A distributed chip system based on a star chain type network

Abstract

The application discloses a kind of distributed chip systems based on star chain type network, comprising: central node and multiple distributed nodes;Wherein, the central node and multiple distributed nodes form 1 or more ring star chain networks, the starting point and end point of each ring star chain network are the central node, and the direction of the ring star chain network passage is reversibly configurable.It can also be configured to the working state of distributed node, to the state of chain or hang chain activation state.The distributed chip system based on star chain type network provided by the embodiment of the application can design the chip system bus into ring star chain network structure, realize low cost, network topology structure is simple, easy to expand and maintain, and by adding star chain passage direction reversibly configurable function and node chain function, the transmission efficiency of bus can be effectively improved, power loss is reduced, and the demand of high-performance chip distributed node efficient access and low latency can be met.

Description

Technical Field

[0001] This invention relates to the field of chip design technology, and in particular to a distributed chip system based on a starchain network. Background Technology

[0002] With the development of System-on-a-Chip (SoC) technology, supported by microelectronic integrated circuit processes and IP core reuse technology, its circuit design has become increasingly complex, especially for very large-scale SoCs. Therefore, advanced integrated circuit design, represented by SoCs, is gradually transitioning from a device-centric design model to an interconnect-centric design model. Consequently, for distributed chip systems, reducing chip cost while improving chip reliability, reducing power consumption and heat dissipation, and enhancing performance has become a critical challenge for chip design. Summary of the Invention

[0003] This application provides a distributed chip system based on a starchain network. To provide a basic understanding of some aspects of the disclosed embodiments, a brief summary is given below. This summary is not intended as a general commentary, nor is it intended to identify key / important components or describe the scope of protection of these embodiments. Its sole purpose is to present some concepts in a simple form as a prelude to the detailed description that follows.

[0004] This application provides a distributed chip system based on a starlink network, comprising:

[0005] Central node and multiple distributed nodes;

[0006] In this system, the central node and multiple distributed nodes form one or more ring-shaped star chains. The central node is the starting point and the central node is the ending point of each ring-shaped star chain network. The path direction of the ring-shaped star chain network is configurable.

[0007] In an optional embodiment, the central node includes:

[0008] The request processing unit is used to receive command data sent by the upstream interface and transmit the command data to the distributed nodes on the target Starlink.

[0009] The response processing unit is used to receive command data or response data sent by distributed nodes on Starlink and return the command data or response data to the upstream interface.

[0010] The interface configuration unit is used to configure the path direction of the Starlink network;

[0011] The Function Configuration Unit is used to configure the overall working status of Starlink, as well as the working status of individual distributed nodes on Starlink.

[0012] In an optional embodiment, configuring the path direction of the Starlink network includes:

[0013] Configure the overall path direction of the Starlink network to counterclockwise, and modify the path direction of a single Starlink to clockwise; or,

[0014] Configure the overall path direction of the Starlink network to clockwise, and modify the path direction of a single Starlink to counterclockwise; or,

[0015] Configure the overall path direction of the Starlink network to be either counterclockwise or clockwise.

[0016] In an optional embodiment, the central node further includes:

[0017] The link data scheduling unit is used to receive requests from distributed nodes of the first starlink network to retrieve data from distributed nodes of the second starlink network.

[0018] Retrieve distributed node data from the second Starlink network as requested;

[0019] Return the distributed node data from the second Starlink network to the distributed nodes of the first Starlink network.

[0020] In an optional embodiment, the distributed node includes:

[0021] The chain-free processing unit is used to configure the working state of distributed nodes, which includes the chain-free state and the chain-attached activation state.

[0022] The clock and power control processing unit is used to configure the system's clock and power consumption mode.

[0023] In an optional embodiment, the distributed node further includes:

[0024] The interface processing unit is used to receive and send interface data;

[0025] The logic processing unit is used to perform logical calculations on the received interface data;

[0026] The data selection unit is used to perform logical calculations on the received interface data through the logic processing unit when the distributed node is in the chain-attached active state, and use the calculated data as the output data; when the distributed node is in the chain-free state, the received interface data is directly used as the output data.

[0027] In one optional embodiment, the communication mode between the central node and the distributed nodes includes:

[0028] Unicast mode, multicast mode, and broadcast mode.

[0029] In an optional embodiment, it further includes:

[0030] The flow control module is used to implement system bus flow control through the end backpressure mechanism.

[0031] In an optional embodiment, when the end-backpressure mechanism is applied to the source distributed node, it includes:

[0032] Obtain the receiving capacity of the current destination distributed node;

[0033] Determine if the receiving capacity of the destination distributed node is sufficient;

[0034] When the receiving capability of the destination distributed node is sufficient, send an access request to the destination distributed node and reduce the receiving capability of the destination distributed node by a preset value; after receiving the response from the destination distributed node, increase the receiving capability of the destination distributed node by a preset value.

[0035] When the receiving capability of the destination distributed node is insufficient, no access request is sent to the destination distributed node.

[0036] In an optional embodiment, when the end-backpressure mechanism is applied to the destination distributed node, it includes:

[0037] Initialize the local end's receiving capacity;

[0038] Receive access requests sent by the source distributed nodes and reduce the local end's receiving capacity by a preset value;

[0039] Send a response to the source distributed node and increase the local end's receiving capacity by a preset value to obtain the current end's receiving capacity.

[0040] The technical solutions provided in this application embodiment may include the following beneficial effects:

[0041] The distributed chip system based on a starchain network provided in this application embodiment can design the chip system bus into a ring starchain network structure, which achieves low cost, simple network topology, and easy expansion and maintenance.

[0042] Furthermore, by adding features such as configurable starlink path direction inverter and node escape chain function, the transmission efficiency of the bus can be effectively improved and power consumption loss can be reduced, thus meeting the requirements of high-performance chips for efficient access to distributed nodes and low latency.

[0043] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit the invention. Attached Figure Description

[0044] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0045] Figure 1 This is a schematic diagram of a star-chain network structure according to an exemplary embodiment;

[0046] Figure 2 This is a schematic diagram of the structure of a central node module according to an exemplary embodiment;

[0047] Figure 3 This is a schematic diagram illustrating the structure of a distributed node module according to an exemplary embodiment;

[0048] Figure 4 This is a schematic diagram illustrating a node escape chain function according to an exemplary embodiment;

[0049] Figure 5 This is a schematic diagram illustrating a Starlink path direction inverter configurable function according to an exemplary embodiment;

[0050] Figure 6 This is a schematic diagram illustrating a data packet format according to an exemplary embodiment;

[0051] Figure 7 This is a schematic diagram illustrating a reverse pressure mechanism process according to an exemplary embodiment. Detailed Implementation

[0052] The following description and accompanying drawings fully illustrate specific embodiments of the invention to enable those skilled in the art to practice them.

[0053] It should be understood that the described embodiments are merely some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0054] In the following description, when referring to the accompanying drawings, the same numbers in different drawings denote the same or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Rather, they are merely examples of systems and methods consistent with some aspects of the invention as detailed in the appended claims.

[0055] This application proposes a structure and design method for a star-chain network distributed system bus, which has low implementation cost, simple network topology, and is easy to expand and maintain. It can overcome the insufficient load capacity of traditional chip buses and solve the problem of heavy load on the central node. Through the star-chain path direction invertible configuration function and node chain escape function, the bus transmission efficiency can be effectively improved and power consumption loss can be reduced, which can meet the requirements of high-performance chips for efficient access and low latency of distributed nodes.

[0056] In one possible implementation, the chip system comprises a central node and multiple distributed nodes, such as... Figure 1 As shown, multiple central nodes and multiple distributed nodes are combined to form one or more ring-shaped star chains. The starting point and ending point of each ring-shaped star chain network are both central nodes, and the path direction of the ring-shaped star chain network is configurable.

[0057] like Figure 1 As shown, Figure 1 (a) is a two-star chain network structure, with the central node in the middle. The ring SR-0 is a star chain, with the central node at both the start and end points, and four distributed nodes A, B, C, and D in between. The ring SR-1 is another star chain, with the central node at both the start and end points, and four distributed nodes E, F, G, and H in between.

[0058] Figure 1 (b) is a four-star chain network structure, with the central node in the middle. Ring SR-0 is one star chain, with the central node at both its start and end points, and containing four distributed nodes: A, B, C, and D. Ring SR-1 is another star chain, with the central node at both its start and end points, and containing four distributed nodes: E, F, G, and H. Ring SR-2 is another star chain, with the central node at both its start and end points, and containing four distributed nodes: M, N, O, and P. Ring SR-3 is yet another star chain, with the central node at both its start and end points, and containing four distributed nodes: I, J, K, and L.

[0059] This system architecture includes a central node and several star chains. The number of star chains can be configured according to the chip requirements. Each sub-star chain contains several distributed nodes. This application does not limit the number of distributed nodes on a star chain and can be configured according to the actual number of IP devices. This includes, but is not limited to, N star chains and N distributed nodes. Figure 1 As shown, the path direction of this system is configured in a counterclockwise direction.

[0060] By setting the internal IP network of the chip into a ring-shaped star chain network structure, it achieves low cost, simple network topology, easy expansion and maintenance, and allows for the addition or reduction of the number of nodes and star chains.

[0061] like Figure 2 As shown in the embodiment of this application, the central node includes multiple input / output interface (intf) modules, which are connected to multiple star chains through the interfaces.

[0062] Specifically, the central node module mainly includes: a request processing unit, used to receive command data sent by the upstream interface and transmit the command data to the distributed nodes on the target Starlink. It also includes a response processing unit, used to receive command data or response data sent by the distributed nodes on Starlink and return the command data or response data to the upstream interface. Finally, it includes an interface configuration unit, used to configure the path direction of the Starlink network.

[0063] The path direction of the Starlink network can be configured through the interface configuration unit of the central node. For example, the overall path direction of the Starlink network can be configured to counterclockwise, and then the path direction of a specific Starlink can be changed to clockwise. Alternatively, the overall path direction of the Starlink network can be configured to clockwise, and then the path direction of a specific Starlink can be changed to counterclockwise. Or, the overall path direction can be configured to clockwise or counterclockwise and then left unchanged.

[0064] The chip system provided in this application embodiment can modify the path direction on some of the starlinks through the interface configuration unit after the path direction of the starlink network is set. By adding the path direction inverter configurable function, the transmission efficiency of the bus can be effectively improved and the power consumption loss can be reduced.

[0065] Figure 5 This is a schematic diagram illustrating a Starlink path direction inverter configurable function according to an exemplary embodiment.

[0066] like Figure 5 As shown in (a), the direction of each Starlink path is counterclockwise. Therefore, the longest link from the central node to any Starlink node is a 4-node path, and the longest link from distribution node A to the farthest distribution node P is an 8-node path. Figure 5 As shown in (b), the path direction of Starlink SR-2 is reversed. When the Starlink path is in the reversed direction, the access path of distributed node A to distributed node P is only 5 nodes, which can reduce the access path, effectively improve the transmission efficiency of the bus, and reduce power consumption loss.

[0067] The central node module also includes a function configuration unit, used to configure the overall operating state of Starlink, as well as the operating state of individual distributed nodes on Starlink, or other function configurations. For example, the function configuration unit can be used to configure the overall operating state of the entire Starlink network, setting the entire Starlink network to an off-chain state or an on-chain active state, or to configure the operating state of a specific distributed node to an off-chain state or an on-chain active state.

[0068] Specifically, when the entire Starlink network or a specific distributed node is in an off-chain state, the off-chain distributed node is invisible to the system, and its data is not processed or computed. When the entire Starlink network or a specific distributed node is in an on-chain active state, the distributed node functions normally.

[0069] In an optional embodiment, the central node further includes: a link data scheduling unit, configured to receive a request from a distributed node of the first Starlink network to retrieve data from a distributed node of the second Starlink network; retrieve the distributed node data of the second Starlink network according to the request; and return the distributed node data of the second Starlink network to the distributed node of the first Starlink network.

[0070] By setting up a link data scheduling unit, data from other links can be retrieved through the central node, enabling rapid data processing and improving system performance.

[0071] Figure 3 This is a schematic diagram illustrating the structure of a distributed node module according to an exemplary embodiment, such as... Figure 3 As shown, the distributed node module includes: an escape chain processing unit, used to configure the working state of the distributed node, which includes the escape chain state and the chain-attached activation state.

[0072] In addition to setting the working status of a node in the functional configuration unit of the central node, the working status of a node can also be set in the chain-free processing unit of the distributed node itself. The chain-free processing unit can be used to set the working status of the node to chain-free state or chain-activated state.

[0073] like Figure 4 As shown, distributed node B is switched to the chain-free state. At this time, node B is not visible to the system. The data of node A is processed by node C. Node B no longer processes the data transmitted by A. The actual data processing path is from A to C. When the configuration is switched to the chain-active state, the system function of the node is restored.

[0074] By reducing the data transmission path, the bus transmission efficiency can be further improved and power consumption loss can be reduced, which can meet the requirements of high-performance chips for efficient access and low latency of distributed nodes.

[0075] It also includes a clock and power control processing unit, which is used to configure the system's clock and power consumption mode. It can set the distributed node to a low-power mode. When the distributed node is in low-power mode, the link data will directly bypass the current distributed node, and the current distributed node is in an out-of-link state.

[0076] In an optional embodiment, the distributed node further includes: an interface processing unit for processing interface data, such as receiving and sending interface data; a logic processing unit for performing logical calculations on the received interface data, primarily for computational functions; and a data selection unit, used when the distributed node is in an active chain state, to perform logical calculations on the received interface data through the logic processing unit and use the calculated data as output data; and when the distributed node is in an off-chain state, to directly use the received raw interface data as output data without performing data calculations on this distributed node, and directly output the data.

[0077] In one possible implementation, the communication modes between the central node and the distributed nodes include: unicast mode, multicast mode, and broadcast mode.

[0078] When the communication mode is set to unicast mode, this includes the access method from the central node to a single distributed node, for example, such as... Figure 1 As shown in (b), the central node accesses Node_A@SR_0, including access methods from distributed nodes to the central node. For example, Node_A@SR_0 accesses the central node, including access methods from distributed nodes to distributed nodes. For example, Node_A@SR_0 accesses Node_N@SR_2.

[0079] When the communication mode is set to multicast mode, it refers to a broadcast access initiated by the central node to all distributed nodes on a particular starlink chain. For example: Figure 1 As shown in (b), when the central node accesses all Node@SR_1 and all Node@SR_2, it means that the central node accesses all distributed nodes on SR_1, or the central node accesses all distributed nodes on SR_2.

[0080] When the communication mode is set to broadcast mode, it refers to a network-wide broadcast access method initiated by the central node to all Starlink nodes and all distributed nodes. For example, Figure 1 (b) shows the central node accessing all Node@SR_0, all Node@SR_1, all Node@SR_2 and all Node@SR_3.

[0081] The above approach allows for flexible configuration of the chip system's access methods.

[0082] The chip system provided in this application embodiment implements communication access based on the packet format. Each packet contains different information transmitted across various fields, primarily including two categories: control information and data information. Figure 6As shown, the data information includes destination address, source address, request / response type, ID information, consistency model, payload data, etc., while the control information includes valid bits, start flag, end flag, length, etc.

[0083] The chip system provided in this application embodiment also includes: a flow control module, used to implement system bus flow control through an end-to-end reverse pressure mechanism.

[0084] In one possible implementation, when the end-to-end backpressure mechanism is applied to the source distributed node, it includes: obtaining the current end-to-end receiving capacity of the destination distributed node; determining whether the end-to-end receiving capacity of the destination distributed node is sufficient; when the end-to-end receiving capacity of the destination distributed node is sufficient, sending an access request to the destination distributed node and reducing the end-to-end receiving capacity of the destination distributed node by a preset value; after receiving a response from the destination distributed node, increasing the end-to-end receiving capacity of the destination distributed node by a preset value; and when the end-to-end receiving capacity of the destination distributed node is insufficient, not sending an access request to the destination distributed node.

[0085] When the end-to-end backpressure mechanism is applied to the destination distributed node, it includes: initializing the end-to-end receiving capacity of the local node; receiving the access request sent by the source distributed node and reducing the end-to-end receiving capacity of the local node by a preset value; sending a response to the source distributed node and increasing the end-to-end receiving capacity of the local node by a preset value to obtain the current end-to-end receiving capacity.

[0086] like Figure 7 As shown, the specific steps are as follows:

[0087] During system configuration, the credit data of the destination distributed node is initialized. This credit data represents the receiving capability of the destination distributed node and can be represented by a quantified numerical value. The source node updates the credit data of the destination distributed node, determining if the credit data is sufficient. If sufficient, it sends an access request to the destination distributed node. After sending the access request, the source node decrements and updates its stored credit data for the destination distributed node. Upon receiving the request, the destination distributed node also decrements and updates its own credit data. Then, it sends a response to the source node, incrementing and updating its own credit data. Upon receiving the response, the source node also increments its stored credit data for the destination distributed node.

[0088] If the destination distributed node does not have enough credit data at this time, wait until the destination distributed node has enough credit data before sending the request.

[0089] In an exemplary scenario, node A wants to access node B. A is the source node and B is the destination node. During system initialization, A knows how many end receivers B has. When A sends a request, A's record decreases by one. When A receives a response from B, A's record increases by one. So when A requests B, for example, it sends 4 requests, but there are only 3 end receivers, so it waits and does not occupy the bus. If there are 4 or more end receivers, it can send the request.

[0090] By configuring this mechanism, access requests are not sent when receiving capacity is insufficient, thus preventing deadlock waiting on the bus and the resulting resource consumption. This fully releases bus resources and resolves bus congestion caused by insufficient resources during the implementation of the interface backpressure mechanism.

[0091] The chip system bus configuration scheme provided in this application embodiment achieves low cost, simple network topology, and easy expansion and maintenance. It also overcomes the insufficient load capacity of traditional chip buses and solves the problem of heavy load on the central node. Through the configurable inverter function of the starlink path direction and the node chain escape function, the bus transmission efficiency can be effectively improved, power consumption loss reduced, and the requirements of high-performance chips for efficient distributed node access and low latency can be met.

[0092] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0093] The above embodiments merely illustrate several implementation methods of the present invention, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this patent should be determined by the appended claims.

Claims

1. A distributed chip system based on a starlink network, characterized in that, include: Central node and multiple distributed nodes; The central node and multiple distributed nodes form one or more ring-shaped star chain networks. The central node is the starting point and the ending point of each ring-shaped star chain network. The path direction of the ring-shaped star chain network is configurable. The central node includes: a request processing unit, used to receive command data sent by the upstream interface and transmit the command data to the distributed nodes on the target starchain; The response processing unit is used to receive command data or response data sent by distributed nodes on Starlink, and return the command data or response data to the upstream interface. The interface configuration unit is used to configure the path direction of the Starlink network; The Function Configuration Unit is used to configure the overall working status of Starlink, as well as the working status of individual distributed nodes on Starlink.

2. The system according to claim 1, characterized in that, The configuration of the Starlink network path directions includes: Configure the overall path direction of the Starlink network to counterclockwise, and modify the path direction of a single Starlink to clockwise; or, Configure the overall path direction of the Starlink network to clockwise, and modify the path direction of a single Starlink to counterclockwise; or, Configure the overall path direction of the Starlink network to be either counterclockwise or clockwise.

3. The system according to any one of claims 1-2, characterized in that, The central node also includes: The link data scheduling unit is used to receive requests from distributed nodes of the first starlink network to retrieve data from distributed nodes of the second starlink network. The request retrieves distributed node data from the second Starlink network. The distributed node data of the second Starlink network is returned to the distributed node of the first Starlink network.

4. The system according to claim 1, characterized in that, The distributed nodes include: The chain-free processing unit is used to configure the working state of the distributed node, which includes the chain-free state and the chain-attached activation state. Clock and power control processing unit, used to configure the system clock and power consumption mode; The interface processing unit is used to receive and send interface data; The logic processing unit is used to perform logical calculations on the received interface data; The data selection unit is used to perform logical calculations on the received interface data through the logic processing unit when the distributed node is in the chain-attached active state, and use the calculated data as the output data; when the distributed node is in the chain-free state, the received interface data is directly used as the output data.

5. The system according to claim 1, characterized in that, The communication modes between the central node and the distributed nodes include: Unicast mode, multicast mode, and broadcast mode.

6. The system according to claim 1, characterized in that, Also includes: The flow control module is used to implement system bus flow control through the end backpressure mechanism; When the aforementioned end-backpressure mechanism is applied to the source distributed node, it includes: Obtain the receiving capacity of the current destination distributed node; Determine whether the receiving capability of the target distributed node is sufficient; When the receiving capability of the target distributed node is sufficient, an access request is sent to the target distributed node, and the receiving capability of the target distributed node is reduced by a preset value; after receiving the response from the target distributed node, the receiving capability of the target distributed node is increased by a preset value. When the receiving capability of the destination distributed node is insufficient, no access request is sent to the destination distributed node.

7. The system according to claim 6, characterized in that, When the aforementioned backpressure mechanism is applied to the destination distributed node, it includes: Initialize the local end's receiving capacity; Receive the access request sent by the source distributed node and reduce the end reception capability of this end by a preset value; Send a response to the source distributed node and increase the local end reception capacity by a preset value to obtain the current end reception capacity.

Images