27 results about "Fat tree" patented technology

Provided in the present application is a data routing system. The system comprises a first topological network, a second topological network and a third topological network, wherein the network structure of the first topological network is a toroidal network topology structure consisting of a first node group, the first node group comprises a first number of network nodes, and the network nodes in the toroidal network topology structure have connection relationships with upstream nodes and downstream nodes in each dimension; the second topological network is a fat-tree network topology structure consisting of the network nodes in the first topological network and a switch; there is a cascade relationship between the network nodes in the first topological network and the switch; the first topological network is nested in the third topological network, and the network structure of the third topological network is a fully connected network topology structure consisting of a second node group; and the second node group comprises a second number of first node groups, and there is a connection relationship between the first node groups. Thus, large-scale or ultra-large-scale networking can be realized at a relatively low cost.

A network on chip for high performance computing includes a first fat-tree network topology that connects each of a number of initiator nodes to one of a number of initiator core switches through one or more levels of initiator edge switches. A second fat-tree network topology connects each of the target nodes to one of a number of target core switches through one or more levels of target edge switches, and the first fat-tree network topology is joined to the second fat-tree network topology through connections between the initiator core switches and the target core switches.

The invention provides a data routing system. The system comprises a first topology network, a second topology network and a third topology network. The network structure of the first topology network is a surrounding network topology structure formed by a first node group, the first node group comprises a first number of network nodes, and the network nodes in the surrounding network topology structure are respectively connected with upstream nodes and downstream nodes in each dimension. And the second topology network is a fat tree network topology structure formed by network nodes and switches in the first topology network. And each network node and the switch in the first topology network have a cascade relationship. The first topology network is nested in the third topology network, and the network structure of the third topology network is a full-interconnection network topology structure formed by the second node groups. The second node groups comprise a second number of first node groups, and the first node groups are connected. Therefore, large-scale or super-large-scale networking can be realized with relatively low cost.

Techniques and architecture are described for a method that includes an in network compute (INC) manager receiving from switches of a fat tree configured network, arithmetic logic unit (ALU) capacity of the switches. Based at least in part on the ALU capacity of the switches and bandwidth, the INC manager determines one or more switches within each tier that are capable of supporting the processing units and based at least in part on the determining, the INC manager selects a first switch as a root, wherein the first switch is included within a tier of switches having intermediate tiers of switches located between the tier and the plurality of processing units within the fat tree configured network. The INC manager creates one or more paths of switches within each of the intermediate tiers from the root to the plurality of processing units to provide a constrained disjoint spanning tree of switches.

The invention discloses a flow forwarding method and a communication system based on a fat tree data center network topology structure, and the method comprises the main steps: a source access layer switch determines a target terminal device and an access layer switch set corresponding to the target flow when receiving the target flow sent by a source terminal device; judging whether the source access layer switch and the target access layer switch are the same access layer switch or not; if not, determining the size of the target flow, and judging whether the target flow is a large flow; if the target flow is the large flow, searching a target large flow fixed path corresponding to the access layer switching unit in a pre-configured large flow fixed path static mapping table, and configuring the target large flow fixed path as a special forwarding path of the target flow; otherwise, the data packet of the target flow is sprayed to the equivalent shortest path which is not occupied by the large flow in all the equivalent shortest paths, the large flow forwarding of the data center is reasonably controlled, and the real-time performance of small flow forwarding is guaranteed.

Provided in the present application is a data routing system. The system comprises a first topological network, a second topological network and a third topological network, wherein the network structure of the first topological network is a toroidal network topology structure consisting of a first node group, the first node group comprises a first number of network nodes, and the network nodes in the toroidal network topology structure have connection relationships with upstream nodes and downstream nodes in each dimension; the second topological network is a fat-tree network topology structure consisting of the network nodes in the first topological network and a switch; there is a cascade relationship between the network nodes in the first topological network and the switch; the first topological network is nested in the third topological network, and the network structure of the third topological network is a fully connected network topology structure consisting of a second node group; and the second node group comprises a second number of first node groups, and there is a connection relationship between the first node groups. Thus, large-scale or ultra-large-scale networking can be realized at a relatively low cost.

The embodiment of the invention provides a data packet transmission method and device, and the method comprises the steps: enabling a target node to transmit a jth response packet to a source node along a first transmission path when a target port receives a jth forward packet transmitted by a source port through the first transmission path, the source port being mounted on the source node, the target port is mounted on a target node, j is an integer greater than or equal to 1, and a plurality of transmission paths exist between the source node and the target node; and when the source node receives the jth response packet and the packet label in the jth response packet is inconsistent with the latest label under the first target entry in the source node routing table, the source port sends the (j + 1) th forward packet to the target port through the first transmission path, the (j + 1) th forward packet is a data packet sent by the source port to the target port after the jth forward packet. According to the invention, the problem of disorder of the data packet transmitted by the leveling fat tree network in the related technology is solved.

Provided in the present application is a data routing system. The system comprises a first topological network, a second topological network and a third topological network, wherein the network structure of the first topological network is a toroidal network topology structure consisting of a first node group, the first node group comprises a first number of network nodes, and the network nodes in the toroidal network topology structure have connection relationships with upstream nodes and downstream nodes in each dimension; the second topological network is a fat-tree network topology structure consisting of the network nodes in the first topological network and a switch; there is a cascade relationship between the network nodes in the first topological network and the switch; the first topological network is nested in the third topological network, and the network structure of the third topological network is a fully connected network topology structure consisting of a second node group; and the second node group comprises a second number of first node groups, and there is a connection relationship between the first node groups. Thus, large-scale or ultra-large-scale networking can be realized at a relatively low cost.

The invention discloses a multi-mode network testing method for AI application east-west flow, and relates to the technical field of network testing, and the method comprises the steps: S1, constructing a testing environment; a non-blocking fat tree architecture is adopted, a high-performance switch and a GPU server are configured, and the high-performance switch and the GPU server are used for processing large-scale data transmission and calculation tasks and simulating a real AI cluster environment; s3, performing a performance isolation test; and S4, testing a congestion scene. According to the AI application east-west flow-oriented multi-modal network test method, a set of complete multi-modal network test system is provided, and the test scene and method of each sub-test are given, so that the problems of quality evaluation such as dynamic routing and congestion control of the AI application east-west flow-oriented multi-modal network are solved, and the test efficiency is improved. The multi-modal network can be evaluated in an omnibearing and multi-angle manner, and the performance of the multi-modal network in an AI application scene can be evaluated comprehensively and accurately.

The invention discloses a data routing system. The data routing system comprises a first topology network and a second topology network. The first topology network is nested in the second topology network. The network structure of the first topology network is a fat tree network topology structure formed by a first node group and a switch, the first node group comprises a first number of network nodes, and each network node in the first topology network is in cascade connection with the switch. The network structure of the second topology network is a full interconnection network topology structure formed by second node groups, the second node groups comprise a second number of first node groups, and the second number of first node groups are connected. According to the system, a multi-stage hierarchical network structure with a nested fat tree network topology and a full-interconnection network topology is established through small switching nodes with a small number of ports, so that a large-scale or super-large-scale network is established, the network cost and power consumption are effectively reduced, and the network expandability is improved.

A network on chip for high performance computing includes a first fat-tree network topology that connects each of a number of initiator nodes to one of a number of initiator core switches through one or more levels of initiator edge switches. A second fat-tree network topology connects each of the target nodes to one of a number of target core switches through one or more levels of target edge switches, and the first fat-tree network topology is joined to the second fat-tree network topology through connections between the initiator core switches and the target core switches.

This invention discloses an adaptive optimization method for edge optical links in a data center based on deep reinforcement learning. Using a fixed fat-tree electrical switching plane as the underlying architecture, it achieves adaptive addition and deletion of direct optical links between edge switches through topology control (DQN) and optical path switching. The DQN and packet-by-packet routing (DQN) form a two-layer collaborative decision-making structure. This invention models the optical link addition / deletion decision as a constrained Markov decision process, employing a dynamic candidate set mechanism to construct a low-dimensional state and action space, significantly reducing model computational complexity and training overhead. Compared to a pure electrical fat-tree architecture, the provided dynamic candidate set DQN scheme effectively reduces average on-network latency, packet rejection rate, and maximum queue length under high-load scenarios. Under the same link resource constraints, the dynamic candidate set scheme outperforms the all-optical link pair scheme, exhibiting higher training efficiency, decision stability, and network scalability. This invention can adaptively sense traffic distribution and schedule optical link resources on demand, making it suitable for real-time control and performance optimization scenarios in high-dynamic and high-throughput hybrid optoelectronic data center network topologies.

The embodiment of the invention provides a data packet transmission method and device, and the method comprises the steps: determining a node mounted on a source port as a source node, and determining a node mounted on a target port as a target node under the condition that the source port transmits a plurality of data packets to the target port; and when the ith node receives the jth forward packet sent by the (i-1) th node and the ith node confirms that the output time of other forward packets at the ith node is later than the output time of the jth forward packet, the ith node sends a confirmation packet to the (i-1) th node. Through application of the method and the device, the problem that the data packets transmitted by the leveling fat tree network are out of order is solved, and the effect of ensuring that the data packets transmitted by the leveling fat tree network are not out of order is achieved.

The invention discloses evaluating available bandwidth between leaf switches in a fat tree network. A system for bandwidth estimation includes an interface and a processor. The interface communicates with a fat tree (FT) network that includes a leaf switch belonging to an underlying layer, a backbone switch belonging to a top layer, and an intermediate switch belonging to one or more intermediate layers. The processor: estimates an available bandwidth between a first leaf switch and a second leaf switch by identifying a first bottom-to-top subtree to which the first leaf switch belongs and a second bottom-to-top subtree to which the second leaf switch belongs; determining a path count for at least some switches in the intermediate layer, wherein the path count for a switch indicates a number of paths through the switch, via the first bottom-to-top sub-tree to the first leaf switch, and via the second bottom-to-top sub-tree to the second leaf switch; and estimating the available bandwidth based on the path count.

The present invention discloses a method for tracing the physical path of data packets in a data center network with a fat-tree topology, which includes the following steps: Step 1) Assign specific tags to each switch node and Pod in the data center network, and perform specific configurations on the server node IPs; Step 2) Based on the assigned tags, deploy certain flow table entries to each switch node in the data center network to implement functions such as pushing tags into the packet header space; Step 3) When the data packet passes through the switch node, the switch performs corresponding operations based on the flow table matching entry of the data packet; at the client or server side, the final path information sequence in the packet header is split and parsed to restore the complete path of the data packet in the data center network. This method can effectively reduce the number of switch flow table entries and the packet header space, and completely restore the physical path of the data packet in the network, and can be used for functions such as quickly measuring the network traffic matrix and monitoring and locating network anomalies.

The application discloses a fat-tree data center network topology-based traffic forwarding method and a communication system. Main steps of the traffic forwarding method are as follows: when a source access layer switch receives target traffic sent by a source terminal device, the source access layer switch confirms a target terminal device and an access layer switch group corresponding to the target traffic, and judges whether the source access layer switch and a target access layer switch are the same access layer switch; if not, the source access layer switch confirms a size of the target traffic, and judges whether the target traffic is a large flow; if the target traffic is a large flow, the source access layer switch searches a target large flow fixed path corresponding to the access layer switch group in a preconfigured large flow fixed path static mapping table, and configures the target large flow fixed path as a dedicated forwarding path of the target traffic; otherwise, the source access layer switch sprays data packets of the target traffic to an equal shortest path which is not occupied by a large flow, so that large flow forwarding of the data center is reasonably controlled, and real-time performance of small flow forwarding is guaranteed.

Disclosed in the present application is a data routing system. The system comprises a first topological network and a second topological network, wherein the first topological network is nested in the second topological network; the network structure of the first topological network is a fat-tree network topology structure consisting of a first node group and a switch, the first node group comprises a first number of network nodes, and there is a cascade relationship between the network nodes and the switch in the first topological network; and the network structure of the second topological network is a full-mesh network topology structure consisting of a second node group, the second node group comprises a second number of first node groups, and there is a connection relationship among the second number of first node groups. In the system, a multilevel hierarchical network structure where a fat-tree network topology is nested in a full-mesh network topology is established by means of small switching nodes having a small number of ports, and thus a large-scale or ultra-large-scale network is established, thereby effectively reducing the network cost and power consumption and improving the network scalability.

The embodiment of the invention provides a method and device for sending a routing request, and the method comprises the steps: obtaining the codes of a source terminal and a target terminal, and enabling the source terminal and the target terminal to be source nodes in a thin fat tree network structure; according to the codes of the source terminal and the target terminal, a target layer where a nearest public father node is located is determined in the thinned fat tree network topology structure, and the public father node is a common father node of the source terminal and the target terminal; and upwards transmitting a target routing request sent by the source terminal to a target father node of the target layer, and downwards transmitting the target routing request to the target terminal according to the idle condition of each port on the target father node. Through application of the method and the device, the problem that a congestion hotspot is easy to appear in a lower-layer routing node in the related technology is solved, and the effect of improving the throughput rate of a network is further achieved.

The invention provides a reconfigurable active interposer, an integrated chip and a configuration method thereof. The reconfigurable active intermediate layer is configured to be connected with a plurality of core particles and comprises a programmable internet network, a flash memory and a four-wire serial peripheral interface, the programmable internet network is based on a multi-layer 2D toroidal grid and fat tree mixed topology layout, and the programmable internet network is configured to dynamically adjust signal transmission of each core particle; the flash memory is connected with the programmable internet, and the flash memory is configured to store configuration data; the four-wire serial peripheral interface is configured to preload a configuration signal to the flash memory. According to the reconfigurable active interposer, dynamic path reconfiguration and selective wiring can be realized, the flexibility of network topology is improved, efficient multiplexing of the interposer is realized, and the configuration power consumption is reduced.

Disclosed in the present application is a data routing system. The system comprises a first topological network and a second topological network, wherein the first topological network is nested in the second topological network; the network structure of the first topological network is a fat-tree network topology structure consisting of a first node group and a switch, the first node group comprises a first number of network nodes, and there is a cascade relationship between the network nodes and the switch in the first topological network; and the network structure of the second topological network is a full-mesh network topology structure consisting of a second node group, the second node group comprises a second number of first node groups, and there is a connection relationship among the second number of first node groups. In the system, a multilevel hierarchical network structure where a fat-tree network topology is nested in a full-mesh network topology is established by means of small switching nodes having a small number of ports, and thus a large-scale or ultra-large-scale network is established, thereby effectively reducing the network cost and power consumption and improving the network scalability.

The embodiment of the invention provides a method and device for sending a routing request, and the method comprises the steps: obtaining the codes of a source terminal and a target terminal; determining a target layer where the nearest common father node is located in the thinned fat tree network topology structure according to the codes of the source terminal and the target terminal; and according to a target sub-network where the source terminal and the target terminal are located, a target routing request sent by the source terminal is sent to a target father node on a target layer through a routing node in the target sub-network, and the target father node sends the target routing request to the target terminal through the routing node in the target sub-network. Through application of the method and the device, the problem that the sub-networks controlled by the double hosts are mutually disturbed in a data transmission process in related technologies is solved, and the effect that the two sub-networks are mutually independent is further achieved.

A computer data network is efficiently configured to forward multicast frames among a collective of hosts using a tree spanning the collective. The computer network topology may be a tree. It may also be a Clos fat-tree configured for flow-zone switching. Operation of a data switch to support collective multicast operation is disclosed.

The embodiment of the invention provides a method and device for sending a routing request, and the method comprises the steps: obtaining the codes of a source terminal and a target terminal; determining a target layer where the nearest public father node is located in the thinned fat tree network structure according to the codes of the source terminal and the target terminal; according to a target sub-network where the source terminal and the target terminal are located and routing nodes included in the first routing node set and the second routing node set, a target routing request sent by the source terminal is transmitted upwards to a target father node on a target layer through the routing nodes in the target sub-network, and the target father node transmits the target routing request downwards to the target terminal through the routing node in the target sub-network. Through the data transmission method and device, the problem that the sub-networks are mutually disturbed in the data transmission process in the prior art is solved, and the effect that the sub-networks are not disturbed mutually is achieved.

The embodiment of the invention provides a torus and fat tree fusion computing-oriented u2u chip interconnection and exchange architecture, method and equipment and a medium. The architecture can comprise a plurality of clusters; the cluster comprises a plurality of chips which are networked and interconnected by adopting a first topological structure; the plurality of clusters are in networking connection by adopting a second topological structure; the chip is used for transmitting the to-be-transmitted data to a target chip through a communication protocol corresponding to the first topological structure, or transmitting the to-be-transmitted data to the target chip through a communication protocol corresponding to the second topological structure; according to the chip, the to-be-transmitted data can be transmitted to the target chip through the communication protocol corresponding to the first topological structure, or the to-be-transmitted data can be transmitted to the target chip through the communication protocol corresponding to the second topological structure, the interconnection modes can be flexibly switched according to different task requirements, and the communication efficiency is improved. According to the method, the requirement for a delay-sensitive real-time calculation task is met, high-bandwidth support can be provided under the condition of huge data volume, and the overall performance of processing the to-be-transmitted data is effectively improved.

The embodiment of the invention provides a method and device for sending a routing request, and the method comprises the steps: obtaining the number of a source routing node and the number of a target routing node; determining the layer number of the nearest public father node in the fat tree network topology structure according to the serial numbers of the source routing node and the target routing node; determining a target father node according to the total number of the routing requests sent to the target routing node by the source routing node and the number of the public father nodes on the layer where the nearest public father node is located; and sending a target routing request sent by the source terminal to the target father node through the source routing node, and sending the target routing request to the target terminal through the target father node. Through application of the method and the device, the problem of congestion hotspots caused by unbalanced link load in related technologies is solved, and the effect of improving the network throughput rate is further achieved.

The embodiment of the application provides a kind of fat tree network topology and the method for sending routing request of thinning, wherein the relationship between the total number of the fat tree network topology and the total number of source node: the relationship between the number of the routing node of the first layer and the number of the routing node of the first layer: Wherein, it is the integer greater than or equal to 1. Through the application, the problem of low link utilization rate of high-level routing node in the related art is solved, and the effect of improving the link utilization rate of high-level routing node is achieved.

A network-on-chip for high-performance computing includes a first fat-tree network topology connecting each of a number of initiator nodes to one of a number of initiator core switches via one or more tiers of initiator edge switches. A second fat-tree network topology connects each of the target nodes to one of a number of target core switches via one or more tiers of target edge switches, and the first fat-tree network topology is connected to the second fat-tree network topology via links between the initiator core switches and the target core switches.