CN201947293U - Network, system and processor using multicast route based on tree - Google Patents

Network, system and processor using multicast route based on tree Download PDF

Info

Publication number
CN201947293U
CN201947293U CN2010206995097U CN201020699509U CN201947293U CN 201947293 U CN201947293 U CN 201947293U CN 2010206995097 U CN2010206995097 U CN 2010206995097U CN 201020699509 U CN201020699509 U CN 201020699509U CN 201947293 U CN201947293 U CN 201947293U
Authority
CN
China
Prior art keywords
node
zone
data
adjacent
destination
Prior art date
Legal status (The legal status 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 status listed.)
Expired - Lifetime
Application number
CN2010206995097U
Other languages
Chinese (zh)
Inventor
王凯峰
朱鹏飞
孙红霞
吴永强
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
STMicroelectronics Beijing R&D Co Ltd
Original Assignee
STMicroelectronics Beijing R&D Co Ltd
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 STMicroelectronics Beijing R&D Co Ltd filed Critical STMicroelectronics Beijing R&D Co Ltd
Priority to CN2010206995097U priority Critical patent/CN201947293U/en
Application granted granted Critical
Publication of CN201947293U publication Critical patent/CN201947293U/en
Anticipated expiration legal-status Critical
Expired - Lifetime legal-status Critical Current

Links

Images

Landscapes

  • Data Exchanges In Wide-Area Networks (AREA)

Abstract

The utility model discloses a modified multicast route scheme based on tree. In a mesh network with a plurality of nodes among which data is transmitted, substantial data transmission can be launched and executed to sent the data to a correct processing node to be executed. In order to send the data to the desired place (for example, the correct destination node), routing algorithm definition is used for defining a rule set for transmitting data with high efficiency among the nodes until reaching the destination node. In order to ensure the purpose that all data is correctly transmitted among the nodes in a reasonable and efficient manner, the routing algorithm can define a node subset as an area, and furthermore, the data can be sent through the area. By identifying the specific neighbouring relation of a group of destination nodes and rerouting data so that the data uses the neighbouring relation by passing through the area different from the residing area of the destination nodes, better overall efficiency can be realized.

Description

Use is based on network, system and the processor of the multicast route of tree
Technical field
The utility model relates to the route in the mesh network.
Background technology
Netted networking is such one type networking, and wherein each node in the network can serve as independently router, and no matter whether this node is connected to another network.Mesh network can be realized by the computational entity of any amount.For example, some wireless computer devices (such as, the intelligent movable phone) can form a mesh network.As another example, the some processing unit on single integrated circuit (IC) chip or a plurality of IC can form a mesh network.This mesh network utilizes the path between a plurality of processing nodes and the node, allow to realize continue connect and around the path that breaks down or block by using the different paths that are used for transfer of data between the node to reconfigure.The whole mesh networks connected to one another of node are full-mesh networks.
A particular aspects of mesh network realizes in the IC that is classified as ultra-large integrated (VLSI) chip.In these IC, the two dimension of processing node (2D) net can be realized on single flat IC chip.This multiprocessor IC is used to use efficiently the transistor of the obtainable more and more quantity of modern VLSI technology more and more widely.Along with the increase of processing node quantity, the realization of network-on-chip (that is 2D mesh network) has promoted communication and the transfer of data between each processing node.Be used to promote that the overall plan of this communication and transfer of data is called as routing algorithm or simple route.Conventional route at the 2D mesh network on the IC provides very simple grid network, and this grid network can obtain short connection the in the chip architecture.Yet, when forming a plurality of communications and data transfer path, have a large amount of problems.When a node may need to some node broadcasts data, the not high and trouble of conventional routing algorithm efficient, even if this is because can obtain the route of short and more efficient, but some data may be replicated in concurrent path or parallel branch path.
The utility model content
According on the other hand of the present utility model, a kind of mesh network is provided, comprising: node matrix equation; And the source node in the described node matrix equation, can operate a plurality of other nodes that are used for to described matrix and send data; Wherein said mesh network is by the routing algorithm management of supporting by adjacent destination node route data.
In one embodiment, described mesh network further comprises having the two-dimension netted network that m is capable and n is listed as.
In another embodiment, described mesh network further comprises three-dimensional netted network.
In another embodiment, described mesh network comprises the multinode processor.
In another embodiment, each node comprises processing node in the described mesh network.
According on the other hand of the present utility model, a kind of processor is provided, comprising: be arranged in a plurality of nodes in the mesh network by row and column, can operate and be used for transmitting and receive data according to routing algorithm.Described routing algorithm comprises: determine source node and a plurality of destinations node that receives data from described source node; Each destination node is associated with a zone in a plurality of zones; Determine whether a destination node in the zone is adjacent with another regional destination node, and if adjacent, then calculate for being from which zone for the data of described source node transmission zone more efficiently; Adjacent destination node in the zones of different is associated with described zone more efficiently; And to institute's node transmission data on purpose.
In one embodiment, described processor further comprises polycaryon processor, and wherein each node all is a processing unit.
In another embodiment, the described a plurality of nodes in the described processor comprise the capable and n of m row, and comprise for the y=even number be l (x, y)=y*n+x and for the y=odd number be l (x, y)=node identification of y*n+n-x-1.
In another embodiment, described processor further comprise by with described source node in going together mutually two adjacent nodes and with four node J1, J2, J3 and the J4s adjacent of two the adjacent nodes definition of described source node in same column with described source node, wherein related with each adjacent node sign comprises the value of following J1>J2>J3>J4.
In another embodiment, described a plurality of zones comprise that the sign of described each node is relevant with the sign of described source node by four zones of the sign of each node definition, wherein: the node that regional 1 comprises and having the sign between [J1, (m * n)-1]; Zone 2: node with the label between [J2, J1-1]; Zone 3: node with the label between [J4+1, J3]; And zone 4: node with the label between [0, J4].
In another embodiment, described processor further comprises the single integrated circuit tube core.
In another embodiment,, further comprise a plurality of integrated circuit leads.
According on the other hand of the present utility model, a kind of system is provided, comprising: a plurality of processing nodes in the network, can operate and be used for according to routing algorithm from first node to a plurality of other nodes transmission data.Described algorithm comprises: each node that will be except described first node is associated with a zone in a plurality of zones; Determine a node in the zone whether with another zone in node adjacent, and if adjacent, then calculate for being from which zone for the data of described first node transmission zone more efficiently; Adjacent node in the zones of different is associated with described zone more efficiently; And to described node transmission data.
In one embodiment, each node comprises the computing equipment with at least one processing entities.
In another embodiment, each node comprises the two-dimension netted network of second grade, and the mesh network of each second grade can be operated and be used for according to described routing algorithm route data.
In another embodiment, described system further comprises by the be coupled memory of at least one node of bus.
In another embodiment, described system comprises the network that constitutes by multistage computing network computing equipment coupled to each other.
Description of drawings
With reference to following detailed, will understand the execution mode of disclosed theme here easier and better in conjunction with the accompanying drawings.
Fig. 1 is used for block diagram based on 4 * 4 networks of the mapping scheme of the two-dimensional network of tree according to having of the execution mode of disclosed theme here.
Fig. 2 is used for block diagram based on 6 * 6 networks of the unmodified routing plan of the two-dimensional network of tree according to having of the execution mode of disclosed theme here.
Fig. 3 is used for block diagram based on 6 * 6 networks of the modified routing plan of the two-dimensional network of tree according to having of the execution mode of disclosed theme here.
Fig. 4 is the block diagram of execution mode of computer system 400 that can realize the mesh network system 300 of Fig. 3.
Embodiment
Following discussion proposes in order to make those skilled in the art make and to use theme disclosed herein.General Principle described herein can be applied under the situation of spirit that does not deviate from current specifically described execution mode and scope except that the execution mode of above detailed description and other execution modes and the application the application.Current disclosure is not to be intended to be limited to the execution mode that illustrates, but should meet with principle and the corresponding to scope the most widely of feature disclosed herein or suggestion.
Before the specific detail that various execution modes are discussed, introduce general introduction to this theme.Have in a plurality of mesh networks, can initiate and carry out mass data and transmit and message (for example, data) is delivered to correct processing node carry out at the node of Data transmission each other.For data being delivered to the place (for example, correct destination node) that these data need be gone, use routing algorithm to be defined in and transmit data between the node efficiently up to the rule set that arrives the destination node.In order to guarantee the purpose of all data with the correct transmission between node of efficient way reasonably, routing algorithm can become the zone with the node subset definition, then data is sent in these zones so that carry out route.As below will discussing in further detail, by discerning one group of special neighbourhood relations between the node of destination, and by re-route data make its by be different from the destination node the zone in resident zone utilize this adjacently, can realize better overall efficient.Therefore, total grade of data transfer activities can obtain reducing by using single path that similar data is routed to a plurality of adjacent destination nodes.These and other notions describe in further detail with reference to figure 1-Fig. 4.
Fig. 1 is used for block diagram based on 4 * 4 matrixes (for example, network) of the mapping scheme of the two-dimensional network 100 of tree according to having of an execution mode of disclosed theme here.In the two-dimensional network based on tree, each node 110 can and finally be that unique identifier in the context of routing algorithm is quoted by the context of mapping scheme.Therefore, the basis of mapping scheme influences the person's character based on the routing algorithm of the sign of each node inherently.
Any given two-dimension netted network will have the capable and n row of m.In order to realize the Multicast Routing Algorithm based on tree, all nodes in the network must at first utilize sign partition function l to identify.The function l that is used for the 2D mesh network of m * n can be as follows according to the x and the y coordinate representation of node:
Figure BSA00000415091700051
In the mapping scheme of this 4 * 4 network, can node 110 be designated in 16 different nodes one with identifier 0-15.First node in this example can be the node of lower left and be designated node 0.Mapping scheme is node 1-3 with node identification one after the other to the right along end row then.Next, identification schemes moves to second row, but rests on right-hand node 4 of this network.Scheme is moved back into left in the process of identification nodes 5-7, move to the third line then similarly and move right once more.Repeat this pattern up to all nodes all by unique node identifier sign-in this example, be the node identifier between 0 and 15 as shown in fig. 1.Utilize this suitable mapping scheme, routing algorithm can use these node identifications that are used for the data route to realize.Below with reference to Fig. 2 and Fig. 3 two these type of routing algorithms are described.
Fig. 2 is used for block diagram based on 6 * 6 matrixes (for example, network) 200 of the simple routing plan of the two-dimensional network of tree according to having of the execution mode of disclosed theme here.For the example illustrative purposes, run through the disclosure and will use an exemplary scene.This example comprises the source node Pi that the some destinations node in network 200 is sent multicast message (for example, data).Therefore, in order to be easy to explanation, in this example, source node Pi is not a fringe node (edgenode) and corresponding to the node that is designated P15.This source node Pi must have four adjacent nodes, and its label that distributes according to sign in the mapping scheme is labeled as J1, J2, J3 and J4 by the mode of successively decreasing.That is, node identifier is determined label, makes J1>J2>J3>J4, and for example, for the source node Pi with P 15 identifiers, J1 will be the maximal contiguous node that is designated node P20, and J2 will be node P16, and J3 will be that node P14 and J4 will be node P8.For this mapping scheme and routing algorithm, adjacent node is defined as the node that is directly adjacent to source node Pi in horizontal or vertical direction, for example, and in identical row or column and without any intermediate node.The diagonal node is not thought adjacent.
Utilization is defined as four adjacent nodes of J1-J4, can with every other node definition for the part of one and the label zone that only is associated (suitably called after zone 1-4) with an adjacent node J1-J4.According to following equation, with the every other node definition in the mesh network 200 in the zone:
■ zone 1: the node of label between [J1, (m * n)-1];
■ zone 2: the node of label between [J2, J1-1];
■ zone 3: the node of label between [J4+1, J3]; And
■ zone 4: the node of label between [0, J4].
When Pi received multicast message, it extracts this message from the packet header of message destination collection D generated four new destination subset D 1, D2, D3 and D4 corresponding to the destination node in each zone then.Rule set can be as follows:
■ generates D1 by select the destination node of zone in 1 from D, and if D1 be empty (for example, having at least one destination node of gathering among the D in the zone 1), then utilize new destination collection D1 information to send message to J1;
■ generates D2 by select the destination node of zone in 2 from D, and if D2 be sky, then utilize new destination collection D2 information to send message to J2;
■ generates D3 by select the destination node of zone in 3 from D, and if D3 be sky, then utilize new destination collection D3 information to send message to J3; And
■ generates D4 by select the destination node of zone in 4 from D, and if D4 be sky, then utilize new destination collection D4 information to send message to J4.
According to this suitable rule set, a suitable message of sending in four adjacent node J1 or J2 or J3 or J4 then from message set D.When the suitable new destination of adjacent node utilization subset information receives multicast message, can these multicast message data of high-speed cache.If this adjacent node is real destination node (that is, this node is the final destination of the particular data appointment among the message set D), then do not need these data are passed to another destination node.Yet,, can repeat this process and make from four new subset D 1, D2, D3 and the D4 of the viewpoint definition new region 1-4 of new source node and four new adjacent nodes and multicast message D for every other data.Repeat this process and all arrive its destination node up to all data.Therefore, the data path of each the destination node in the example shown in Figure 2 has disclosed the path by the institute of the arrow between different nodes mark.
In order to continue example above, source node P15 can bear the task of sending multicast message to destination set of node D={P2, P5, P6, P9, P13, P22, P29, P33, P35}.Before subsequent node transmission multicast message, source node P15 will at first generate four new destination collection D1, D2, D3 and D4.For source node P15, its four adjacent node J1, J2, J3 and J4 are P20, P16, P14 and P8, correspondingly corresponding to four regional 1-4, these four zones have following node corresponding in each zone: [P20-P35], [P16-P19], [P9-P14] and [P0-P8].Therefore can calculate four new destination collection D1-D4:D1={P22, P29, P33, P35}; D2={ sky }; D3={P9, P13} and D4={P2, P5, P6}.
Because destination collection D1, D3 and D4 are not empty, so source node P15 will utilize the information of new destination collection D1, D3 and D4 to J1, J3 and J4 message transfer.Source node P15 does not need the message transfer to J2, because D2 is empty, this means in the zone 2 does not have the destination node.
The respective subset that all receives multicast message D as P20, P14 or P8 (for example, be respectively D1, D3 or D4) time, each node in these nodes all becomes the source node of bearing the task of sending the new multicast message of its subclass that is associated conduct now, and this new multicast message has its corresponding destination collection of determining in the mode identical with mode discussed above.Therefore, repeat this two steps process up to institute on purpose collection be empty (for example, all message have all reached its corresponding target ground node).
Though the Multicast Routing Algorithm of describing among Fig. 2 based on tree has successfully been transmitted multicast message D to all appropriate purpose ground nodes, this execution does not utilize node architecture.That is,, the route of multicast message just is limited to fully path by single zone in case place a destination to concentrate the destination node.Therefore, multicast message for the attainment of one's purpose the ground node non-destination node that must pass through quantity (promptly, the quantity of node transmission-run through remainder of the present disclosure, the quantity of node transmission is called as " jumping (hop) " and is represented by the arrow between the node among Fig. 2) may be bigger, though some destination node is adjacent one another are, is arranged in different prime areas.
Consider the example among Fig. 2, destination node P2 and P9 and destination node P13 and P22 are the right examples of adjacent destination node that appears in the zones of different (about original source node Pi).Destination node P2 and P9 are respectively in zone 4 and zone 3.Similarly, destination node P13 and P22 are respectively in zone 3 and zone 1.Therefore, about the comparison of destination node P13 and P22, destination node P22 will receive multicast message from node P21, and P13 will receive multicast message D from P14 via double bounce.In order to arrive destination node P22, although node P20 and P21 are not the destination nodes, multicast message D must experience three and jump from source node P15 by node P20 and node P21.Therefore, by redundancy and parallel route to a certain degree, need 5 jumpings multicast message could be delivered to this two adjacent nodes (P13 and P22) altogether.In fact, can obtain higher performance by making destination node P22 directly receive multicast message D from destination node P13.The transmission multicast message only needs a jumping from node P13 to node P22, and needs double bounce from node P20 to node P21 to node P22 transmission multicast message D.Jumping figure analysis in the routing algorithm shown in Figure 2 disclosed add up to 19.Therefore, discuss with reference to figure 3, can realize routing algorithm more efficiently by the adjacent destination node in the identification zones of different as following.
Fig. 3 is the basis block diagram of 6 * 6 networks 300 of the execution mode of disclosed theme here, and this 6 * 6 network 300 uses the routing algorithm of revising to carry out data communication in the two-dimensional network based on tree.Routing algorithm example shown in Fig. 3 is supported the direct communication between the node of adjacent destination in the zones of different.Therefore, can realize high efficiency, only be 17 because use total jumping figure of algorithm shown in Fig. 3, and it is less than using 19 required jumpings of routing algorithm shown in Fig. 2.This can finish by adding additional step in the first pass analysis during from the multicast message D of source node Pi in the subclass of definite destination.
As previously mentioned, when Pi receives multicast message, it extracts this message from the packet header of message destination collection D (D={P2, P5, P6, P9, P 13, P22, P29, P33, P35}) generates four new destination collection D1, D2, D3 and D4 corresponding to destination node in each zone then.Use and identical before rule set:
■ generates D1 by select the destination node of zone in 1 from D, and if D1 be empty (for example, having at least one destination node of gathering among the D in the zone 1), then utilize new destination collection D1 information to send message to J1;
■ generates D2 by select the destination node of zone in 2 from D, and if D2 be sky, then utilize new destination collection D2 information to send message to J2;
■ generates D3 by select the destination node of zone in 3 from D, and if D3 be sky, then utilize new destination collection D3 information to send message to J3; And
■ generates D4 by select the destination node of zone in 4 from D, and if D4 be sky, then utilize new destination collection D4 information to send message to J4.
Next, carry out analysis, so that determine whether to exist any destination node adjacent one another are and in zones of different to the destination node.According to this new step in this routing algorithm, may exist the node in the zone 1 adjacent with the node in zone 2; May exist the node in the zone 1 adjacent with the node in the zone 3; May exist the node in the zone 3 adjacent with the node in the zone 4; And may exist the node in the zone 2 adjacent with the node in the zone 4.In the process of inspection multicast message destination collection D, according to definition, the node that zone 1 can not have in any node and the zone 4 is adjacent, and the nodes that zone 2 can not have in any node and the zone 3 are adjacent.Therefore, with between the zones of different, still all fringe node called after destination fringe nodes (DEN) of destination node.Therefore, in this example, determine that destination fringe node collection comprises:
■ is at zone 1, and the fringe node to regional 2 is DEN 1,2={ P 29, and be DEN to zone 3 fringe node 1,3={ P 22;
■ is at zone 2, and the fringe node to regional 1 is DEN 2,1={ }, and be DEN to zone 4 fringe node 2,4={ };
■ is at zone 3, and the fringe node to regional 1 is DEN 3,1={ P 13, and be DEN to zone 4 fringe node 3,4={ P 9; And
■ is at zone 4, and the fringe node to regional 2 is DEN 4,2={ P 6, and be DEN to zone 3 fringe node 4,3={ P 2.
Next, this process can (for example, from destination collection D1 to destination collection D2 move destination node) from an area change to another zone with the destination node based on the comparison of the jumping figure that adjacent destination node is required.By checking destination fringe node collection DEN 1,2, DEN 1,3, DEN 2,1, DEN 2,4, DEN 3,1, DEN 3,4, DEN 4,2And DEN 4,3, can determine that adjacent destination node is right, be called edge, destination phase adjacency pair (DEAP).In the example depicted in fig. 3, edge, destination phase adjacency pair collection is DEAP={{P13, P22}, { P2, P9}}.
If DEAP is empty, then this algorithm does not change any destination collection (D1, D2, D3 or D4), and all-multicast message with above carry out route in the same manner about the example of Fig. 2.Yet, right if DEAP is not empty at each adjacent node, each node of the centering identified is carried out the comparison of jumping figure.This can by calculate at each adjacent destination of this centering node apart from source node Pi to the x-y distance finish.For example, for P13 and P22, x-y distance (that is jumping figure) is respectively 2 and 3.Therefore, provide multicast message D more efficient to destination node P22, because total jumping figure reduces to 3 from 5 via destination node P13.Therefore, destination node P22 is moved to destination collection D3 from destination collection D1.Similarly, destination node P2 is moved to destination collection D3 from destination collection D4.The destination collection of results modification becomes: D1={P29, P33, P35}, D2={ sky }, D3={P2, P9, P13, P22} and D4={P5, P6}.
The routing algorithm of this modification is dominant, because use less resource routing multicast message D in mesh network 300.In the unmodified algorithm that does not utilize adjacent destination node, total jumping figure is 19 as shown in the example of Fig. 2.Use utilizes the routing algorithm of the modification of adjacent destination node, and this total jumping figure reduces to 17.The minimizing of jumping figure means the resource that is used between the different nodes of mesh network the transmission data and the minimizing of power.This may advantageous particularly for the processor on the integrated circuit for handling nuclear, polycaryon processor and being arranged in.In addition, by using the routing algorithm of this modification, any computer network can be benefited because of the minimizing of flow between the node.In addition, although referring to figs. 2 and 3 the example application that illustrates in two-dimension netted network, identical principle can be applied to three-dimensional network or multidimensional network, as long as run through the definition that whole routing algorithm keeps " adjacent node (adjacent node) ".Can exist any amount wherein can realize utilizing the scene of routing algorithm of the modification of adjacent destination node.A this example has been shown among Fig. 4.
Fig. 4 is the block diagram of execution mode of computer system 400 that can realize the mesh network system 300 of one or more Fig. 3.In this system implementation mode, system 400 can comprise computing equipment 405, and this computing equipment 405 has the first processor 410 and second processor 412 that is coupled to local storage 425 by system bus.Each processor 410 and 412 can comprise the mesh network of discussing with reference to figure 3 as mentioned 300.Each processor 410 and 412 can operate between the parts (for example, node) that are used at mesh network 300 the transmission data and and memory 415 between during the transmission data, control storage 415 and each respective wire mesh network 300.In addition, can use additional data repository and communication channel (such as, by network interface 430) come and mesh network 300 transmission data, this mesh network 300 can be the part of remote computing device 450.
In another embodiment, each computing equipment 405 and 450 can also be regarded as the node in the scene of bigger mesh network (not shown), makes each bigger " computing equipment " node to be regarded as the third dimension with respect to the two-dimension netted network 300 in each computing equipment 405 and 450 processors.In addition, each mesh network 300 can comprise single integrated circuit tube core (die) or a plurality of integrated circuit lead.
Shown in the drawings and describe the execution mode of some examples explanation of theme discussed in this article hereinbefore in detail, but it is easy to make various improvement and alternative constructions.Yet, should be understood that us and unintentionally claim is limited in particular forms disclosed, and antithesis, all spirit and scope that belong to claims interior improvement, alternative constructions and equivalents are contained in our being intended that.

Claims (17)

1. mesh network comprises:
Node matrix equation; And
Source node in the described node matrix equation can be operated a plurality of other nodes that are used for to described matrix and send data;
Wherein said mesh network is by the routing algorithm management of supporting by adjacent destination node route data.
2. mesh network according to claim 1 further comprises having the two-dimension netted network that m is capable and n is listed as.
3. mesh network according to claim 1 further comprises three-dimensional netted network.
4. mesh network according to claim 1, wherein said mesh network comprises the multinode processor.
5. mesh network according to claim 4, wherein each node comprises processing node.
6. processor comprises:
Be arranged in a plurality of nodes in the mesh network by row and column, and be used for the route device that transmits and receive data according to routing algorithm between described a plurality of nodes, wherein said route device comprises:
Be used for determining source node and the device that receives a plurality of destinations node of data from described source node;
Be used for the device that zone be associated of each destination node with a plurality of zones;
Be used for determining whether the destination node in a zone is adjacent with another regional destination node, and if adjacent, then calculate for being from which zone for the data of described source node transmission the device in zone more efficiently;
Be used for adjacent destination node and the described more regional device that is associated with zones of different; And
Be used for node apparatus for transmitting data on purpose to institute.
7. processor according to claim 6 further comprises polycaryon processor, and wherein each node all is a processing unit.
8. processor according to claim 6, wherein said a plurality of nodes comprise the capable and n of m row, and comprise for the y=even number be l (x, y)=y*n+x and for the y=odd number be l (x, y)=node identification of y*n+n-x-1.
9. processor according to claim 8, further comprise by with described source node in going together mutually two adjacent nodes and with four node J1, J2, J3 and the J4s adjacent of two the adjacent nodes definition of described source node in same column with described source node, wherein related with each adjacent node sign comprises the value of following J1>J2>J3>J4.
10. processor according to claim 9, wherein said a plurality of zones comprise that the sign of described each node is relevant with the sign of described source node by four zones of the sign definition of each node, wherein:
Zone 1 comprises and having the node of the sign between [J1, (m * n)-1];
Zone 2: node with the label between [J2, J1-1];
Zone 3: node with the label between [J4+1, J3]; And
Zone 4: node with the label between [0, J4].
11. processor according to claim 6 further comprises the single integrated circuit tube core.
12. processor according to claim 6 further comprises a plurality of integrated circuit leads.
13. a system comprises:
A plurality of processing nodes in the network, and be used for according to routing algorithm from the route device of first node to a plurality of other nodes transmission data, wherein said route device comprises:
A device that the zone is associated that is used for each node and a plurality of zones that will be except described first node;
The node that is used for determining a zone whether with another zone in node adjacent, and if adjacent, then calculate for being from which zone for the data of described first node transmission the device in zone more efficiently;
Be used for adjacent node and the described more regional device that is associated with zones of different; And
Be used for to described node apparatus for transmitting data.
14. system according to claim 13, wherein each node comprises the computing equipment with at least one processing entities.
15. system according to claim 13, wherein each node comprises the two-dimension netted network of second grade, and the mesh network of each second grade can be operated and be used for according to described routing algorithm route data.
16. system according to claim 13 further comprises by the be coupled memory of at least one node of bus.
17. system according to claim 13, wherein said system comprise the network that constitutes by multistage computing network computing equipment coupled to each other.
CN2010206995097U 2010-12-30 2010-12-30 Network, system and processor using multicast route based on tree Expired - Lifetime CN201947293U (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN2010206995097U CN201947293U (en) 2010-12-30 2010-12-30 Network, system and processor using multicast route based on tree

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN2010206995097U CN201947293U (en) 2010-12-30 2010-12-30 Network, system and processor using multicast route based on tree

Publications (1)

Publication Number Publication Date
CN201947293U true CN201947293U (en) 2011-08-24

Family

ID=44474627

Family Applications (1)

Application Number Title Priority Date Filing Date
CN2010206995097U Expired - Lifetime CN201947293U (en) 2010-12-30 2010-12-30 Network, system and processor using multicast route based on tree

Country Status (1)

Country Link
CN (1) CN201947293U (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102546380A (en) * 2010-12-30 2012-07-04 世意法(北京)半导体研发有限责任公司 Modified tree-based multicast routing scheme
CN106789675A (en) * 2016-12-05 2017-05-31 清华大学 Routing multicast method and system based on 2D mesh networks

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102546380A (en) * 2010-12-30 2012-07-04 世意法(北京)半导体研发有限责任公司 Modified tree-based multicast routing scheme
CN102546380B (en) * 2010-12-30 2014-12-10 世意法(北京)半导体研发有限责任公司 Modified tree-based multicast routing scheme
CN106789675A (en) * 2016-12-05 2017-05-31 清华大学 Routing multicast method and system based on 2D mesh networks
CN106789675B (en) * 2016-12-05 2019-04-26 北京灵汐科技有限公司 Routing multicast method and system based on 2D mesh network

Similar Documents

Publication Publication Date Title
CN102546380B (en) Modified tree-based multicast routing scheme
CN106254254B (en) Mesh topology structure-based network-on-chip communication method
Li et al. Efficient collective communications in dual-cube
CN105075199B (en) Straight-forward network system with multiple distributed connections to each resource
CN103491006A (en) Method for forwarding data of virtual network router in centralized mode
CN112104558A (en) Method, system, terminal and medium for implementing block chain distribution network
EP3559833B1 (en) Best-efforts database functions
CN106161270B (en) A kind of network deployment method
CN105072194B (en) A kind of storage data in distributed file system repair structure and restorative procedure
CN108449269A (en) Data center network load-balancing method based on SDN
CN107306224A (en) A kind of routed path update method and network administration apparatus
CN105530206B (en) A kind of Torus network system and its working method with double access infrastructures
EP3560148B1 (en) Database functions-defined network switch
CN105050149A (en) Routing method for overcoming congestion of tree type Zigbee network
CN107018545A (en) A kind of ad hoc network method being applied in wireless communication system
CN102404818A (en) Method for generating and updating routing list of satellite network
CN201947293U (en) Network, system and processor using multicast route based on tree
CN102377826B (en) Method for optimal placement of unpopular resource indexes in peer-to-peer network
CN106104503A (en) Unified API is used to program the server for forwarding and framework for the fine granularity network optimization
CN116886591B (en) Computer network system and routing method
CN105282036A (en) Routing node and route exchange method and system
CN105634947A (en) Message forwarding method based on hotspot in opportunistic mobile social network
CN106341331B (en) Routing multicast method and system based on 2D mesh network
CN107707471A (en) The energy-saving scheme of minimum conversion link number based on extensive SDN
CN106789712B (en) A kind of heuristic network load balancing method

Legal Events

Date Code Title Description
C14 Grant of patent or utility model
GR01 Patent grant
CX01 Expiry of patent term

Granted publication date: 20110824

CX01 Expiry of patent term