均衡的方式通过多条传输路径传输等，本申请实施例对此不作具体限制。
类似的，如图5中的(a)所示，若该多个NP晶粒20中的第二NP晶粒21(即源NP晶粒)与该多个SW晶粒10中的任一个SW晶粒均不相连，则第二NP晶粒21具体用于：在通过外部端口接收第一报文后并通过内部端口发送第二报文时，通过内部端口向第三NP晶粒22发送第二报文，第三NP晶粒22与该多个SW晶粒20中的一个(比如，第一SW晶粒11)和第二NP晶粒21均相连。相应的，第三NP晶粒22用于：接收第二报文，并向第一SW晶粒11转发第二报文。
可选的，如图5中的(b)所示，若该多个NP晶粒20中存在至少两个第三NP晶粒22(以两个NP晶粒为例且表示为22a和22b)均与第二NP晶粒21和第一SW晶粒11相连，即第二NP晶粒21和第一SW晶粒11之间存在多条传输路径，则第二NP晶粒21具体用于：在通过内部端口向第三NP晶粒22发送第二报文时，向至少两个第三NP晶粒22中的一个发送第二报文。比如，第二NP晶粒21向第三NP晶粒22a发送第二报文，或者第二NP晶粒21向第三NP晶粒22b发送第二报文。
当第二NP晶粒21和第一SW晶粒11之间存在多条传输路径时，第二NP晶粒21可以按照一定的流量控制策略从该多条传输路径中选择一条传输路径向第一SW晶粒11传输第二报文。比如，该流量控制策略可以包括一条数据流通过一条传输路径传输，或者一条数据流按照负载均衡的方式通过多条传输路径传输等，本申请实施例对此不作具体限制。
进一步的，上述第二SW晶粒12可能满足该预设条件，也可能不满足该预设条件。当第二SW晶粒12满足该预设条件时，第二SW晶粒12可以向连接的第一NP晶粒23发送第二报文。当第二SW晶粒12不满足该预设条件，第二SW晶粒12可以向与第二SW晶粒12连接的其它SW晶粒发送第二报文，以使该其它SW晶粒向连接的第一NP晶粒23发送第二报文。
在一种可能的实施例中，如图6中的(a)所示，该多个SW晶粒10还包括第三SW晶粒13，第三SW晶粒13分别与第二SW晶粒12和第一NP晶粒23相连，则第二SW晶 粒12还用于：当确定第二SW晶粒12不满足该预设条件时，向第三SW晶粒13发送第二报文。相应的，第三SW晶粒13用于：接收来自第二SW晶粒12的第二报文，并根据第二报文中的目的信息确定第三SW晶粒13满足该预设条件，向第一NP晶粒23发送第二报文。
可选的，如图6中的(b)所示，该多个SW晶粒10中存在至少两个第二SW晶粒12(以两个SW晶粒为例且表示为12a和12b)与第一SW晶粒11和第三SW晶粒23均相连，即第一SW晶粒11和第三SW晶粒23之间存在多条传输路径，则第一SW晶粒11具体用于：在确定第一SW晶粒11不满足该预设条件，需要向第二SW晶粒12发送第二报文时，可以向至少两个第二SW晶粒12中的一个发送第二报文。比如，第一SW晶粒11向第二SW晶粒12a发送第二报文，或者第一SW晶粒11向第二SW晶粒12b发送第二报文。
可选的，当第一SW晶粒11与第三SW晶粒23之间存在多条传输路径时，第一SW晶粒11可以按照一定的流量控制策略从该多条传输路径中选择一条传输路径向第三SW晶粒23传输第二报文。比如，该流量控制策略可以包括一条数据流通过一条传输路径传输，或者一条数据流按照负载均衡的方式通过多条传输路径传输等，本申请实施例对此不作具体限制。
对于上述任一种可能的实施例，第一NP晶粒23不仅可用于通过外部端口向外部发送第二报文，也可以用于通过内部端口将第二报文发送给其他NP晶粒。为了保证第一NP晶粒23能够正确地传输该第二报文，与第一NP晶粒23相连的SW晶粒(或称为向第一NP晶粒23发送第二报文的SW晶粒)还用于：向第一NP晶粒23发送路径指示信息，该路径指示信息用于指示该第二报文的发送路径。相应的，第一NP晶粒23还用于：接收该路径指示信息，并根据该路径指示信息通过外部端口向外部发送第二报文，或者，根据该路径指示信息通过内部端口将第二报文发送给其他NP晶粒。进一步的，当与第一NP晶粒23相连的SW晶粒与该目的NP晶粒之间的传输路径中存在多个NP晶粒时，该路径指示信息还可用于指示第二报文在该多个NP晶粒中的发送路径，即该多个NP晶粒可根据该路径指示信息对第二报文进行转发，直至将第二报文转发至该目的NP晶粒。
在一种示例中，以图4所示的通信芯片为例，若NP晶粒21为源NP晶粒，NP晶粒21发送的第二报文通过第一SW晶粒11到达第二SW晶粒12，第二SW晶粒12确定满足该预设条件，第二SW晶粒12可以向第一NP晶粒23发送第二报文和该路径指示信息。若目的NP晶粒为第一NP晶粒23，该路径指示信息可用于指示第一NP晶粒23通过外部端口向外部发送第二报文，则第一NP晶粒23在接收到第二报文和该路径指示信息时，通过外部端口向外部发送第二报文。若目的NP晶粒为NP晶粒24，该路径指示信息可用于指示第一NP晶粒23通过内部端口将第二报文转发至NP晶粒24，则第一NP晶粒23在接收到第二报文和该路径指示信息时，根据该路径指示信息将第二报文转发至NP晶粒24。
在另一种示例中，以图6所示的通信芯片为例，若NP晶粒21为源NP晶粒，NP晶粒21发送的第二报文依次通过第一SW晶粒11和第二SW晶粒12到达第三SW晶粒13，第三SW晶粒13确定满足该预设条件，第三SW晶粒13可以向第一NP晶粒23a发送第二报文和路径指示信息。若目的NP晶粒为第一NP晶粒23a，该路径指示信息可用于指示第一NP晶粒23a通过外部端口向外部发送第二报文，则第一NP晶粒23a在接收到第二报 文和该路径指示信息时，通过外部端口向外部发送第二报文。若目的NP晶粒为NP晶粒24，该路径指示信息可用于指示第一NP晶粒23a通过内部端口将第二报文转发至NP晶粒24，则第一NP晶粒23a在接收到第二报文和该路径指示信息时，根据该路径指示信息将第二报文转发至NP晶粒24。
下面以图7所示的通信芯片为例，对第一NP晶粒23的具体结构和第一NP晶粒23根据该路径指示信息所执行的操作、以及第二NP晶粒21的具体结构和第二NP晶粒21选择传输路径所执行的操作进行举例说明。如图7所示，该通信芯片可以包括四个SW晶粒且分别表示为11、12a、12b和13，以及12个NP晶粒且分别表示为21、22a、22b、23a、23b、24、25a、25b、26、27a、27b和28。其中，SW晶粒11和SW晶粒13均与SW晶粒12a和SW晶粒12b连接，NP晶粒21分别通过NP晶粒22a和22b与SW晶粒11连接，NP晶粒24分别通过NP晶粒23a和23b与SW晶粒13连接，NP晶粒26分别通过NP晶粒25a和25b与SW晶粒12a连接，NP晶粒28分别通过NP晶粒27a和27b与SW晶粒12b连接。
第一、对第一NP晶粒23的具体结构，以及第一NP晶粒23根据该路径指示信息所执行的操作进行举例说明。
在一种可能的实施例中，如图7所示，以第一NP晶粒23为NP晶粒23a为例，第一NP晶粒23a包括第一选择开关231。其中，第一选择开关231包括连接端Q0、第一选择端Q1和第二选择端Q2，第一选择开关231的连接端Q0与第三SW晶粒13连接，第一选择开关231的第一选择端Q1和第二选择端Q2分别与第一NP晶粒23a的外部端口和该NP晶粒24连接。
具体的，以第三SW晶粒13向第一NP晶粒23a发送第二报文和路径指示信息为例，若该目的NP晶粒为第一NP晶粒23a，该路径指示信息用于指示第一NP晶粒23a通过外部端口向外部发送第二报文时，第一NP晶粒23a可以导通第一选择开关231的连接端Q0与第一选择开关231的第一选择端Q1。若该目的NP晶粒为NP晶粒24，该路径指示信息用于指示第一NP晶粒23a通过内部端口将第二报文转发至该NP晶粒24时，第一NP晶粒23a可以导通第一选择开关231的连接端Q0与第一选择开关231的第二选择端Q2。
可选的，第一选择开关231还可以包括至少两个第二选择端Q2，该至少两个第二选择端Q2分别与第一NP晶粒23a的至少两个内部端口相连，该至少两个内部端口分别与该多个NP晶粒中20的至少两个相连。此时，第一NP晶粒23还用于：在根据该路径指示信息通过内部端口将第二报文转发给其他NP晶粒时，导通第一选择开关221的连接端与该路径指示信息指示的发送路径对应的第二选择端Q2。
第二、对第二NP晶粒21的具体结构和第二NP晶粒21选择传输路径所执行的操作进行举例说明。
在一种可能的实施例中，如图7所示，第二NP晶粒21包括第二选择开关211。其中，第二选择开关211包括连接端P0和至少两个选择端P1至Pi(i为大于1的正整数)，第二选择开关211的连接端P0与该NP晶粒21的外部端口连接，第二选择开关211的至少两个选择端P1至Pi分别与至少两个NP晶粒连接。图7中以该至少两个NP晶粒包括两个NP晶粒且分别表示为22a和22b，该至少两个选择端P1至Pi包括两个选择端且分别表示为P1和P2，NP晶粒22a与选择端P1连接，NP晶粒22b与选择端P2连接为例进行说明。
具体的，第二NP晶粒21还用于：控制第二选择开关211的连接端P0与第二选择开关211的至少两个选择端P1至Pi中的一个选择端导通，以实现从上述至少两个NP晶粒中选择一个NP晶粒发送第二报文。示例性的，结合图7，该NP晶粒21可用于控制第二选择开关211的连接端P0与第二选择开关211的选择端P1导通，以从NP晶粒22a和NP晶粒22b中选择NP晶粒22a发送第二报文，即该NP晶粒21选择通过NP晶粒22a向第一SW晶粒11转发该第二报文。
可选的，该NP晶粒21和第一NP晶粒23a除了包括选择开关，还可以包括其他功能单元。示例性的，该NP晶粒21和第一NP晶粒23a还可以包括媒体接入控制(media access control)汇聚单元(aggregation)、网络处理器(network processor，NP)和流量管理器(traffic manager，TM)，媒体接入控制汇聚单元可以简称MAG。其中，MAG可用于感知光信号或电信号，将该光信号或电信号转发为数据帧(即报文)，并对该数据帧进行合法性检查。NP的上行处理可以包括报文解析、数据流分析、报文转发、以及将MAC/IP地址映射到TM队列号等；NP的下行处理上可以包括报文封装和出口处理等。TM可用于提供大容量内存、拥塞避免、流量整形、多级流量调度和流量拥塞控制等功能，还用于将TM队列号映射到目的NP晶粒的标识。
类似的，上述图7所示的多个NP晶粒中的任一NP晶粒均可以包括媒体接入控制汇聚单元MAG、网络处理器NP和流量管理器TM。进一步的，当多个NP晶粒中的任一NP晶粒不仅可用于接收外部的报文或向外部转发报文，而且可用于为其他NP晶粒或者SW晶粒转发报文时，上述NP晶粒中还可以包括与第一选择开关231或第二选择开关211类似的选择开关，本申请实施例对此不作具体限制。
此外，该多个NP晶粒20除了包括上述图7所示的4个SW晶粒和12个NP晶粒之外，还可以包括更多数量的SW晶粒和NP晶粒，任一SW晶粒均可以与其它的SW晶粒相连，任一NP晶粒均可以与其它的NP晶粒相连、或者同时与其它的NP晶粒和SW晶粒相连，每个相连NP晶粒均包括选择开关、MAG、NP和TM。
需要说明的是，上述NP晶粒21至NP晶粒24的相关描述，均是以NP晶粒在数据交换中的不同角度为例进行说明的。在该通信芯片中，该多个NP晶粒10中的任一个NP晶粒均有可能执行上述NP晶粒21至NP晶粒24中任意一个NP晶粒的功能，具体与需要交换的报文和该NP晶粒在该通信芯片中的位置(或称为连接关系)有关。
可选的，当上述NP晶粒包括选择开关、媒体接入控制汇聚单元MAG、网络处理器NP和流量管理器TM等多个不同功能单元时，该NP晶粒也可以被拆分成多个不同的晶粒，每个晶粒用于对应上述一个或者多个功能单元。示例性的，每个NP晶粒可以包括两个晶粒，这两个晶粒包括与MAG对应的晶粒，和NP、TM及选择开关对应的晶粒。
本申请实施例中的多个SW晶粒10的结构可以相同，也可以不同；类似的，多个NP晶粒20的结构可以相同，也可以不同。其中，当该多个SW晶粒10的结构相同，或者该多个NP晶粒20的结构相同时，可以通过一次流片或者相同的流片过程得到的完全相同的多个晶粒，通过配置和旋转该多个晶粒即可得到该通信芯片中不同位置上的SW晶粒或者NP晶粒，这样可以大大降低SW晶粒和NP晶粒的开发成本。
进一步的，当第一报文的源NP晶粒和目的NP晶粒为同一NP晶粒时，该源NP晶粒还可以用于通过外部端口发送第一报文。示例性的，以图4至图7中的任一图示所提供的 通信芯片中的第二NP晶粒21为例，若第一报文的源NP晶粒和目的NP晶粒均为第二NP晶粒21，则第二NP晶粒21还可用于：在接收到第一报文，且确定该目的NP晶粒为第二NP晶粒21时，通过外部端口发送第一报文。
此外，当第一报文的源NP晶粒和目的NP晶粒为相连的两个NP晶粒时，该源NP晶粒还可以用于通过内部端口向该目的NP晶粒发送第一报文。示例性的，以图5或图6中的第二NP晶粒21和NP晶粒22a为例，第二NP晶粒21还用于：在接收到第一报文，且确定该目的NP晶粒为该NP晶粒22a时，通过内部端口向该NP晶粒22a发送第一报文。
进一步的，该通信芯片还可以包括多个分组，该多个分组中的任一分组包括该多个NP晶粒20中的至少一个NP晶粒，和/或该多个SW晶粒10中的至少一个SW晶粒。在一种可能的实施例中，该多个分组中的任一分组包括该多个SW晶粒10中的至少一个SW晶粒，或者该任一分组包括该多个SW晶粒10中的至少一个SW晶粒和至少一个NP晶粒，该至少一个NP晶粒中的任一NP晶粒通过内部端口与同一分组中的一个SW晶粒相连，或者通过内部端口与同一分组中相邻的NP晶粒相连。
示例性的，以图7所示的通信芯片为例，该通信芯片可以包括4个分组，第1分组中可以包括SW晶粒11、NP晶粒21、NP晶粒22a和NP晶粒22b，第2分组中可以包括SW晶粒12a、NP晶粒26、NP晶粒25a和NP晶粒25b，第3分组中可以包括SW晶粒12b、NP晶粒28、NP晶粒27a和NP晶粒27b，第4分组中可以包括SW晶粒13、NP晶粒24、NP晶粒23a和NP晶粒23b。
可选的，当该多个NP晶粒20中相邻的两个NP晶粒属于不同的分组时，该相邻的两个NP晶粒可以不相连。示例性的，以图7所示的四个分组为例，NP晶粒22a和NP晶粒25a可以不相连，NP晶粒22b和NP晶粒27b可以不相连，NP晶粒25b和NP晶粒23b可以不相连，NP晶粒27a和NP晶粒23a可以不相连。在实际应用中，当该多个NP晶粒20中相邻的两个NP晶粒属于不同的分组时，该相邻的两个NP晶粒也可以相连，上述图7中仅以不同分组中相邻的两个NP晶粒不相连为例进行说明。
在一种可能的实施例中，当源NP晶粒和目的NP晶粒均属于一个分组时，该源NP晶粒接收到的报文可以通过该分组中的NP晶粒和/或SW晶粒传输至目的NP晶粒；当源NP晶粒和目的NP晶粒属于不同的分组时，该源NP晶粒接收到的报文可以通过多个分组中的NP晶粒和/或SW晶粒传输至目的NP晶粒。
为便于理解，下面以图8-图10所示的通信芯片为例，对该通信芯片中不同位置上的NP晶粒的报文对应的交换路径进行举例说明。该不同位置可以包括源NP晶粒和目的NP晶粒属于同一分组和不同分组，源NP晶粒为与SW晶粒直连的NP晶粒(简称直连NP晶粒)，源NP晶粒为与SW晶粒不直连的NP晶粒(简称非直连NP晶粒)，目的NP晶粒为直连NP晶粒和非直连NP晶粒。图8-图10中以该通信芯片包括4个SW晶粒(表示为SW1至SW4)和12个NP晶粒(分别表示为NP1-NP12)，SW1至SW4以及NP1-NP12被划分为四个分组，每个分组包括一个SW晶粒、两个直连NP晶粒和一个非直连NP晶粒为例进行说明。
在一种可能的示例中，如图8所示，对于源NP晶粒和目的NP晶粒均属于同一分组内，且源NP晶粒为直连的NP晶粒(比如，源NP晶粒为SW1直连的NP5)时可能的交换路径进行举例说明。具体的，若目的NP晶粒为直连NP晶粒，比如NP5，则报文对应 的交换路径S11为NP5-SW1-NP5，比如NP2，则报文对应的交换路径S12为NP5-SW1-NP2；若目的NP晶粒为非直连NP晶粒，比如NP1，则报文对应的交换路径S13a为NP5-SW1-NP2-NP1，或者交换路径S13b为NP5-SW1-NP5-NP1。对于源NP晶粒和目的NP晶粒均属于同一分组内，且源NP晶粒为非直连的NP晶粒(比如，源NP晶粒为SW3直连的NP9)时可能的交换路径进行举例说明。具体的，若目的NP晶粒为非直连NP晶粒，比如NP9，则报文的交换路径S21a为NP9-NP7-SW3-NP7-NP9，或者交换路径S21b为NP9-NP10-SW3-NP10-NP9，或者交换路径S21c为NP9-NP7-SW3-NP10-NP9或NP9-NP10-SW3-NP7-NP9；若目的NP晶粒为直连NP晶粒，比如NP7，则报文的交换路径S22a为NP9-NP7-SW3-NP7，或者交换路径S22b为NP9-NP10-SW3-NP7。
在另一种可能的示例中，如图9所示，对于源NP晶粒和目的NP晶粒属于不同分组，且源NP晶粒为直连的NP晶粒(比如，源NP晶粒为SW1直连的NP5)时可能的交换路径进行举例说明。具体的，以源NP晶粒和目的NP晶粒所属的两个分组相邻为例，若目的NP晶粒为直连NP晶粒，比如NP3，则报文的交换路径S31为NP5-SW1-SW2-NP3；若目的NP晶粒为非直连NP晶粒，比如NP4，则报文的交换路径S32为NP5-SW1-SW2-NP6/NP3-NP4。以源NP晶粒和目的NP晶粒所属的两个分组不相邻为例，若目的NP晶粒为直连NP晶粒，比如NP8，则报文的交换路径S41为NP5-SW1-SW2/SW3-SW4-NP8；若目的NP晶粒为非直连NP晶粒，比如NP12，则报文的交换路径S42为NP5-SW1-SW2/SW3-SW4-NP8/NP11-NP12。
在又一种可能的示例中，如图10所示，对于源NP晶粒和目的NP晶粒属于不同分组，且源NP晶粒为非直连的NP晶粒(比如，源NP晶粒为NP1)时可能的交换路径进行举例说明。具体的，以源NP晶粒和目的NP晶粒所属的两个分组相邻为例，若目的NP晶粒为直连NP晶粒，比如NP3，则报文的交换路径S51为NP1-NP5/NP2-SW1-SW2-NP3；若目的NP晶粒为非直连NP晶粒，比如NP4，则报文的交换路径S52为NP1-NP5/NP2-SW1-SW2-NP6/NP3-NP4。以源NP晶粒和目的NP晶粒所属的两个分组不相邻为例，若目的NP晶粒为直连NP晶粒，比如NP8，则报文的交换路径S61为NP1-NP5/NP2-SW1-SW2/SW3-SW4-NP8；若目的NP晶粒为非直连NP晶粒，比如NP12，则报文的交换路径S62为NP1-NP5/NP2-SW1-SW2/SW3-SW4-NP8/NP11-NP12。
需要说明的是，上述图8-图10中以源NP晶粒和目的NP晶粒为同一NP晶粒、或者为同一分组内相邻的两个NP晶粒时，对应的交换路径需要经过SW晶粒为例进行说明。可选的，当源NP晶粒和目的NP晶粒为同一NP晶粒、或者为同一分组内相邻的两个NP晶粒时，对应的交换路径也可以不经过SW晶粒，从而直接通过NP晶粒输出。示例性的，以源NP晶粒和目的NP晶粒均为NP4为例，NP4可以在接收到报文并确定该报文的目的NP晶粒为NP4时，将该报文通过对应的外部端口输出。
上述图4-图10仅是示例性的，当该通信芯片用于实现更大容量的数据交换时，该通信芯片还可以包括更多数量的NP晶粒和更多数量的SW晶粒。其中，该多个NP晶粒可用于接收外部的报文，或者向外部发送报文；该多个NP晶粒中的一部分NP晶粒还可以用于为另一部分NP晶粒转发报文，以将源NP晶粒接收到的报文转发至该多个SW晶粒；该多个NP晶粒中的一部分NP晶粒还可以用于为该多个SW晶粒转发报文，以将报文转发至目的NP晶粒。该多个SW晶粒中的每个SW晶粒可用于交换报文，不同的SW晶粒 之间还可用于转发报文。此外，当该通信芯片包括更多数量的NP晶粒和更多数量的SW晶粒时，该通信芯片的多个分组中每个分组所包括的晶粒的类型和数量可以相同，也可以不同。
在一种可能的实施例中，如图11所示，该通信芯片可以包括9个SW晶粒(表示为SW1至SW9)和16个NP晶粒(表示为NP1至NP16)，每个SW晶粒可以与至少一个NP晶粒相连，每个SW晶粒还可以与至少一个SW晶粒相连，每个NP晶粒可以与一个SW晶粒相连、或者与至少一个NP晶粒相连、或者同时与一个SW晶粒和至少一个NP晶粒相连。在一种示例中，如图11中的(a)所示，该通信芯片可以包括9个分组，其中4个分组中的每个分组可以包括三个NP晶粒和一个SW晶粒，另外4个分组中的每个分组可以包括一个NP晶粒和一个SW晶粒，最后1个分组可以包括一个SW晶粒。或者，在另一种示例中，如图11中的(b)所示，该通信芯片可以包括5个分组，其中4个分组中的每个分组可以包括四个NP晶粒和一个SW晶粒，最后1个分组可以包括一个SW晶粒。
上述图11所示的多个分组仅是示例性的，在实际应用中，该多个分组还可以按照其它方式划分的，本申请实施例对此不作具体限制。另外，图11中的(a)以相邻的两个NP晶粒属于不同分组时，该相邻的两个NP晶粒不相连为例进行说明；图11中的(b)以相邻的两个NP晶粒属于不同分组时，该相邻的两个NP晶粒相连为例进行说明。
进一步的，该通信芯片还可以包括：至少一个存储晶粒，该多个NP晶粒中的任意一个NP晶粒与该至少一个存储晶粒中的一个或者多个存储晶粒连接。可选的，该通信芯片包括多个存储晶粒，该多个存储晶粒与该多个NP晶粒一一对应，即每个NP晶粒与对应的一个存储晶粒连接。可选的，该存储晶粒的存储介质类型包括但不限于：动态随机存取存储器(dynamic random access memory，DRAM)、静态随机存取存储器(static random access memory，SRAM)、铁电随机存取存储器(ferroelectric random access memory，FeRAM)或者磁性随机存储器(magnetic random access memory，MRAM)等。
其中，当该至少一个存储晶粒包括多个存储晶粒时，该多个存储晶粒的存储介质类型可以相同或不同；当一个NP晶粒与多个存储晶粒耦合时，该NP晶粒对应耦合的该多个存储晶粒的存储介质类型可以相同或不同，本申请实施例对此不作具体限制。
本申请实施例中通信芯片包括多个SW晶粒和多个NP晶粒，该多个SW晶粒和该多个NP晶粒在该通信芯片内可实现连接，且该多个NP晶粒可用于接收外部的报文或者向外部发送报文，该多个SW晶粒可用于实现报文的分布式交换，从而在该通信芯片内部即可实现数据交换，与现有技术中通过多个NP芯片和多个交换芯片组成的数据交换网络相比，能够大大降低功耗和体积，同时减小采用该通信芯片的设备的散热成本和部署空间，以满足当前对于绿色低碳和可持续发展的需求。
基于此，本申请实施例还提供一种数据交换装置，该数据交换装置包括背板、以及设置在该背板上的至少一个通信芯片，该通信芯片为上文所提的任意一种通信芯片。可选的，该数据交换装置可以为交换盒子；或者，该数据交换装置可以为交换框。其中，当该数据交换装置为交换盒子时，该数据交换装置可以为单设备形态，不支持扩容，从而在该背板上可以不设置背板接口。当该数据交换装置为交换框时，该数据交换装置可以不是单设备形态，支持扩容，从而在该背板上可以设置相应的背板接口。
在一种示例中，当该数据交换装置为交换框时，该交换框包括与该至少一个通信芯片 对应的至少一个线卡，该至少一个通信芯片分别通过该至少一个线卡与该背板上的背板接口连接，以将该至少一个通信芯片设置在该背板上。
基于此，本申请实施例还提供一种交换设备，该交换设备可以为路由器或者交换机，该交换设备包括上文所提供的数据交换装置。
需要说明的是，上述关于通信芯片的详细描述均可对应援引到该数据交换装置和该交换设备中的，本申请实施例在此不再赘述。
最后应说明的是：以上所述，仅为本申请的具体实施方式，但本申请的保护范围并不局限于此，任何在本申请揭露的技术范围内的变化或替换，都应涵盖在本申请的保护范围之内。因此，本申请的保护范围应以所述权利要求的保护范围为准。

Claims (19)

  1. 一种通信芯片，其特征在于，所述通信芯片包括：多个交换晶粒和多个网络处理晶粒，所述多个交换晶粒中的任一交换晶粒与其它交换晶粒相连，所述多个网络处理晶粒中的任一网络处理晶粒包括外部端口和内部端口，所述内部端口与其他网络处理晶粒相连、或者与其它网络处理晶粒和所述多个交换晶粒中的一个交换晶粒均相连；
    所述多个网络处理晶粒中的任一网络处理晶粒，用于：通过所述外部端口接收来自所述通信芯片外部的第一报文，获取所述第一报文的目的信息，通过所述内部端口发送第二报文，所述第二报文包括所述第一报文和所述目的信息，所述目的信息用于指示所述第一报文的目的网络处理晶粒；
    所述多个交换晶粒中的任一交换晶粒，用于：接收所述第二报文，并根据所述目的信息确定不满足预设条件时，向相连的交换晶粒发送所述第二报文；或者，在确定满足所述预设条件时，向相连的第一网络处理晶粒发送所述第二报文；
    所述第一网络处理晶粒，用于：接收来自相连的交换晶粒的所述第二报文，并通过所述外部端口向外部发送所述第二报文，或者，通过所述内部端口将所述第二报文发送给其他网络处理晶粒；
    其中，所述预设条件包括：是所述多个交换晶粒中与所述目的网络处理晶粒相连的交换晶粒、或者是通过其它网络处理晶粒与所述目的网络处理晶粒相连且连接的网络处理晶粒的数量最少的交换晶粒。
  2. 根据权利要求1所述的通信芯片，其特征在于，所述多个交换晶粒包括第一交换晶粒和第二交换晶粒，所述第二交换晶粒与所述第一交换晶粒相连；
    所述第一交换晶粒，用于：接收所述第二报文，并根据所述目的信息确定所述第一交换晶粒不满足所述预设条件时，向所述第二交换晶粒发送所述第二报文；
    所述第二交换晶粒，用于：接收来自所述第一交换晶粒的所述第二报文，并根据所述目的信息确定所述第二交换晶粒满足所述预设条件时，向所述第一网络处理晶粒发送所述第二报文。
  3. 根据权利要求2所述的通信芯片，其特征在于，所述多个交换晶粒还包括第三交换晶粒，所述第三交换晶粒与所述第二交换晶粒相连；
    所述第二交换晶粒，还用于：当确定所述第二交换晶粒不满足所述预设条件时，向所述第三交换晶粒发送所述第二报文；
    所述第三交换晶粒，用于：接收来自所述第二交换晶粒的所述第二报文，并根据所述目的信息确定所述第三交换晶粒满足所述预设条件时，向所述第一网络处理晶粒发送所述第二报文。
  4. 根据权利要求3所述的通信芯片，其特征在于，所述多个交换晶粒中存在至少两个所述第二交换晶粒与所述第一交换晶粒和所述第三交换晶粒均相连；
    所述第一交换晶粒，具体用于：在向所述第二交换晶粒发送所述第二报文时，向所述至少两个第二交换晶粒中的一个发送所述第二报文。
  5. 根据权利要求1-4任一项所述的通信芯片，其特征在于，
    所述多个交换晶粒中的任一交换晶粒，还用于：在向所述第一网络处理晶粒发送 所述第二报文时，向所述第一网络处理晶粒发送路径指示信息，所述路径指示信息用于指示所述第二报文的发送路径；
    所述第一网络处理晶粒，还用于：接收所述路径指示信息，并根据所述路径指示信息通过所述外部端口向外部发送所述第二报文，或者，根据所述路径指示信息通过所述内部端口将所述第二报文发送给其他网络处理晶粒。
  6. 根据权利要求5所述的通信芯片，其特征在于，所述第一网络处理晶粒包括第一选择开关，所述第一选择开关包括连接端、第一选择端和第二选择端，所述第一选择开关的连接端与所述多个交换晶粒中的一个相连，所述第一选择开关的第一选择端与所述外部端口相连，所述第一选择开关的第二选择端与其它网络处理晶粒相连；
    所述第一网络处理晶粒，还用于：在根据所述路径指示信息通过所述外部端口向外部发送所述第二报文时，导通所述第一选择开关的连接端与所述第一选择开关的第一选择端；或者，
    所述第一网络处理晶粒，还用于：在根据所述路径指示信息通过所述内部端口将所述第二报文发送给其他网络处理晶粒时，导通所述第一选择开关的连接端与所述第一选择开关的第二选择端。
  7. 根据权利要求6所述的通信芯片，其特征在于，所述第一选择开关包括至少两个所述第二选择端，所述至少两个所述第二选择端分别与所述第一网络处理晶粒的至少两个所述内部端口相连，所述至少两个内部端口分别与所述多个网络处理晶粒中的至少两个相连；
    所述第一网络处理晶粒，还用于：在根据所述路径指示信息通过所述内部端口将所述第二报文发送给其他网络处理晶粒时，导通所述第一选择开关的连接端与所述路径指示信息指示的发送路径对应的所述第二选择端。
  8. 根据权利要求2-7任一项所述的通信芯片，其特征在于，所述目的网络处理晶粒与所述多个交换晶粒中的任一交换晶粒不相连，所述多个网络处理晶粒中存在至少两个所述第一网络处理晶粒均与所述第二交换晶粒和所述目的网络处理晶粒相连；
    所述第二交换晶粒，还用于：在向所述第一网络处理晶粒发送所述第二报文时，向所述至少两个所述第一网络处理晶粒中的一个发送所述第二报文。
  9. 根据权利要求1-8任一项所述的通信芯片，其特征在于，所述多个网络处理晶粒中存在第二网络处理晶粒与所述多个交换晶粒中的任一个均不相连；
    所述第二网络处理晶粒，具体用于：在通过所述外部端口接收所述第一报文后并通过所述内部端口发送所述第二报文时，通过所述内部端口向第三网络处理晶粒发送所述第二报文，所述第三网络处理晶粒与所述多个交换晶粒中的一个交换晶粒和所述第二网络处理晶粒均相连；
    所述第三网络处理晶粒，用于：接收所述第二报文，并向所述交换晶粒转发所述第二报文。
  10. 根据权利要求9所述的通信芯片，其特征在于，所述多个网络处理晶粒中存在至少两个所述第三网络处理晶粒；
    所述第二网络处理晶粒，具体用于：在通过所述内部端口向所述第三网络处理晶粒发送所述第二报文时，向所述至少两个第三网络处理晶粒中的一个发送所述第二报 文。
  11. 根据权利要求10所述的通信芯片，其特征在于，所述第二网络处理晶粒还包括第二选择开关，所述第二选择开关包括连接端和至少两个选择端，所述第二选择开关的连接端与所述外部端口相连，所述第二选择开关的所述至少两个选择端分别与所述至少两个第三网络处理晶粒相连；
    所述第二网络处理晶粒，还具体用于：导通所述第二选择开关的连接端与所述至少两个选择端中的一个选择端，以从所述至少两个第三网络处理晶粒中选择一个发送所述第二报文。
  12. 根据权利要求11所述的通信芯片，其特征在于，
    所述第二网络处理晶粒，还用于：在所述目的网络处理晶粒为所述第二网络处理晶粒时，通过所述外部端口发送所述第一报文。
  13. 根据权利要求1-12任一项所述的通信芯片，其特征在于，所述通信芯片包括多个分组，所述多个分组中的任一分组包括所述多个网络处理晶粒中的至少一个网络处理晶粒，所述多个分组中任意两个分组中的网络处理晶粒不相连；
    其中，同一分组中的至少一个网络处理晶粒通过所述内部端口连接同一所述多个交换晶粒中的一个交换晶粒，或通过所述内部端口连接同一分组中的相邻网络处理晶粒。
  14. 根据权利要求1-13任一项所述的通信芯片，其特征在于，所述多个交换晶粒和所述多个网络处理晶粒位于同一晶元中。
  15. 根据权利要求1-14任一项所述的通信芯片，其特征在于，所述通信芯片还包括：多个存储晶粒，所述多个网络处理晶粒中的任意一个网络处理晶粒与所述多个存储晶粒中的一个或者多个存储晶粒连接。
  16. 一种数据交换装置，其特征在于，所述数据交换装置包括背板、以及设置在所述背板上的至少一个通信芯片，所述通信芯片为权利要求1-15任一项所述的通信芯片。
  17. 根据权利要求16所述的装置，其特征在于，所述数据交换装置包括交换盒子或交换框。
  18. 根据权利要求17所述的装置，其特征在于，当所述数据交换装置包括交换框时，所述交换框包括与所述至少一个通信芯片对应的至少一个线卡，所述至少一个通信芯片分别通过所述至少一个线卡设置在所述背板上。
  19. 一种交换设备，其特征在于，所述交换设备包括如权利要求16-18任一项所述的数据交换装置。
PCT/CN2023/095556 2022-07-18 2023-05-22 一种通信芯片及数据交换装置 WO2024016825A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202210843102.4 2022-07-18
CN202210843102.4A CN115514704B (zh) 2022-07-18 2022-07-18 一种通信芯片及数据交换装置、交换设备

Publications (1)

Publication Number Publication Date
WO2024016825A1 true WO2024016825A1 (zh) 2024-01-25

Family

ID=84501756

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2023/095556 WO2024016825A1 (zh) 2022-07-18 2023-05-22 一种通信芯片及数据交换装置

Country Status (2)

Country Link
CN (2) CN117424852A (zh)
WO (1) WO2024016825A1 (zh)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117118902A (zh) * 2022-04-29 2023-11-24 华为技术有限公司 一种通信芯片及数据交换装置
CN117424852A (zh) * 2022-07-18 2024-01-19 华为技术有限公司 一种通信芯片及数据交换装置

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20020181452A1 (en) * 2001-06-01 2002-12-05 Norman Richard S. Cell-based switch fabric architecture implemented on a single chip
JP2013258299A (ja) * 2012-06-13 2013-12-26 Canon Machinery Inc ウエハリング交換装置およびチップ実装装置
CN113079090A (zh) * 2020-01-06 2021-07-06 华为技术有限公司 一种流量传输的方法、节点和系统
CN114116595A (zh) * 2020-08-28 2022-03-01 华为技术有限公司 数据传输的方法和系统
CN115514704A (zh) * 2022-07-18 2022-12-23 华为技术有限公司 一种通信芯片及数据交换装置

Family Cites Families (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9065722B2 (en) * 2012-12-23 2015-06-23 Advanced Micro Devices, Inc. Die-stacked device with partitioned multi-hop network
KR102229942B1 (ko) * 2014-07-09 2021-03-22 삼성전자주식회사 멀티 다이들을 갖는 멀티 채널 반도체 장치의 동작 방법 및 그에 따른 반도체 장치
US20180048562A1 (en) * 2016-08-09 2018-02-15 Knuedge Incorporated Network Processor Inter-Device Packet Source ID Tagging for Domain Security
US10708127B1 (en) * 2017-12-29 2020-07-07 Arista Networks, Inc. Low-latency network switching device with latency identification and diagnostics
US11422708B2 (en) * 2019-07-05 2022-08-23 SK Hynix Inc. Memory interface, data storage device including the same and operating method thereof
US11476211B2 (en) * 2019-12-19 2022-10-18 Nepes Co., Ltd. Semiconductor package and manufacturing method thereof
WO2022041158A1 (zh) * 2020-08-28 2022-03-03 华为技术有限公司 一种多芯片封装结构、交换机
EP4198752A4 (en) * 2020-08-31 2023-09-20 Huawei Technologies Co., Ltd. CHIP AND CHIP PACKAGE

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20020181452A1 (en) * 2001-06-01 2002-12-05 Norman Richard S. Cell-based switch fabric architecture implemented on a single chip
JP2013258299A (ja) * 2012-06-13 2013-12-26 Canon Machinery Inc ウエハリング交換装置およびチップ実装装置
CN113079090A (zh) * 2020-01-06 2021-07-06 华为技术有限公司 一种流量传输的方法、节点和系统
CN114116595A (zh) * 2020-08-28 2022-03-01 华为技术有限公司 数据传输的方法和系统
CN115514704A (zh) * 2022-07-18 2022-12-23 华为技术有限公司 一种通信芯片及数据交换装置

Also Published As

Publication number Publication date
CN115514704B (zh) 2023-08-22
CN115514704A (zh) 2022-12-23
CN117424852A (zh) 2024-01-19

Similar Documents

Publication Publication Date Title
WO2024016825A1 (zh) 一种通信芯片及数据交换装置
TWI284469B (en) A network system having a plurality of switches capable of improving transmission efficiency and method thereof
US7266112B1 (en) Dial access stack architecture
RU2461140C2 (ru) СПОСОБ РАСШИРЕНИЯ ФИЗИЧЕСКОЙ ОБЛАСТИ ДЕЙСТВИЯ СЕТИ InfiniBand
US9614759B2 (en) Systems and methods for providing anycast MAC addressing in an information handling system
US7145866B1 (en) Virtual network devices
CN101325497B (zh) 在不存在自动协商标准的接口上的自动协商
WO2022094771A1 (zh) 网络芯片和网络设备
CN109474534B (zh) 基于fpga的路由加速转发系统、路由器及转发方法
CN102904825B (zh) 一种基于Hash的报文传输方法和设备
EP4184937A1 (en) Method, apparatus, and system for communication in data centre
WO2023207310A1 (zh) 一种通信芯片及数据交换装置
WO2020093871A1 (zh) 下行报文发送、转发方法和装置
WO2022253087A1 (zh) 一种数据传输方法、节点、网络管理器及系统
EP4325800A1 (en) Packet forwarding method and apparatus
WO2022147792A1 (zh) 一种交换系统、交换网络和交换节点
WO2021012902A1 (zh) 一种处理网络拥塞的方法以及相关装置
WO2008128464A1 (fr) Dispositif et procédé pour réaliser le transfert d'un flux unique par de multiples unités de traitement en réseau
Cisco Catalyst 3000 Theory of Operation
Cisco Catalyst 3000 Theory of Operation
CN101232510A (zh) 一种多网口设备组播的实现方法及系统
CN111434079A (zh) 一种数据通信方法及装置
Cisco Catalyst 3000 Theory of Operation
Cisco Catalyst 3000 Theory of Operation
Cisco Catalyst 3000 Theory of Operation

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 23841894

Country of ref document: EP

Kind code of ref document: A1